Introduction To Computers and Programming

Introduction to Computers and Programming
using
C++ and MATLAB
Alex F. Bielajew
The University of Michigan
Department of Nuclear Engineering and Radiological Sciences
2927 Cooley Building (North Campus)
2355 Bonisteel Boulevard
Ann Arbor, Michigan 48109-2104
U. S. A.
Tel: 734 764 6364
Fax: 734 763 4540
email: bielajew@umich.edu
c 20002011 Alex F Bielajew

Revision: June 29, 2011
Preface
This book arises out of a course I teach, a four-credit (52 hour), freshman-level course
Introduction to Computers and Programming being taught in the College of Engineering at
the University of Michigan. The book is in reasonably rough shape at this stage. It was
assembled from my lecture notes several years ago and is under constant revision. I may
never nish it! A wise person once said, Old age happens when you dwell more on the past
than on the future. By this denition, I have found eternal youth, insofar as this book is
concerned.
My educational objectives are quite simple. This is not really a course in computing. It
is a course in thinking, technical thinking, logical thinking, about formulating a problem,
a mathematical problem, a physics problem, a game, and then, executing a solution and
making it work. I consider the computer and the ability to program it as a kind of
laboratorya laboratory to investigate practically, the theories and ideas of other technical
courses. It is possible, for example, to teach a lot of Calculus to students without ever
mentioning the word, and there are several examples throughout this book.
This course is not about syntax. Hence, the book introduces the minimum amount syntax
to get through a problem. Indeed, I even keep some syntax hidden, to encourage students
to discover their own algorithms. So, if you are thinking of using this book as a technical
reference in C++ or Matlab, I anticipate that you will be disappointed.
The greatest value in this book, if there is any to be found, is in the exercises, problems and
projects at the back of almost every chapter. The ideal way to learn a computer language
is to learn a little syntax and then try it out on a computer. The ideal way to think is
not to read about it, but to actually do it! The book reects this. The material in the
chapters, separated from the exercises, is worse than useless, for reading the material and
not doing the problems is just a waste of time. Do the problems! Moreover, dont ask me
for the solutions! There is much more pedagogical value in a well-posed question than a
well-articulated answer.
OK, Ill step o my soap box now!
Several professors and many students have contributed to the ideas put into this book.
Professor James Holloway and I have had endless discussion on the general problem of
teaching algorithmic thinking to freshman. We still have not come to any conclusions except
i
ii
that it is highly challenging and rewarding. As for how it actually gets done, well, to be sure
it is a work-in-progress. I spent a very enjoyable term co-teaching this course (the rst time
I taught it) with Professor Ken Powell. Thanks, Ken, for nursing me through that rst year!
Both Jamess and Kens diverse computational backgrounds are reected to some degree in
this book.
I owe a debt of gratitude to the dozens of Graduate Student Instructors who have taught
with me on this course. One of them, Dan Osborne, deserves special recognition. He is one of
the most gifted and committed teachers I have ever encountered. Dan and I spent hour after
hour discussing the challenges of teaching computing and thinking skills to undergraduates.
To the 5000 or so undergraduates, mostly freshmen, who have taken my course: Every time
I teach this course I learn something newabout computing, about teaching, about the joy
of making an early deection in a students career.
Finally, I am one of those people who tends to see the forest, instead of the trees. Consequently, spelling and sentence structure usually yield to an enthusiasm to present the material
in an interesting way. Its not my only fault, but maybe my most visible one. To this end,
please inform me if you spot any errors. I am also deeply indebted to my wife-to-be, Linda
Park, for her unimaginable attention to detail, her proofreading and suggestions. Linda,
every time you undertake a proofreading project on my behalf, my mind boggles, truly.
AFB, June 29, 2011
Contents
1 Introduction to the course
1.1
What this course is about . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.2
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Data representations
2.1
Bits, nibbles, bytes and words . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2
4-bit Binary, Hexadecimal and Decimal Digits . . . . . . . . . . . . . . . . .
2.3
Binary arithmetic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.4
Converting from binary to decimal . . . . . . . . . . . . . . . . . . . . . . .
10
2.4.1
Whole numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
10
2.4.2
Real numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
11
Binary integer arithmetic on computers . . . . . . . . . . . . . . . . . . . . .
11
2.5.1
Twos-complement integer arithmetic . . . . . . . . . . . . . . . . . .
12
2.6
32-bit Binary, Hexadecimal, Unsigned and Signed Integers . . . . . . . . . .
13
2.7
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
15
2.5
3 Algorithms and Pseudocodes
23
3.1
What is an algorithm? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
23
3.2
Flowchart representation . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
3.3
Pseudocode representation . . . . . . . . . . . . . . . . . . . . . . . . . . . .
24
3.4
Decisions, conditionals, branch instructions
. . . . . . . . . . . . . . . . . .
25
3.5
Looping or jump instructions . . . . . . . . . . . . . . . . . . . . . . . . . .
25
iii
iv
CONTENTS
3.6
Sequencing, branching and looping . . . . . . . . . . . . . . . . . . . . . . .
26
3.7
A mini-summary before the examples . . . . . . . . . . . . . . . . . . . . . .
27
3.8
Some examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
27
3.8.1
The Breakfast algorithm, or, Bielajews Sunday Morning Internationallyfamous pancakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3.8.2
Solve for x: Ax2 + Bx + C = 0 where A, B, C are arbitrary constants
29
3.8.3
Iteration: A summing loop . . . . . . . . . . . . . . . . . . . . . . . .
33
3.8.4
Iteration: A product loop . . . . . . . . . . . . . . . . . . . . . . . .
36
An aside on computer architecture . . . . . . . . . . . . . . . . . . . . . . .
40
3.9.1
What does S = S + 1 mean? . . . . . . . . . . . . . . . . . . . . . . .
40
3.10 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
41
3.11 Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
47
3.9
4 Getting started in C++
49
4.1
Simple input/output (I/O): A rst program in C++ . . . . . . . . . . . . . .
49
4.2
Compiling, linking, loading and running . . . . . . . . . . . . . . . . . . . .
52
4.3
Declaring and initializing variables . . . . . . . . . . . . . . . . . . . . . . .
55
4.4
Integer math in C++ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
57
4.5
Floating point math in C++ . . . . . . . . . . . . . . . . . . . . . . . . . . .
59
4.6
The if/else if/else construct . . . . . . . . . . . . . . . . . . . . . . . . .
60
4.7
Logical expressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
63
4.7.1
Logical expressions with AND or OR . . . . . . . . . . . . . . . . . .
66
4.7.2
Mixed arithmetic and logical expressions . . . . . . . . . . . . . . . .
67
4.8
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
4.9
Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
73
5 Loops
83
5.1
The while loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
83
5.2
The do/while loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
86
5.3
The for loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
89
CONTENTS
5.4
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5.5
Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
6 Early AbstractionFunctions
94
123
6.1
Motivation for functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
6.2
User-dened functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
6.3
Example: A problem tackled with teamwork . . . . . . . . . . . . . . . . . . 134
6.4
Call by value, call by reference, reference parameters . . . . . . . . . . . . . 138

6.4.1
The address of a variable . . . . . . . . . . . . . . . . . . . . . . . . . 138
6.4.2
Call-by-value vs. call-by-reference . . . . . . . . . . . . . . . . . . . . 140
6.5
The rules of scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
6.6
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
6.7
Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160
7 More Variable Types, Data Abstraction

7.1
167
Representation of oating-point numbers . . . . . . . . . . . . . . . . . . . . 167

7.1.1
When is one not one? Floating point anomalies. . . . . . . . . . . . . 174
7.2 char: the character variable . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

7.2.1
7.3
7.4
Character strings; the string class . . . . . . . . . . . . . . . . . . . . 177
The vector class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

7.3.1
Introduction to vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 179
7.3.2
Declaring and using vectors . . . . . . . . . . . . . . . . . . . . . . . 180
7.3.3
Vector syntax and rules: . . . . . . . . . . . . . . . . . . . . . . . . . 182
7.3.4
Vectors and functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 187
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
8 More Data Abstraction, Arrays, Structures

8.1
199
Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199
8.1.1
Declaring a 2-dimensional array . . . . . . . . . . . . . . . . . . . . . 200
8.1.2
Initializing a 2-dimensional array . . . . . . . . . . . . . . . . . . . . 201
vi
CONTENTS
8.1.3
8.2
8.3
Passing a two-dimensional array to a function . . . . . . . . . . . . . 204
Character arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

8.2.1
One dimensional character arrays . . . . . . . . . . . . . . . . . . . . 209
8.2.2
Two dimensional character arrays . . . . . . . . . . . . . . . . . . . . 211
Structures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
8.3.1
The structure denition . . . . . . . . . . . . . . . . . . . . . . . . . 212
8.3.2
Where can structures be dened? . . . . . . . . . . . . . . . . . . . . 213
8.3.3
The members of a structure . . . . . . . . . . . . . . . . . . . . . . . 213
8.3.4
Declaring structure variables . . . . . . . . . . . . . . . . . . . . . . . 213
8.3.5
Assigning values to structure members . . . . . . . . . . . . . . . . . 213
8.3.6
Re-assigning values of structure members . . . . . . . . . . . . . . . . 214
8.3.7
Function call-by-value of a structure . . . . . . . . . . . . . . . . . . 214
8.3.8
Function call-by-reference to a structure . . . . . . . . . . . . . . . . 215
8.3.9
Structure arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215
8.3.10 Example using structures and arrays: Ion transport . . . . . . . . . . 218

8.4
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 222
8.5
Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228
9 Miscellaneous Topics
9.1
239
Generating random numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . 239

9.1.1
Example: Integrating functions by random sampling
. . . . . . . . . 246
9.2
Simple Sorting: The bubble or sinking sort . . . . . . . . . . . . . . . . . . . 248
9.3
RecursionA function calling itself . . . . . . . . . . . . . . . . . . . . . . . 249
9.4
Input and Output using les . . . . . . . . . . . . . . . . . . . . . . . . . . . 252

9.4.1
File and stream handling in C++ . . . . . . . . . . . . . . . . . . . . 252
9.4.2
Creating sequential access les and writing to them . . . . . . . . . . 255
9.4.3
Reading from sequential access les . . . . . . . . . . . . . . . . . . . 258
9.5
Command line arguments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261
9.6
Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 264
CONTENTS
10 A Potpourri of Applications
vii
267
10.1 Finding a zero using the binary chop . . . . . . . . . . . . . . . . . . . . . 267

11 Programming in MATLAB
271
11.1 A taste of MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 271

11.2 Arrays in Matlab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 273
11.3 Loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
11.3.1 The for loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 279
11.3.2 The while loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
11.4 The if/else construct . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282
11.5 Some Matlab terminology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283
11.6 Operators in MATLAB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 284
11.6.1 Math operators in Matlab . . . . . . . . . . . . . . . . . . . . . . . . 284
11.7 Pointwise operators in MATLAB . . . . . . . . . . . . . . . . . . . . . . . . 285
11.7.1 Logical Operators in Matlab . . . . . . . . . . . . . . . . . . . . . . . 286
11.8 M-les . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
11.8.1 Script M-les . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
11.8.2 Function M-les . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 292
11.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
11.10Projects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 327
11.10.1 Supplementary material: Trajectories without air resistance . . . . . 346
11.10.2 Supplementary material: Trajectories with air resistance . . . . . . . 346
11.10.3 Supplementary material: Stepping algorithms . . . . . . . . . . . . . 347
11.10.4 Supplementary material: Trajectories of objects in the universe . . . 354
11.10.5 Supplementary material: Stepping algorithms . . . . . . . . . . . . . 355
12 Graphics
367
12.1 Two Dimensional Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367

12.1.1 A basic plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 367
12.1.2 Using dierent line colors
. . . . . . . . . . . . . . . . . . . . . . . . 367
viii
CONTENTS
12.1.3 Making titles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
12.1.4 Axis labels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
12.1.5 Dierent line styles . . . . . . . . . . . . . . . . . . . . . . . . . . . . 370
12.1.6 Point plots, plotting with point symbols . . . . . . . . . . . . . . . . 372
12.1.7 Plotting more than one thing at once . . . . . . . . . . . . . . . . . . 377
12.1.8 Subplots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 378
12.2 Three dimensional graphics . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
12.2.1 Plot3 plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
12.2.2 Mesh plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 379
12.2.3 Axis labelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 381
12.2.4 meshc and meshz plots . . . . . . . . . . . . . . . . . . . . . . . . . . 382
12.2.5 Surface plots using surf . . . . . . . . . . . . . . . . . . . . . . . . . . 382
12.2.6 Surface plots using surfc . . . . . . . . . . . . . . . . . . . . . . . . . 383
12.2.7 Surface plots using sur . . . . . . . . . . . . . . . . . . . . . . . . . 383
12.2.8 Contour plots using contour . . . . . . . . . . . . . . . . . . . . . . . 384
12.2.9 Contour plots using contour3 . . . . . . . . . . . . . . . . . . . . . . 384
12.2.10 Contour plots using pcolor . . . . . . . . . . . . . . . . . . . . . . . . 384
12.2.11 Contour plots using contourf . . . . . . . . . . . . . . . . . . . . . . . 385
12.2.12 Contour plots with labels . . . . . . . . . . . . . . . . . . . . . . . . . 385
13 Miscellaneous topics
387
13.1 Pitfall review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387

13.1.1 Review of for loops . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
13.1.2 Review of functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
13.2 The MATLAB way . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
13.3 Selected Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 395
13.3.1 Free fall with air resistance . . . . . . . . . . . . . . . . . . . . . . . . 395
13.3.2 Diusion of charged ions in an electric eld . . . . . . . . . . . . . . . 400
13.3.3 Calculation of electric elds . . . . . . . . . . . . . . . . . . . . . . . 402
CONTENTS
13.4 Summary of the Course
ix
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 404
14 Programming Style Guide for C++
405
15 Syntax reference for beginning C++
413
16 Syntax reference for more advanced C++
417
17 Syntax reference for MATLAB
421
CONTENTS
Chapter 1
Introduction to the course
Computers are stupid. They only give answers.
Computers are blindly faithful. They only do what you tell them to do.
Computers are dumb. They do not do what you want them to do.
1.1
What this course is about
This course is called ENG101: Introduction to Computers and Programming.

The College of Engineering Bulletin describes it as follows (emphasis is mine):
Algorithms and programming in C++ and MATLAB, computing as a tool in engineering,
introduction to the organization of digital computers.
Lets dene a few things rst...
What is an engineer? What is engineering?
The American Heritage Dictionary denes engineering as The application of science to practical ends, such as the design, manufacture, and operation of structures, machines, and
systems. An engineer is dened as someone who practices engineering as a profession.
Our denition is somewhat broader. Engineering is a broad range of technical endeavor
that starts with the basic sciences (trying to characterize and understand the basic laws of
nature), integrates and systematizes this knowledge, builds things based on these principles,
manufactures these things on a large scale, and makes them available for societal benet.
If we broaden our denition of things to ideas, methods, and algorithms (a prescription
for doing something, like a recipe in cooking), as well as physical devices, then that captures
the essence of what engineering is today.
1
CHAPTER 1. INTRODUCTION TO THE COURSE
In our College of Engineering you will encounter faculty and sta who might otherwise have
been called Mathematicians, Physicists, Chemists and Biologists.
What is a computer?
Again, we appeal at rst to some standard reference texts.
American Heritage Dictionary
1. One (a person) who computes.
Some of the best examples:
- Richard Feynmans computersteams of people working out calculations during the
Manhattan Project
- Calculation of ballistics tables for WWII
- Calculation of sine/cosine/logarithm tables
2. A device (a machine) that computes, especially a programmable electronic machine
that performs high-speed mathematical or logical operations or that assembles, stores,
correlates, or otherwise processes information.
Oxford English Dictionary
1. One who computes; a calculator, reckoner; spec. a person employed to make calculations in an observatory, in surveying, etc. (Earliest reference 1646).
2. A calculating-machine; esp. an automatic electronic device for performing mathematical or logical operations; freq. with dening word prexed, as analogue, digital,
electronic computer (see these words). (Earliest reference 1897).
McGraw-Hill Encyclopedia of Science and Technology
A device that receives, processes, and presents information. The two basic types of
computers are analog and digital. Although generally not regarded as such, the most
prevalent computer is the simple mechanical analog computer, in which gears, levers,
ratchets, and pawls perform mathematical operations for example, the speedometer
and the watt-hour meter (used to measure accumulated electrical usage). The general
public has become much more aware of the digital computer with the rapid proliferation
of the hand-held calculator and a large variety of intelligent devices, ranging from
typewriters to washing machines.
Our denition A device that
1. accepts some form of input (usually disorganized or of some highly specic nature),
2. responds to this input information in a pre-determined (programmed) way,
3. produces some form of output (usually more organized or coherent than the input)
Some common examples of computers
1.1. WHAT THIS COURSE IS ABOUT
Personal: Wristwatch, electronic car keys, phone, radio, GPS device, calculator, laptop...
In the home: Stove, refrigerator, freezer, microwave oven, clocks and timers, toaster,
coee machine, thermostat, washer and dryer, TVs, DVD/Blu-Ray players, audio
equipment ... the list is endless.
Automotive: fuel delivery, air/fuel mixture, spark advance, brakes (ABS), traction control, suspension adjustment, climate control, navigation, parking assist, lane variation
warning, proximity warnings, computer-controlled driving (coming soon)...
Mechanical analog computers: Wind-up clocks, toys, mechanical calculators...
Electrical analog computers: Integrators, dierentiators, dierential equation solvers,
Christmas light ashers (cheap ones!)...
Digital computers: Calculators, PDAs, Palm computers, laptops, desktops, minicomputers, mainframe computers....
Digital computers?
Digital computers are the simplest of all! They only understand two things, which we
label 1 and 0, or on or o, but more precisely, states of high or low voltage.
Computers switch between these two states very, VERY quickly. Modern run-of-the-mill
digital computers can throw about 100 billion switches per second! Thats enough to write
about 10 million pages of text in one second.
The input is rarely a collection of 0s and 1s, 011110000110... It can be in the form of
text that is converted to 0s and 1s or voltage that is digitized with an analog-todigital converter. (Musical recording, for example.)
The output is similarly rarely a collection of 0s and 1s. It can be 1D data (words,
numbers), 2D data (functions, charts, pictures, music), 3D data (video, surfaces, timeevolving functions)...
The processing or computing or programming part is what the bulk of the course
is concerned with. The programming responds to the input in some pre-determined
way. In some digital computers the programming is xed, for example, a digital clock
or a digital thermometer. In the most interesting cases, the program can be changed.
In this course we will learn to program programmable digital computers.
How are digital computers programmed?

You can either start o the computer with the 0s and 1s in the right place, or, you use a programming language. Programming languages provide the prescription, the pre-determined
steps that instruct the computer how to respond to the various input it gets. We will learn
two languages in this course.
C++ is a very popular (one of the most popular and common) language, a general-purpose
language that can be used for all aspects of computing. It is the best language to use
if one wants to teach many of the basic operations that a computer can perform.
MATLAB is a language that is suited for scientic and engineering applications, designed
for getting answers quickly, with the minimum of eort, with reasonable guarantee
that the answer is correct. It also provides graphical output with a minimum of eort.
What this course is REALLY about?
The hidden agenda
This course is all about putting ideas into action. It is about taking understood notions of
how things work and building something, developing a numerical solution for some specic
problem. This is what engineering is all about. The course will try to put your knowledge
of mathematics and physics to practical use. Some of the problems we may solve in this
course...
How can I program a computer to play a simple game?
How far can a golf ball y into a 50 mile/hour wind?
How can one nd the shortest distance between two points when the landscape is not
at?
If I take 10,000 steps in random directions, how far, on average, will I be from where
I started?
How can I do calculus on a computer (integrate, dierentiate, nd maxima and minima)?
Where should I place a radiator in my room so that I am warm and do not have to
pay too much for the heat?
How can I make a vacuum?
1.1. WHAT THIS COURSE IS ABOUT
What this course is NOT REALLY about?

The ugly stu
It is not primarily a course on computer language syntax. It may seem like it at times,
but it really isnt. Learning syntax is just a means to an end.
It is not really a course on operating systems, computer hardware, or the ne detail of
how computers work. We will discuss this but only enough to get the job done.
Educational Objectives
Engineering 101, Introduction to Computers and Programming, is intended to give you an
exposure to, and a chance to practice, algorithmic thinking. It does so by asking you to
actually solve problems using algorithms. This may be dierent from other courses that you
have taken, because we do not tell you exactly how to solve the assigned problems. Rather,
we provide you with tools that will allow you to solve them and some examples of how
problems are solved. But you must creatively assemble the solution yourself.
Note that neither C++ nor MATLAB languages per se are the keys to the course. We
teach you these languages so that you will have a means to express algorithms for execution
on a computer, but a great many other computer languages could have been selected to
support our central purpose of getting you to think about solving problems through a set
of procedural steps. Because programming languages such as C++ and MATLAB contain
many constructs and demand great attention to detail, it is easy to lose sight of the true
intellectual center of the course. But keeping a focus on algorithmic thinking will make
writing programs less dicult and less time consuming.
The educational objectives of Engineering 101 include:
To introduce students to algorithmic thinking in their approach to problem solving.
To teach students to implement algorithms in ANSI C++.
To teach students to implement algorithms in MATLAB.
To have students apply their knowledge of elementary physics and calculus.
To provide students with a foundation on which to base their later applications of
computers in engineering.
Students who successfully complete Engineering 101 (receive a grade of C or better) should:
Design algorithms to accomplish clearly specied tasks.

Display knowledge of ANSI C++ syntax and semantics.
Display knowledge of MATLAB syntax and semantics.
Successfully implement clearly stated algorithms in C++.
Successfully implement clearly stated algorithms in MATLAB.
Be prepared to engage in discipline-specic instruction in computing.
1.2
Problems
1. In the broad denition of computer in this chapter, how many computers can you
nd: On yourself? In your room? In a car? In your home?
2. A typical 3rd grader is a lot smarter than any computer. They can read, do simple
math, play a musical instrument, memorize poetry and write stories. Generally, with
sucient motivation, they will also do as they are told. How many ways could you
instruct a 3rd grader to determine the value of without the aid of a computer or
calculating device?
Chapter 2
Data representations
2.1
Bits, nibbles, bytes and words
bit The digital computer processing hardware is made up of transistors, each of which can
exist in only one of two possible states, a state of high voltage or low voltage.
This state of high/low (or on/o) is represented abstractly as 1/0 (one/zero). A bit
is some piece of microscopic (usually!) hardware that can exist in either of these two
states as determined by a programmer or computer designer. In addition to the voltage
states in a transistor in a computers memory chips or processing chips, there are many
others forms: magnetic domains in a oppy disk, hard disk or magnetic tape, pits
in a CD or DVD, holes in paper tape or computer cards. Nowadays, there are billions
or trillions of bits of various forms in a typical computer. A bit may be simple, but
it is usually so small, typically well below a micrometer across (1 micrometer = 106
m), that we can conceive of cramming billions of them into a relatively small volume
of space. Some bits, particularly those in the processing chips of a computer, can also
be made to change states very rapidly, well under one nanosecond. (1 nanosecond =
109 s.) So, we can do a lot of dierent things with lots of bits in a very short span of
time.
nibble A nibble is a collection of 4 bits. A nibble can be in one of 24 = 16 separate
combinations of bits.
byte A byte is a collection of 2 nibbles, or 8 bits. A byte can be in one of 28 = 256
separate combinations of bits. This is enough to, for example, associate with each
unique pattern one of the characters of the English alphabet (upper and lower case)
plus all the common special characters we may use, for example, in writing English
prose or a technical document, e.g. !$&(),:; ?. And, there would be lots of room left
over to make other interesting associations.
7
CHAPTER 2. DATA REPRESENTATIONS
word A word is a collection of bytestypically 4 for 232 = 4, 294, 672, 296 possible combinations. We can do a lot with all the dierent representations of words. We will
see, later in the lecture, how whole numbers can be associated with the various word
patterns.
The notion that many bits can conspire, in very large numbers, to produce wonderfully
intricate objects is illustrated in the following pictures of my dog, Matilda Skip Bielajew II.
(Yes, that is her real name!) The rst picture was generated from a le that was about 12
Megabytes, containing 96 million bits. It looks like real life. Yet, even when we zoom a little,
we start to see the eects of the underlying bit structure. Look at how her whiskers appear
jagged. Under extreme magnication (one of her whiskers, again) we see that the picture is
made up of at, uniformly colored tiles. Each one of these tiles is represented by 24 bits,
8 bits each for red, green and blue intensity, that, to all appearances, aords an apparently
uniform transition of color from tile to tile.
2.2. 4-BIT BINARY, HEXADECIMAL AND DECIMAL DIGITS
2.2
4-bit Binary, Hexadecimal and Decimal Digits
4-bits of data (a nibble) can represent or be represented by several dierent things. It is

common for computer programmers to use hexadecimal notation (0, 1, 2...9, A, B...F) as
a shorthand for (0000, 0001, 0010...1001, 1010, 1011...1111). See the following table.
However, collection of a certain number of bits has to be given an interpretationone that
is sensible to humans. One such assignment (which is completely arbitrary and is decided
upon by the computer programmer or designer) is to assign decimal numbers to the strings
of bits. Such an assignment could look like:
Binary
Hexadecimal
representation representation
0000
0
0001
1
0010
2
0011
3
0100
4
0101
5
0110
6
0111
7
1000
8
1001
9
1010
A
1011
B
1100
C
1101
D
1110
E
1111
F
Decimal
assignment
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
We will use this particular representation in doing some simple arithmetic in the exercises
and the assignments.
2.3
Binary arithmetic
Let us pretend that we have to do some arithmetic (additions and multiplications) using only
binary numbers. The numbers 2,3,....9 do not exist! Only 0 and 1 exist. We also assume
that the operations of addition and multiplication are commutative (i.e. independent of the
order of the mathematical operation).
The addition of ctional binary numbersbase-2 arithmetic looks like:
10
0
1
10
11
10
.
.
1111
.
.
.

+
+
+
+
+
?
1
1
1
10
=
=
=
=
=
?
10
1
1
10
+ 0
= ?
+ 10
+ 11
+ 10
= 11
= 100
= 100
+ 111 = 111 + 1111 = 10110
The multiplication of
0
?
=
1
?
=
10
10 =
10
11 =
11
11 =
.
.
.
1111 111 =
.
.
.
ctional
?
10
11
1001
111
binary numbersbase-2 arithmetic looks like:

0
= 0
1
= ?
10
= 100
10
= 110
1111 = 1101001
The arithmetic procedure is analogous to the more familiar base-10 system (that arose simply
because of our anatomy10 ngers). Additions of columns of numbers, subtraction, multiplication, the procedure of long division, can be accomplished by analogous procedures.
There is no limit to how long a ctional base-2 number can be and we may have to carry
along some extra information with each numberits sign.
2.4
2.4.1
Converting from binary to decimal

Whole numbers
Humans are usually quite awful at doing binary arithmetic. If you are doing such a calculation, you can check by converting the binary numbers into their decimal equivalents.
In general, to convert a whole binary number to a whole decimal number, consider a binary
number of the form:
bn bn1 bn2 b2 b1 b0 ,
2.5. BINARY INTEGER ARITHMETIC ON COMPUTERS
11
where the bi s are either 0 or 1. We compute a decimal number using the following formula:
d10 = bn 2n + bn1 2n1 + bn2 2n2 b2 22 + b1 21 + b0 20 .
For example, the result of the two more dicult calculations above can be checked as follows:
101102 = 1 24 + 0 23 + 1 22 + 1 21 + 0 20 = 24 + 22 + 21 = 1610 + 410 + 210 = 2210
is the result of one of the additions [11112 (1510 ) + 1112 (710 )] above, while
11010012 = 26 + 25 + 23 + 20 = 6410 + 3210 + 810 + 110 = 10510
is the result of one of the multiplications [11112 (1510 ) 1112(710 )] above. Note that the
subscript 2 indicates a binary or base-2 number while the subscript 10 indicates a
decimal number.
2.4.2
Real numbers
To convert a real(non-whole) binary number to a real decimal number, consider a binary

number of the form:
bn bn1 bn2 b2 b1 b0 . b1 b2 b(m2) b(m1) bm ,
where the bi s are either 0 or 1 and a period, ., separates the whole part, bi where i 0
from the fractional part, bi where i < 0. We compute a real decimal number using the
following formula:
r10 = bn 2n +bn1 2n1 b1 21 +b0 20 +b1 21 +b2 22 +b(m1) 2(m1) +b(m) 2m .
It is not a whole lot dierent from the decimal numbers you are used to. However, sometimes
the formulae can look a little messy.
2.5
Binary integer arithmetic on computers
Computers, because of their comprehension of data in binary terms, naturally do their

arithmetic using the base-2 representation. The circuitry within a computer is most easily
set up this way. However, when it comes to base-2 arithmetic and computers, there is a
subtle dierence! For practical reasons, computers must deal with integers that are nite
in length. 16 bits, representing 65,536 separate possibilities, used to be the norm. Today,
12
32 bits is common (4294967296 possibilities). Soon, 64 bits (about 181018 ) will be the
standard. Some computers nowadays permit hardware 64-bit integer arithmetic.
Let us consider 32-bit computers onlythe most common type of hardware available today.
In the course of doing an arithmetic calculation, any bits that are set beyond the 32-bit
boundary are lostthey fall into the bit bucket and disappear.
2.5.1
Twos-complement integer arithmetic
A string of 32-bits can represent many things but we will focus on hexadecimal, unsigned
and signed integers. These signed integers must also carry information as to the sign of the
number. Historically, there have been several ways to do this. The industry has settled
on twos-complement arithmetic because it simplies the process of adding positive, negative or mixed numbers. In this scheme, to negate a signed integer, one takes the complement
(change all the 1s to 0s and all the 0s to 1s) of the binary representation and adds 1 to it.
This is seen in the table at the end of this chapter.
Lets give the notation i to the complement of the integer i. So, for any i,
i + i = 11111111111111111111111111111111. That is:
any 32-bit pattern
+ complement of the above
= 11111111111111111111111111111111
and
11111111111111111111111111111111
+ 00000000000000000000000000000001
= 00000000000000000000000000000000
since the 33rd bit can not be set! In twos-complement arithmetic the expression i j is
calculated as i + j + 1. So you see, computers can not really add and subtract, they can only
add (and take complements)!
A few examples help illustrate the simplicity of the twos-complement scheme. Consider a
4-bit representation of the number 310 = 00112 and 410 = 01002 . There are two tricks you
should apply here: a) every time you see a negative sign in front of a bit pattern, i, to be used
in a calculation, convert it to a positive sign, change i to i + 1 and do the binary addition,
and, b) if a nal result has a leading order (leftmost) bit set to 1, it is a negative number,
so, extract the minus sign and convert the resultant bit pattern j to j + 1.
310 + 410 = 00112 + 01002 = 01112 = 710
310 410 = 00112 01002 = 00112 + 11002 = 11112 = 00012 = 110
310 410 = 00112 01002 = 11012 + 11002 = 10012 = 01112 = 710
2.6. 32-BIT BINARY, HEXADECIMAL, UNSIGNED AND SIGNED INTEGERS
13
See how easily it all worked out! This is why the twos-complement approach has been
adopted. In order to handle negative integers one can convert them to their twos-complement
counterparts and add. Otherwise, special circuitry would have to be developed for handling
negative numbers and that would be less ecient.
2.6
32-bit Binary, Hexadecimal, Unsigned and Signed

Integers
32-bit computers can represent 232 = 4, 294, 672, 296 dierent things but ONLY 4, 294, 672, 296
dierent things. Now we consider the conventional assignments to unsigned and signed
integers. There is also the hexadecimal representation that is associated with every 4-bit
sequence, as indicated in the table below.
The unsigned integers associate the decimal number 0 to a string of 32 0 bits. The rest are
formed by the addition by 1 in either binary or decimal. The maximum unsigned integer,
therefore, is 4, 294, 672, 295. Adding a one to this causes all the 32 bits to go back to zero,
exactly as if you had cycled the odometer in your car! Its the same principle. There is
no 33rd bit, so you can imagine it just falls o the end.
The signed integers have the same sequence as the unsigned integers up to 231 1 =
2147483647 which is the bit pattern 01111111111111111111111111111111. Adding a one
to this gives a bit pattern of 10000000000000000000000000000000 which is interpreted as
231 = 2147483648. Note that these two are complements of each other and that they add
up to
11111111111111111111111111111111, which has the interpretation -1, consistent with the
twos-complement scheme described above.
14

Binary
Hexadecimal
00000000000000000000000000000000
00000000
00000000000000000000000000000001
00000001
00000000000000000000000000000010
00000002
00000000000000000000000000000011
00000003
00000000000000000000000000000100
00000004
.
.
.
.
.
.
00000000000000000000000000001111
0000000F
00000000000000000000000000010000
00000010
00000000000000000000000000010001
00000011
.
.
.
.
.
.
01111111111111111111111111111110
7FFFFFFE
01111111111111111111111111111111
7FFFFFFF
10000000000000000000000000000000
80000000
10000000000000000000000000000001
80000001
.
.
.
.
.
.
11111111111111111111111111111100 FFFFFFFC
11111111111111111111111111111101 FFFFFFFD
11111111111111111111111111111110 FFFFFFFE
11111111111111111111111111111111 FFFFFFFF
Unsigned Int Signed Int

0
0
1
1
2
2
3
3
4
4
.
.
.
.
.
.
15
15
16
16
17
17
.
.
.
.
.
.
2147483646
2147483646
2147483647
2147483647
2147483648
-2147483648
2147483649
-2147483647
.
.
.
.
.
.
4294967292
-4
4294967293
-3
4294967294
-2
4294967295
-1
2.7. PROBLEMS
2.7
15
Problems
1. octal + octal = octal

Note: An octal is a number represented by only 3 bits.
Consider a ctitious computer, one that can only do 3-bit binary arithmetic. (Anything
past the 3rd bit can not be represented. See the notes for Lecture 2 for some related
examples.) Complete the following addition table, that is, add the octal in ith row to
the octal in the jth column and put the resultant octal in the box located at the ith
row and jth column. There are 3 examples worked out for you. For these examples,
show your calculations in long-hand.
000 001 010 011 100 101 110 111
000
001
010
011
100
101
110
111
010
011
110
2. octal octal = octal

This is a similar exercise to the above except that you are to multiply the octals. Be
sure to show your work for the 3 examples given.
000 001 010 011 100 101 110 111
000
001
010
011
100
101
110
111
001
100
001
16

3. Bits, nibbles and integers. Is it a hex?
Your imaginary computer represents integers using 4 and only 4 bits! The result of an
arithmetic operation with two integers always results in a 4-bit result. Higher-order
bits are lost.
The following table shows the binary, hexadecimal, unsigned integer, and signed integer representations for all the possible bit patterns that can be represented by your
computer.
Binary Hexadecimal
0000
0
0001
1
0010
2
0011
3
0100
4
0101
5
0110
6
0111
7
1000
8
1001
9
1010
a or A
1011
b or B
1100
c or C
1101
d or D
1110
e or E
1111
f or F

0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
-8
9
-7
10
-6
11
-5
12
-4
13
-3
14
-2
15
-1
Do your calculations in binary arithmetic and SHOW YOUR WORK! Quote your
results of your calculations in binary, integer, unsigned and hexadecimal integer representations.
(a) What is the sum of the integers 2 and 3?
(b) What is the sum of the integers -2 and -3?
(c) What is the sum of the integers -8 and -8?
(d) What is the product of the integers 2 and 3?
(e) What is the product of the integers -2 and -3?
(f) What is the product of the integers -8 and -8?
2.7. PROBLEMS
17
4. The interbreeding hex: hex + hex = hex

Consider a ctitious computer, one that can only do 4-bit binary arithmetic. (Anything
past the 4th bit can not be represented. See the notes for Lecture 2 for some related
examples.) Complete the following addition table, that is, add the hex in ith row to
the hex in the jth column and put the resultant hex in the box located at the ith
row and jth column. Some of the answers are given to you. Fill in the empty boxes.
Remember that addition commutes. Therefore the actual work is halved and you need
not do the boxes that are lled with Xs.
+ 0 1 2 3 4 5 6 7 8 9 A B C D E F
0 0 X X X X X X X X X X X X X X X
1 1 2 X X X X X X X X X X X X X X
2 2 3 4 X X X X X X X X X X X X X
3 3 4
6 X X X X X X X X X X X X
4 4 5
8 X X X X X X X X X X X
5 5 6
a X X X X X X X X X X
6 6 7
c X X X X X X X X X
7 7 8
e X X X X X X X X
8 8 9
0 X X X X X X X
9 9 a
2 X X X X X X
A a b
4 X X X X X
B b c
6 X X X X
C c d
8 X X X
D d e
a X X
E e f
c X
F f 0
e
18

5. Oh, hex, wont it ever stop: hex hex = hex
This is a similar exercise to the previous one except that you are to multiply the hexes.
X 0 1 2 3 4 5 6 7 8 9 A B
0 0 X X X X X X X X X X X
1 0 1 X X X X X X X X X X
2 0 2 4 X X X X X X X X X
3 0 3
9 X X X X X X X X
4 0 4
0 X X X X X X X
5 0 5
9 X X X X X X
6 0 6
4 X X X X X
7 0 7
1 X X X X
8 0 8
0 X X X
9 0 9
1 X X
A 0 a
4 X
B 0 b
9
C 0 c
D 0 d
E 0 e
F 0 f
C D E F
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
X X X X
0 X X X
9 X X
4 X
1
2.7. PROBLEMS
19
6. A stful of (binary) digits

Your imaginary computer represents integers using 5 and only 5 bits! Fill in the blank
spots in the following table. The signed integers follow the twos-complement rule
discussed in class.
Binary Hexadecimal
00000
00
00001
01
00010
02
00011
03
00100
04
00101
05
00110
06
00111
07
01000
08
01001
09
01010
0A
01011
0B
01100
0C
01101
0D
01110
0E
01111
0F
10000
10
10001
11
10010
12
10011
13
10100
10101
10110
10111
11000
11001
11010
11011
11100
11101
11110
11111

0
0
1
1
2
2
3
3
4
4
5
5
6
6
7
7
8
9
10
11
12
13
14
15
16
17
-3
-2
-1
20

7. These digits will give you the dgets.
Your imaginary computer still represents integers using 5 and only 5 bits! The result of
an arithmetic operation with two integers always results in a 5-bit result. Higher-order
bits are lost.
Do your calculations in binary arithmetic and SHOW YOUR WORK! Quote your
results of your calculations in binary, signed integer (2s-complement scheme), unsigned
and hexadecimal integer representations. You may use the table that you completed
in the previous question.
(a) What is the sum of the unsigned integers 3 and 4?
(b) What is the sum of the unsigned integers 16 and 17?
(c) What is the sum of the signed integers -2 and -3?
(d) What is the sum of the signed integers -16 and -16?
(e) What is the product of the unsigned integers 3 and 4?
(f) What is the product of the signed integers -2 and -3?
(g) What is the product of the signed integers -16 and -16?
2.7. PROBLEMS
21
8. Bits and bytes? Again!

In the following table, the column 1 represents bit patterns on a 6-bit computer.
(a) In the 2nd column, write in the complement of the bit pattern in column 1
(b) In the 3rd column, write in the twos complement of the bit pattern in column 1
(c) In the 4th column, write in the twos complement of the bit pattern in column 3
(d) In the 5th column, put a check mark if the bit pattern in column 1 represents a
positive signed int.
(e) In the 6th column, put a check mark if the bit pattern in column 1 represents an
odd unsigned int.
(f) In the 7th column, put a check mark if the bit pattern in column 1 represents an
unsigned int that is divisible by 2.
(g) In the 8th column, put a check mark if the bit pattern in column 1 represents an
unsigned int that is divisible by 4.
(h) If the bit pattern in column 1 represents an odd unsigned int, write its unsigned
int representation in column 9.
Column 1 Column 2 Column 3 Column 4 Col 5 Col 6 Col 7 Col 8 Column 9

000010
111001
110110
000001
001101
110010
010010
110110
101111
101010
22

9. A bit challenging
In the following table:
1) Convert the rst 5 32-bit bit patterns to hex
2) indicate which bit patterns are divisible by 2 (/2)
3) indicate which bit patterns are divisible by 4 (/4)
4) indicate which bit patterns represent positive or negative integers in twos-complement
representation (+/-?)
5) indicate which bit pattern represents the largest integer in twos-complement representation (>?)
6) indicate which bit pattern represents the smallest integer in twos-complement representation, negative with the greatest magnitude (<?)
1111
1000
1011
0010
0010
1110
1111
0110
0011
0000
0110
1010
0000
0011
1100
1101
1101
1111
0100
0111
1101
1000
0100
0011
1110
1001
1011
1001
0001
0010
0001
0001
1000
1100
0111
0111
1110
1110
1111
1000
0010
1100
0011
0001
0011
0011
1101
1011
1001
1101
0100
1110
1111
1000
1111
1010
32-bit bit patterns

0010 0110 1110 0101
0000 1100 1000 0010
1000 1011 0100 1111
1101 1001 0100 1101
1010 0001 0010 1010
1000 1000 0011 0101
0000 1100 0011 1010
0100 0000 0101 1000
0000 0011 1100 0110
1101 1001 1111 1011
0001 1010 0010 1111
0000 1110 1111 1010
0011 0010 0100 1100
1010 0111 1011 1011
0100 1011 0110 1011
1011 1110 1011 1010
1110 1011 0100 0101
0001 0111 1110 1110
1110 1110 0000 0011
1010 1111 0100 0010
1101 1111 1101 1101
1010 0101 0110 1010
0110 1111 0001 0010
1010 1100 1100 0111
0111 1100 1001 0101
1010 0101 1000 0010
1000 0100 0011 0101
0000 0011 1111 1010
hex
0100
1011
0000
0111
0101
1110
0010
1011
0000
0101
0110
0010
0001
0111
1111
0000
0100
0011
1011
0010
1010
1111
0111
0010
0001
1011
0011
0110
0110
1011
0111
1110
0111
1101
X
1101
X
0000
X
1101
X
1000
X
1000
X
0011
X
0101
X
0000
X
0011
X
0001
X
1111
X
0000
X
1011
X
0101
X
1101
X
0110
X
0001
X
0111
X
1101
X
1010
X
0111
X
1111 9A03FA6F
/2? /4? -/+? >? <?
Chapter 3
Algorithms and Pseudocodes
In this chapter we introduce algorithms and ways with which algorithms can be represented.
At the end of the chapter, we also talk a little about computer architecture, as a way of
illustrating the operations that can happen inside a computer, when a part of an algorithm
is executed.
3.1
What is an algorithm?
An algorithm is a set of well-dened, unambiguous steps or instructions to be carried out in

sequence to accomplish some task.
An algorithm usually has a START point and a STOP point. Certainly any algorithm
we shall encounter in this book will. The START may require some inputs and the
STOP may provide some output.
An algorithm is made of individual instructions.
An instruction consists of a well-dened operation usually on some input (what the
instruction receives) and usually producing output (what the instruction produces).
Each instruction is well-dened and its outcome predictable if the instruction operates
on valid input.
There is a direction of logic ow or sequencing. Once an instruction is executed, it
passes control to another instruction.
There can only be a nite set of instructions.
When executed (with valid input) an algorithm is guaranteed to terminate in a sensible
way.
23
24
3.2
CHAPTER 3. ALGORITHMS AND PSEUDOCODES
Flowchart representation
START
STEP 1
STEP 2
STEP 3
STOP
The above gure is a representation of a simple sequence of 3 steps with a beginning and an
end. The START and STOP are represented by circles. A sequence step is represented by
a rectangle. The logic ow is indicated by lines and arrows. This kind of representation is
called a owchart or schematic diagram. It looks a lot like a circuit diagram in electronics.
3.3
Pseudocode representation
An algorithm is a well-dened description of actions and their ordering. Pseudocode is another way of expressing the actions and ordering to be taken place on a computer in an
exact way. Yet, pseudocode does it in an informal way, without getting bogged-down in the
details or syntax of a particular computer language like C++, MATLAB or something else.
Two programmers should be able to agree on pseudocode and write functionally equivalent
computer programseither in the same computer language or in dierent ones. Pseudocode
can be written in any human language in the form of an ordered (or numbered) list. For
example, a pseudocode representation of the above owchart would be:
START = Step 1 = Step 2 = Step 3 = STOP
A more conventional representation has the instruction ow (also called logic ow) going
from top to bottom:
3.4. DECISIONS, CONDITIONALS, BRANCH INSTRUCTIONS
25
START
Step 1
Step 2
Step 3
STOP
3.4
Decisions, conditionals, branch instructions
A conditional instruction or statement, also called a decision instruction, a branch or decision statement, is one in which a decision is made and the logic ow can follow dierent
paths, depending on the answer. In pseudocode it can be represented as follows:
IF such-and-such THEN DO
this-and-that
OTHERWISE DO
something-else
One could do the same thing with a owchart, and it would look like:
YES
THIS
&
THAT
SUCH
&
SUCH
NO
SOMETHING
ELSE
Note that we have introduced a new shape to handle the case where, depending on the input,
a decision is made and the logic ow can follow two dierent paths. This is represented by a
diamond shape. A branch statement or a decision statement is characterized by only one
path in and always two paths out.
3.5
Looping or jump instructions
A nal ingredient is needed before we complete our abstract discussion of algorithms the
ability for a statement to transfer control not to the next statement, but somewhere else in
26
the algorithm. This is called a looping or jump instruction.

Consider the following pseudocode:
ITEM 3
.
.
a set of other steps
.
.
JUMP TO: ITEM 3
The stu between ITEM 3 and JUMP TO: ITEM 3 inclusive, denes a loop. This is
represented in a owchart as follows with the direction of logic ow made clear with the lines
and arrows.
ITEM 3
JUMP TO
ITEM 3
3.6
Sequencing, branching and looping
It can be shown that the 3 constructs of sequencing, branching and looping, are all that one
needs to construct any computer algorithm! Thus, the theory of algorithmic ow boils
down to understanding just these three things. There are many, many ways of expressing
these three things in a computer language like C++ or Matlab. But remember that it all
boils down to just sequencing, branching and looping. These are the three pillars of
algorithmic design!
Of course there is a lot of other stu to be considered to make a computer language useful.
Well see a lot of this in the course. However, as far as the ow of logic goes, we have done
it all at this point! You can now go and write an algorithm to do anything you want and
express it in either pseudocode or owchart form.
3.7. A MINI-SUMMARY BEFORE THE EXAMPLES
3.7
27
A mini-summary before the examples
PSEUDOCODE: The language and organization of pseudocode is not rmly established.

Make any convention for yourself that makes your algorithm clear. Note the use of spacing
and indenting, bold and italic fonts, all ways of emphasizing meaning.
FLOWCHARTS: It is also possible to write pseudocode diagrammatically in terms of ow
charts that represent the ow of computer instructions, or logic ow. Here there are some
accepted (but not rigorously adopted) standards. These charts look very much like circuit
diagrams. It is considered good practice to describe every variable you use and provide a
well-dened starting point and ending point.
Pseudocode and ow charts are indispensable aids in getting a computer program up and
running with a minimum of eort.
3.8
Some examples
28
3.8.1
The Breakfast algorithm, or, Bielajews Sunday Morning

Internationally-famous pancakes
1. Start of The Breakfast Algorithm

2. If the day is not Sunday, have your usual boring bowl of Granola. For added excitement,
slice in half a banana. Otherwise,
(a) In a big bowl, whisk together 1 and 1/2 cups white our and 2 heaping teaspoons
of baking powder.
(b) In a separate bowl, beat two whole eggs (no shells please) together for one minute
or so.
(c) Add 1 cup of milk, preferably skimmed, to the eggs and mix.
(d) Add 1 tablespoon of Canola oil, 3 tablespoons of maple syrup, and one teaspoon
of vanilla to the milk and eggs and mix.
(e) Add the wet stu to the dry stu.
(f) Take a big spoon and,
(g) Mix the stu in the big bowl one whole turn.
(h) If there are lots of lumps in the mixture , go back to step 2.(g), otherwise go to
step 2.(i). Technical note: Dont go overboard and make it completely smooth.
Leave it a little lumpy for texture.
(i) Slice in a Granny Smith apple (no peel) and (optionally) a half cup of frozen
cranberries (well rinsed).
(j) Add a little bit of Canola oil to an electric fry pan preheated to 350 F. Put in
just enough to coat the bottom of the pan.
(k) Put the pancake dough into the pan, making approximately four 34 diameter
circles. Turn the heat down to 300 F. Fry for exactly 3 minutes with the pan
covered, ip and fry for 3 more minutes with the pan uncovered.
(l) Take them out of the frying pan and place on a napkin to absorb excess oil.
(m) Eat with maple syrup and sliced strawberries on top. I DARE you to eat more
than 3! (It has never been done...)
3. Clean up your dishes.
4. End of The Breakfast Algorithm
This algorithm produces about 8 pancakes, enough for 46 people.
3.8. SOME EXAMPLES
3.8.2
29
Solve for x: Ax2 + Bx + C = 0 where A, B, C are arbitrary

constants
Denitions: x is an unknown real number, A, B, C are xed constants that can take on any
valid real value. We are trying to nd the values of x for which Ax2 + Bx + C is zero. If we
plot the function f (x) = Ax2 + Bx + C, a parabola, it can look as shown in the gure below,
for certain choices of A, B, and C. The x-axis can intersect the curve f (x) = Ax2 + Bx + C
only 0, 1 or 2 times, depending on the values of A, B, and C.
NO SOLUTION
X
1 SOLUTION
X
2 SOLUTIONS
X
First, we do the math in excruciatingly ne detail. It is essential to understand completely

the mathematics of what you are doing if you are to have a chance at writing a successful
algorithm.
Ax2 + Bx + C

B
C
A x2 + x +
A
A
B
C
x2 + x +
A
A
2
2
B
B
B
C
x2 + x +
+
A
4A2 4A2 A
= 0
= 0
= 0
= 0
N.B.
A = 0
30

B2
B
x+
A
4A2
B2
B
x2 + x +
A
4A2

B 2
x+
2A
B
x+
2A
x2 +
=
=
=
=
x =
x =
B2
C
2
4A
A
B 2 4AC
4A2
B 2 4AC
(2A)2
B 2 4AC
N.B. B 2 4AC 0
2A
B
B 2 4AC
2A
2A
2
B B 4AC
A = 0, B 2 4AC 0
2A
So, under some circumstances, (A = 0, B 2 4AC > 0), we have two solutions:
B + B 2 4AC
B B 2 4AC
and x =
x=
2A
2A
corresponding to the last case in the preceding gure.
What if A = 0, B 2 4AC = 0? We have one solution:
x=
B
2A
corresponding to the middle case in the preceding gure.

What if B 2 4AC < 0? No real solution. This corresponds to the rst case in the preceding
gure.
What if A = 0? Re-analyze!
Bx + C = 0 (y = Bx + C Is an equation for a straight line!)
C
x =
N.B. B = 0
B
This is the place where a straight line crosses the x-axis.
What if A = 0 and B = 0? No solution for x! A straight line with zero slope does not
intercept the x-axis. The equation is nonsense unless C = 0 as well. If A = B = C = 0 it is
correct but not usually interesting!
Now we give the pseudo-code for this example, the solution to the quadratic equation.
START A, B, C are inputs, known to be real, xed constants and x is the output, expected
to be real but unknown at the start
3.8. SOME EXAMPLES

IF (B 2 4AC < 0)
PRINT (No solution because B 2 4AC < 0)
STOP
ELSE (i.e. B 2 4AC 0)
IF (A = 0)
IF (B 2 4AC = 0)
x = B/(2A)
PRINT (One solution: x, because B 2 4AC = 0)
STOP
ELSE (i.e. B 2 4AC > 0)
x1 = (B + B 2 4AC)/(2A)
x2 = (B B 2 4AC)/(2A)
PRINT (Two solutions are: x1 and x2 )
STOP
ELSE (i.e. A = 0)
IF (B = 0)
x = C/B
PRINT (Only one solution x, because A = 0)
STOP
ELSE (i.e. B = 0)
PRINT(No solution because A = 0 and B = 0)
STOP
END OF ALGORITHM
Note:
IFs can be nested
Use of indenting
Multiple STOP-points!
Not every piece of pseudocode is used every time
If enough dierent A, B, Cs are tried, every piece of pseudocode will be used
It will work for any value of A, B, C. The algorithm is robust
Can represent in terms of a graphical ow chart
This is the ow chart for the quadratic formula:
31
32
START
A,B,C ARE FIXED

X IS UNKNOWN
IS
B2-4AC >- 0
?
NO
SOLUTION
STOP
Y
IS
A =/ 0
?
IS
B =/ 0
NO
SOLUTION
STOP
IS
B2-4AC = 0
?
X = -C / B
X = -B / (2A)
PRINT X
PRINT X
STOP
B2-4AC
X 1 = -B +
2A
B2-4AC
X 2 = -B -
2A
STOP
PRINT ( X 1 , X 2 )
STOP
Note that, although there is no reason to expect it, the owchart designer expects that
there will be two real solutions every time. This is reected in the owchart by this case
running vertically down the page, with the exceptions hanging o to the right. This makes
the owchart particularly easy to read.
3.8. SOME EXAMPLES
3.8.3
33
Iteration: A summing loop
Before you read this section, if you have never had any experience with programming, the
statement S = S + 1 should justiably seem to be completely wrong, since mathematically
it is equivalent to saying 0 = 1! Just cancel the Ss from either side! So, if this is the case,
kindly go and read the next section An aside on computer architecture and then come
back. Although mathematical statements and computer code statement may look the same,
they are completely dierent. Mathematics expresses an idea, while a line of computer code
represents an operation, or a set of individual operations.
Problem: Sum the positive integers 1, 2, 3 N up to N.

A dumb way that works:
START N is an integer (xed) and S is an integer (unknown)
S = 1
S = S + 2
S = S + 3
.
.
.
S = S + i (for the ith statement)
.
.
.
S = S + N
PRINT(S)
END OF ALGORITHM
This is the ow chart for the dumb way:
34
START
N FIXED NUMBER
S IS UNKNOWN
S=1
S=S+2
S=S+3
.
.
.
S=S+i
.
.
.
S=S+N
PRINT S
STOP
N statementslots of repetition, very long if N is large.

A better way:
START N is an integer (xed), S is an integer that will contain the sum . At the start, S is
initialized to 1, that is, its starting value will be 1. Another integer, i, is employed to
contain the value that is to be added to S. It is initialized to 1.
Start of counting loop:
i = i + 1 (get the new value for i)
S = S + i (add it to S)
IF i = N jump to End of counting loop:
3.8. SOME EXAMPLES
35
Jump to the rst statement in this loop

End of counting loop:
PRINT(S)
END OF ALGORITHM
Only a few statements to express the same idea
Loops can be nested as well
Can represent in terms of a graphical ow chart
This is the ow chart for the better way:
START
N FIXED NUMBER
i = 1; S = 1
i=i+1
S=S+i
NO
i=N?
YES
PRINT S
STOP
There is an even better way! There is a mathematical solution to this that was discovered
by Gauss (Karl Friedrich Gauss (17771855 AD)) as a schoolboy!
S=
N(N + 1)
2
36
If N is even:
[1, 2, 3, ..., (N 2), (N 1), N]

[1, 2, 3, ..., (N/2)] + [(N/2 + 1), (N/2 + 2), (N/2 + 2), ..., N]
[1, 2, 3, ..., (N/2)] + [N, (N 1), (N 2), ..., (N/2 + 1)]
[(N + 1), (N + 1), (N + 1), ..., (N + 1)] (N/2 terms)
N
=
(N + 1)
2
S =
=
=
=
quod erat demonstratum

For odd N the proof is left to the reader.
3.8.4
Iteration: A product loop
Problem: Compute P = k N .
A very, VERY, dumb way (but one that works):
START k is an integer or a real (xed), N is an integer (xed) and P is an integer or a real
number (unknown)
P = k
P = P k
P = P k
.
.
.
P = P k (for the ith statement)
.
.
.
P = P k (for the Nth statement)
PRINT(P)
END OF ALGORITHM
Here is the ow chart for the very, VERY, dumb way:
3.8. SOME EXAMPLES
37
START
N FIXED NUMBER
k FIXED NUMBER
P IS UNKNOWN
P=k
P=P*k
P=P*k
.
.
.
P=P*k
.
.
.
P=P*k
PRINT P
STOP
N statementslots of repetition of the same statement

A better (but still dumb) way:
START k is an integer or a real (xed), N is an integer (xed), P = 1 is a number that
will contain the product (initialized to 1) and i is an integer (initialized to 0) used as
an index or counter to indicate how many times we have passed through the loop.
Start of loop:
38

i = i + 1 (increment the counter)
P = P k (accumulate the sum)
IF i = N jump to End of loop:
jump to Start of loop:
End of loop:
PRINT(P)
END OF ALGORITHM
START
Here is the ow chart for the better way:

N FIXED NUMBER
k FIXED NUMBER
i = 0; P = 1
i=i+1
P=P*k
NO
i=N?
YES
PRINT P
STOP
Only a few statements to express the same idea.

A much, MUCH better way: Al-Kashi (1390 - 1450 AD)
alKashi, Ghiyath alDin Jamshid Masud
Born in Kashan (Iran), died in Samarkand (now called Uzbekistan)
Self-proclaimed inventor of decimal fractions
Calculated to 16 decimal places
Theory of numbers and computation
First discovered the binomial theorem
Inventor of a calculating machine to predict lunar orbits
Discovered the following algorithm in 1414 AD:
3.8. SOME EXAMPLES
39
START k is an integer or a real (xed), N is an integer (xed), P = 1 is a number that

will contain the product (initialized to 1)
Start of loop:
IF (N is even)
N = N/2
k = k k
ELSE (i.e. N is odd)
N = N - 1
P = P k
IF (N < 1) jump to End of loop:
ELSE (i.e. N 1) jump to Start of loop:
End of loop:
PRINT(P)
END OF ALGORITHM
This is the ow chart for the Al-Kashi Method:
START
N FIXED
k FIXED
P = 1 (PRODUCT)
N>
- 1?
(N < 1)
NO
PRINT P
YES
YES
N=N/2
k=k*k
N even
?
NO
N=N-1
P=P*k
Danger, danger: The algorithm redenes N and k.
STOP
40
3.9
3.9.1
An aside on computer architecture

What does S = S + 1 mean?
If youre doing math, it means nonsense! It is the same as saying 0 = 1!

If youre writing a computer program it means something else and describing what is means
requires a small aside on computer architecture, the real nuts and bolts of what is going
on inside a computer.
First some denitions. See the gure below.
MEMORY The memory is the physical place in the computer where bit patterns are stored.
Memory is best thought of as a stack of words, 32-bit bit patterns.
VARIABLE A symbol for a certain bit pattern (a 32-bit integer, say) that the programmer
can dene and change at will in a program. The symbol (S in this case) really refers
to the location in memory where the bit pattern is stored.
CONSTANT A symbol for a certain bit pattern that the programmer can dene and use
but not change.
CPU The Central Processing Unit in a computer. The CPU is the boss. It follows the
steps in a program and tells all the other parts of a computer what they should be
doing. It is the controller.
REGISTER A special place in the CPU where bit patterns are stored.
ADDER A functional unit in a computer (usually located right on the CPU chip) that
performs additions on 32-bit bit patterns.
MEMORY
ADDER
CPU
R1 R2 R3 R4
00000000000000000000000000000000
S: 00000000000010010000100000100000
00000000000000000000000000000000
00000000000000000000000000000000
C: 00000000000000000000000000000001
00000000000000000000000000000000
00000000000000000000000000000000
3.10. PROBLEMS
41
In a computer program, the statement: S = S + 1 gets translated into the following

operations that the CPU performs or controls.
FETCH S,R1 Copy the bit pattern that is stored at the location referred to by S and put
it in the 1st CPU register R1.
FETCH C,R2 Copy the bit pattern that is stored at the location referred to by C and put
it in the 2nd CPU register R2. Previously in the program, the constant C would have
to have been dened and the bit pattern for C (00000000000000000000000000000001)
would have to have been stored in C.
ADD R1,R2,R3 Copy the constants of the bit patterns stored in R1 and R2 into the
ADDER, do the binary arithmetic and copy the result into the 3rd CPU register R3.
STORE R3,S make a copy of the bit pattern in the 3rd CPU register R3 and place it in
the memory location referred to by S.
So, just the simple increment (add by 1) of a variable results in 4 operations performed by
the CPU. The preceding steps are exactly what an assembly language program would have
to do to interpret the statement S = S + 1. Remember that when we write statements like
S = S + 1 it is really a shorthand for a set of steps that a computer has to carry out. We
will be clear in the future as to what is math and what is computer language. Math script
will be used for math (e.g. S = S +1) while type script will be used for computer instructions
(e.g. S = S + 1).
3.10
Problems
1. The trinity of algorithmic design

What are the three most important capabilities that are needed to construct algorithms?
The ability to declare variables
The ability to assign values to variables
The ability to pass control from one operation to another in a prescribed way
A way of making a True or False decision
The existence of ints and floats
The ability to do mathematical operations (e.g. +, - )
The ability to go back to a previous operation
42

A compiler that can convert human-readable text into a language (such as
binary) that a computer can understand
2. What is an algorithm?
In 35 words or less, answer the question, What is an algorithm?.
The three pillars of algorithm design are 1) sequencing, 2) branching, and 3) looping.
In 25 words or less for each, describe what they mean.
That is:
What is sequencing? (25 words or less)
What is branching? (25 words or less)
What is looping? (25 words or less)
4. Toss your cookies at this one!
In this Chapter, a recipe for Bielajews Sunday Morning Internationally-famous pancakes was givenin the form of pseudocode. Describe another recipe for a dish that
you likein the form of pseudocode. It has to have a Start, a Stop, more than 5 Steps,
at least one Decision (or branch) point, and at least one Loop.
5. In the 70s, man, if you were not hip, you were ...
Here is an algorithm in pseudo-code:
START
OUTPUT Enter a positive integer
INPUT N
i=1
WHILE (i i < N)
i=i+1
END WHILE
IF (i i = N)
OUTPUT YES
ELSE
OUTPUT NO
END IF
STOP
For which of the following inputs will the algorithm say YES?
(a) 28
(b) 36
3.10. PROBLEMS
43
(c) 54
(d) 81
What property of the input is the algorithm determining?
6. Summing the squares
Describe an algorithm in either pseudocode or owchart form, to sum the squares of
the integers up to and including some N, where N is an input to the algorithm. That
is,
S = 12 + 22 + 32 + + N 2
where S is the sum. You must express your algorithm in terms of a loop that is traversed
N times. (There is a closed mathematical expression S = N(N + 1)(2N + 1)/6 that
you must not exploit in your algorithm. Pretend the closed mathematical expression
does not exist.) Make sure to declare and initialize all the variables you use.
7. Dont you love triangles?
A number N is called triangular if you can take N little asterisks and make an
equilateral triangle from them. For example, here are the rst few triangular numbers:
*
* *
3
*
* *
* * *
6
*
* *
* * *
* * * *
10
*
* *
* * *
* * * *
* * * * *
15
*
* *
* * *
* * * *
* * * * *
* * * * * *
21
Here is an algorithm that is attempting to detect triangularity in the input. But it has
a bug. Fix the algorithm.
START
INPUT N
i=1
S=0
WHILE (S < N)
i=i+1
S =S+i
END WHILE
IF (S = N)
OUTPUT Triangular
44

ELSE
OUTPUT Not triangular
END IF
STOP
8. De-Gauss this!
In this chapter we learned Mr. Gausss formula for the computation of the Nth triangular number:
1 + 2 + 3 + ... + N =
N(N + 1)
2
Here is another algorithm attempting to detect triangularity of the input. Do you

think this one is OK? If not, x it.
START
INPUT N
i=1
WHILE [(i (i + 1))/2 < N]
i=i+1
END WHILE
IF [(i (i + 1))/2 = N]
OUTPUT Not triangular
ELSE
OUTPUT Triangular
END IF
STOP
9. The oscillating sign

Write an algorithm in pseudocode that accepts a positive integer N and outputs a 1
(one) if N is even and -1 (minus one) if N is odd. Hint: -1 multiplied by itself an even
number of times is +1, while -1 multiplied by itself an odd number of times is -1.
10. Summing an oscillating series
Describe an algorithm in both pseudocode and owchart form, to perform the following
sum
S = 1 x + x2 x3 + x4 + (x)N
3.10. PROBLEMS
45
where x and N are inputs and S is the sum, an output. You must express your
algorithm in terms of a loop that is traversed N times. Express your algorithm in
terms of pseudocode and a owchart.
11. Lets talk about... the factorials of life
The factorial occurs in many places in mathematics and engineering applications and
life in general. The symbol for the factorial is the exclamation (!) mark. So, we would
call 3!, three factorial. Factorials are easy to calculate.
1! = 1
2! = 2 1
3! = 3 2 1
4! = 4 3 2 1
and so on.
If N is a whole number,
N! = N (N 1) (N 2) 2 1.
Note that N! = N (N 1)!, a neat property of factorials.
You will describe an algorithm to calculate the factorial of N where N is any whole
number greater than 1. You will describe an algorithm as if you were going to program
it on a computer, that is, you will specify a set of instructions that even a machine
can understand. The algorithm should include instructions on how to start and end.
You can assume that the machine on which the algorithm will be implemented can
store variables, loop, do branches, and do simple arithmetic. Express your algorithm
in terms of pseudocode and a owchart.
12. Decay and dissipation
You will describe an algorithm to calculate an approximation to the exponential function, ex . The following sum is a really good approximation to the exponential function,
ex , as long as the right hand side contains enough terms.
S =1
x x2 x3 x4
+
+
+
1! 2!
3!
4!
Let x and N, the number of terms on the right hand side, be inputs to the algorithm
and S, the sum, be the output. Note the use of the factorials in the above sum. Express your algorithm in terms of pseudocode and a owchart. Note that the algorithm
to raise x to a power is given in Chapter 3, Section 5 of the book.
Hint: The most ecient way to solve this problem is to rewrite S in the following
fashion:
S = s0 + s1 + s2 + s3 + s4 +
46

where s0 = 1 and note that
sn =
x
sn1
n
13. Whats YOUR sine?

You will describe an algorithm to calculate the sine function, as if you were going to
program it on a computer, that is, you will specify a set of instructions that even a
machine can understand. The algorithm should include instructions on how to start
and end. You can assume that the machine on which the algorithm will be implemented
can store variables, loop, do branches, and do simple arithmetic.
The following sum is a really good approximation to the sine function, as long as the
right hand side contains more than 4 terms in the sequence shown:
S=
x3 x5 x7
x
+
1!
3!
5!
7!
Let x and the number of terms on the right hand side be inputs to the algorithm and
S, the sum, be the output. Note the use of the factorials in the above sum. Express
your algorithm in terms of pseudocode and a owchart.
14. The rich get richer, the poor get poorer!
The nancial investment rm of Getridge-Quigg, Ann Arbor, has a special deal for
students. If you deposit your money with them they will give you 0.5% interest per
month (a large amount these days) but then they will charge you a $5.00 per month
service charge applied against your account. If you have less than $5.00 in your account,
they take the remainder and your account is closed. So, if you deposited $1000.00 and
leave it there, your account will always have the same balance at the end of the month.
If you deposit more than $1000.00 and leave it there you may gain. Deposit less than
$1000.00 and your account may eventually be reduced to zero.
Write an algorithm that starts with an input of how much is to be deposited. You
intend to leave that money in the account and never add or withdraw from it. Your
algorithm will predict how many months it will take for you to double your money
or to reduce your account to zero. If the balance in your account never changes, you
have to state that as well. Note that Getridge-Quigg rst applies the interest and then
applies the service charge. After every calculation, Getridge-Quigg rounds the result
to the nearest penny. For example, the monthly interest on a balance of $1001.00 is
$5.01 while the monthly interest on a balance of $1000.99 is $5.00. You have to build
this fact into your algorithm.
Express your algorithm in terms of pseudocode and a owchart.
3.11. PROJECTS
3.11
47
Projects
1. Follow the bouncing ball

You will describe an algorithm as if you were going to program it on a computer, that
is, you will specify a set of instructions that even a machine can understand. Your
algorithm should specify its purpose, the input data (if any), the output data (if any),
and the set of steps that comprise the algorithm itself. The algorithm should include
instructions on how to start and end. You can assume that the machine on which the
algorithm will be implemented can store variables, loop, and do branches.
The problem is to determine the total distance a ball travels having been dropped from
a height h (and traveling a total distance h on its way to the oor) and rebounding to
a fraction of h which we will call ch (and traveling a total distance h + ch). c is called
the coecient of restitution. The ball will then drop to the oor (total distance =
h + 2ch) and bounce up to a height c2 h (total distance = h + 2ch + c2 h) and so on.
You can assume for simplicity that the ball is a point, that is, its diameter is exactly
zero (a mathematical ction to make our lives easier) and that c falls anywhere in the
region 0 < c < 1. Write an algorithm that determines the distance the ball travels
before coming to a rest. You can make any reasonable assumption you like about what
at rest means for the ball. However, the assumption has to be reasonable and
justiable either on the basis of numerics, mathematics or physics. This is a subtle
point that is discussed below and is discussed in class.
This problem has an exact mathematical solution:
d=h
1+c
1c
that you must not employ in your algorithm to determine the total distance traveled.
You have to obtain the solution by developing an algorithm that sums individually
every bounce of the ball.
Some discussion:
A ball made from putty will have c = 0 and an ideal ball (which does not really exist
in reality) would have c = 1. A fresh out-of-the-can tennis ball has c = 0.5 or so, an
old tennis ball has c = 0.25 or so. A ping-pong ball has c = 0.90 or so.
This problem is very much like Xenos paradox. The paradox is that mathematically
the total distance traveled by the ball is nite (as long as c is less than 1) but mathematically the total number of bounces is innite! However, after many bounces the
recovery distance becomes so small that accumulating them really does not matter in
any practical sense any more. If digital computers could really represent real numbers
to innite precision then it would be possible to write an algorithm that accumulates
the distance properly but it would never stop unless we introduced some articial means
of stopping the execution. Part of the purpose of this exercise is to get you thinking
48

about this paradox, to realize that computers can only provide an approximation to
mathematical ction and physical reality.
Chapter 4
Getting started in C++
4.1
Simple input/output (I/O): A rst program in C++
Here is a rst program in C++. You can type it in yourself, or download it from the web
from the lecture note distribution area for this class.
//File: goBlue.cpp
#include <iostream>
using namespace std;
int main(void)
{ cout << "Go Blue!\n";
// Multiple line statement
cout <<
"\n\n\t\"Go\t\t\\\n\t\t\t\tBlue!\"\a\a\a\r";
return(0);
}
There are several common features of every C++ program
Comments Comments start with /* and end with */. For example,
/* This is a comment */
49
50
CHAPTER 4. GETTING STARTED IN C++

Comments are put there for humans only. They do not aect or modify what the
computer does. The computer does not care if you comment or not. But humans
do, especially your Lab Instructors and other people who hand out grades or
pay your salary. Comments should explain what you are trying to accomplish and
how you are going about doing it. It is appropriate and usually correct to assume
that the person reading your code is a complete idiot and that you have to explain
in detail what you are doing. After you have forgotten how or why you programmed
your own code, you will nd your own comments very useful to understand what you
have done. The start of any program or program segment should begin with a few
lines of explanation, along with an identication of the programmer and the date it
was created. Put comments through the rest of the code wherever you think that the
reader (aka the idiot) might need some explanation. Less is not more in this case. But,
dont go overboard too far. When in doubt, comment, comment, comment.
Comments can begin and end anywhere and can span more than one line.
// This is also a comment, from the "//" to the end of the line
which is useful when your comment is only one line or you want a comment to start
somewhere in a line and continue to the end of the line.
Use both kinds of comments and make your code readable.
Preprocessor directives Preprocessor directives instruct the compiler in advance of the

actual compilation of your code. Preprocessor directives are indicated by a # beginning a line. In this case the preprocessor directive
#include <iostream>
instructs the preprocessor to include the contents of the ANSI standard library le
iostream. This header le contains other #include statements that call other library
les that dene the standards C++ input and output functions.
There are other common header les. We shall soon encounter cmath which is included
via the preprocessor directive
#include <cmath>
which denes a lot of the common math functions. There are other types of preprocessor directives called symbolic constants and macros which we shall encounter
later.
If you want to have a look at these library les, look in the directory called
/usr/include/g++-3
where they reside. Boredom alert! This is not for the faint-of-heart!
4.1. SIMPLE INPUT/OUTPUT (I/O): A FIRST PROGRAM IN C++
51
using namespace std; Tells the compiler that we will be using that standard namespace
in the le that contains the line
All this really means is that standard library les and denitions are being used. It is
possible to dene your own namespaces if you want to make sure that your variables
and denitions do not conict with either the standard ones, or someone elses, someone
working on the same project. For this course, we shall be using the standard namespace
for everything we do. If you go on to work in a professional programming environment,
writing software by teams, you will almost certainly be using your own namespaces.
int main(void){} All C++ programs begin processing at the rst executable statement in
main. The ()s indicate that main is a block of code called a function. The int in
front of main declares main to be a function that is expected to return an integer.
(More on this later.) main is a special function that the operating system recognizes
as the place to begin execution of a C++ program.
The lines of code in between the {}s is called the function body. It is considered good
programming practice to indent the text in the function body by a certain amount of
space. A common recommendation is 3 spaces which is reasonably pleasing to the eye.
We will adopt this convention for this course. In our example, the function body starts
with the statement
cout << "Go Blue!\n";
which instructs the C++ standard library function cout function to print the series
of characters Go Blue! (not including the double quotes) to the screen and then
positions the cursor at the beginning of the next line. This positioning is eected by
a special sequence of characters \n called an escape sequence. Escape sequences are
necessitated when special characters are required or to undo the meaning of special
characters, like the double quote () in the cout statement above. Here is a list of
some commonly used escape sequences employed in cout statements.
\n
\t
\r
\a
\\
\
Newline. Move cursor to beginning of next line

Horizontal Tab. Move cursor to next tab stop
Return. Move cursor to beginning of current line
Alert. Sound the bell
Print a backslash (\)
Print a double quote ()
As an extreme example contrived to show all the possibilities,
52

cout <<
"\n\n\t\"Go\t\t\\\n\t\t\t\tBlue!\"\a\a\a\r";
which throws 2 newlines, goes to the rst tab stop, prints a double quote followed
by Go, advances two more tab stops writes a backslash, throws a newline, goes to
the third tab stop and prints Blue! followed by a double quote, sounds the alarm
3 times (You will probably only hear one beep, depending on your computer.) and
issues a carriage return. Because a new line was not thrown, further keyboard input
could overwrite the Blue! text. This just goes to show how far a professor will go to
contrive an example. Here is what the output looks like:
"Go
red%
\
Blue!"
Note that the statement starting with cout is terminated with a semicolon (;). This
is a special character in C++ called a delimiter that signies the end of a statement.
There can be more than one statement per line in a C++ code or a single statement
can span several lines. Generally it is considered to be good programming practice
to put at most only one statement per line of code. If a statement is long, put it on
several lines, space things nicely, and indent the carry-over lines 3 spaces relative to
the rst line, like in the example above.
return(0); When this line is encountered, processing in the CPU returns to the operating
system. Since main is a function of the int type, an integer value must be returned
to the process that called it, in this case the O/S. A return value of 0 usually means
normal execution, by convention. A return other than 0 can be treated as an error
condition by the O/S. For example, if your program had prompted for user input and
the user typed in something that the program was not prepared to process, one could
signify this to the O/S by returning a 1, for example,
return(1);
You can have as many returns as you need in a C++ code, just as you can have many
stop points in an algorithm.
4.2
Compiling, linking, loading and running
Now that we have created a program, we have to tell the computer what to do with it! After
all, computers only understand binary code and our program is written (coded) in terms of
characters.
To compile the program, which we will call goBlue.cpp, the following statement on a unix
machine attempts a compilation:
4.2. COMPILING, LINKING, LOADING AND RUNNING
53
red% g++ goBlue.cpp

The C++ compiler is another program that converts text instructions into binary code that
a computer understands. Compilers have evolved over a long period of time. The original
computers (dating to the middle and late 40s) were actually programmed in binary code that
was processed by the computer directly. This was an extremely laborious process. Early on
it was recognized that it would be easier to program in a more human-compatible way, in a
text-based format, and convert this into machine code using a compiler.
The origins of C++ date back to the late 60s and early 70s to the C language. The original
intent for C was to develop a language for writing computer operating systems. Operating
system (OS) code is special software for allowing users and programs to interact with data
stored on a computer and instructs the operational units of a computer what to do with
this data. Later, in the 80s and 90s, C++ was developed as an evolution of the C language.
(Any C++ compiler will compile C code.) The number of improvements is very greattoo
many to mention.
C++ code is designed to be as hardware-independent as possible so that C++ code can
migrate between dierent computer architectures. C++ code is translated into assembler
or assembly language which represents a text version (i.e. human-readable) version of binary code. Assembly language is always machine dependent and addresses (speaks to) the
computer hardware and its subcomponents. C++ is a higher-level language that deals with
hardware and data in a more abstract way.
C and C++ have evolved beyond their original intent as OS-builders, although this role is
still very important. C and C++ are powerful general-purpose languages for most aspects
of computing. It is probably true that C and C++ are the most commonly-used computer
languages in the world. C++ is certainly the one that is taught the most. There are other
high-level languages for special purpose applications in business and numerical computation but probably C++ will supplant these as new business and numerical computation
applications evolve.
After an algorithm is developed through the conceptual, pseudocode and owchart process, a
C++ program (or other computer language program) is written. During a C++ compilation
and execution, several steps occur:
Edit A human writes a computer program, e.g. goBlue.cpp Any editor will suce. Common unix editors are vi, emacs, pico, nedit or kwrite. Use whatever you like. Your
lab instructor will inform you of your options and probably ask you to stick with one
of these.
Preprocess, precompile Preprocessor directives, like #include and #define are handled.
This can be inclusion of les or substitution of text throughout the C++ program.
54
Compile After the preprocessor, the compilation converts the code to object code also
known as binary code. There are actually several stages. The rst is the parsing stage
where the compiler checks the syntax of your code. If this stage is passed successfully
the next stage is the production of the assembly code from the C++ code which is
syntax-error free. Then the assembly code is converted to binary code, also called
object code.
Link After the object code is created the link stage is entered. At this point, references
to functions that may have been referenced in the C++ program are established.
These missing parts, in the form of object code, are either put in place directly where
referenced (this is called static linking) or their transfer locations are established so
that processing can transfer to these locations and be resumed. This is called dynamic
linking. The le created by the linker is called the load module. By default, the load
module is called a.out by the unix operating system. A user can override this name if
desired.
Load The loader takes the load module (usually located on disk) and puts it into memory in
a place where the operating system expects to see executable code. On a unix system
the loading phase is started when a user types the name of the load module resident
on disk.
Run In the run phase, the binary code in the memory-resident load module is transferred to
the central processing unit (CPU) for execution. The load module contains instructions
for the CPU that directs it to assign tasks to the various operational units of the
computer. The load module also contains data and references to data that are to
be operated on. The run phase is where conceptual errors in a computer program
are exposed. Code that has incorrect syntax, and syntax errors are quickly found by
the compilers parser, are usually easy to correct. Code that has correct syntax but
incorrect design are much harder to nd. This is where pseudocode and ow charts
can come to the rescue.
This model of editing, preprocessing, parsing, object-code generation, linking, loading
and running is followed by all computer code-development environments, from the most
simple Go Blue! C++ program to the most complex, like the telephone management
system, international banking systems and the national security systemsirrespective
of the language it is written in. The following gure includes not only the code compilation process, but also describes the whole process of code development. It is general,
and mostly independent of the computer language being employed.
4.3. DECLARING AND INITIALIZING VARIABLES
55
THINK
PSEUDO CODE
FLOWCHART
BAD DESIGN
TYPE IT IN / EDIT
STEAL AS MUCH
AS POSSIBLE
1) PRECOMPILING
COMPILE
FAILURE
2) PARSING
3) CREATES OBJECT
LINK / CREATE
LOAD MODULE
RUN
FAILURE
REFINEMENT
SUCCESS!
4.3
Declaring and initializing variables
Consider the following C++ program that does mathematical operations with two integers:
//File: integerMath.cpp
#include <iostream>
56

int main(void)
{ int i1 = 5; // Define and initialize
int i2 = 2; // Define and initialize
int iResult; // Define only
iResult = i1 + i2; // sum
cout << "5 + 2 = " << iResult << "\n";
iResult = i1 - i2; // difference
cout << "5 - 2 = " << iResult << "\n";
iResult = i1 * i2; // product
cout << "5 * 2 = " << iResult << "\n";
iResult = i1 / i2; // division
cout << "5 / 2 = " << iResult << "\n";
iResult = i1 % i2; // remainder
cout << "5 mod 2 is " << iResult << "\n";
return(0);
}
There are some new features of this program.
The statement
int i1 = 5;
is a declaration statement. The name i1 is an identier for the variable i1. This declaration
statement species that i1 is an int, a signed integer. It also provides an initialization for
i1, giving it the value of 5.
An identier is a series of characters you can type at a keyboard that can consist of letters,
digits, and underscores ( ). It is a bad idea to start an identier with an underscore, even
though it is legal, because most identiers in libraries accessed by C++ use this. An identier
can not start with a digit, however. According to the draft ANSI C++ standard, an identier
may be any length. ANSI [American National Standards Institute] C++ is a built upon
ANSI C, and in ANSI C only the rst 31 characters are required to be recognized by C
compilers conforming to the ANSI standard for C. Dierent C++ compiler writers determine
how many characters in an identier are signicant. I recommend NOT using more than
4.4. INTEGER MATH IN C++
57
31, for safety. Beyond the limit of 31 or greater, your compiler may treat the characters
beyond it as signicant or insignicant (i.e. ignores them). For absolute safety, use 31
characters or less. Identiers in C++ are case sensitive so that i1 and I1 are recognized
as dierent variables. Unless your program is very short, you should name your identiers
in some recognizable wayit cuts down on the amount of commenting you have to do. I
also like humpback notation, numberOfStudents, rather than number_of_students. Note
that humpback identiers start with lowercase letters (usually) and new words within the
identier start with an uppercase letter. I nd underscores in the middle of identiers ugly,
wasteful and hard to type.
The statement
int i2 = 2;
is another declaration statement that identies the name i2 and declares it to be a signed
integer initialized to the value 2.
The statement
int iResult;
is another declaration statement that identies the name iResult and declares it to be a
signed integer. In this case the variable is not initialized.
4.4
Integer math in C++
The statement
iResult = i1 + i2;
contains two mathematical operators, = and +. The operator = is called the assignment
operator . Do not call it the equal sign! The assignment operator assigns the result of
the operation(s) to the right of it to the value of the identier to the left of it, iResult in
this example. = is called a binary operator because it operates on only two quantities, the
variable to the left which gets the assignment and the quantity to the right that provides
the assignment (after the other mathematical operations are completed).
The operator + is called the addition operator . It is also a binary operator and provides the
mathematical sum of the quantities to its left and right.
There are 4 other binary operators to consider now, the subtraction, -, the product, *, the
division, /, and the modulus, %. Of these the division requires special mention with regards
58
to its operation with integers. The integer division returns the integer part of the result. For
example, 5/2 = 2, not 2.5. The remainder of a division can be obtained from the modulus
function %. Thus 5 % 2 = 1.
The various mathematical operations, when employed within the same statement, follow
certain ordering rules. These are called the rules of precedence. The *, / and % operations
take precedence over + and -. *, / and % have the same precedence, and + and - have the
same precedence as well. For operators with the same precedence, the order of operation is
left to right. For example,
1 + 10 * 5 - 9 / 3 = 1 + 50 - 3 = 48
If a programmer desires a dierent interpretation of the above statement or wishes to make
the operation clearer, parentheses may be employed. Parentheses take precedence over *, /,
%, + and -. Inner parentheses are evaluated rst. For example,
(1 + (10 * 5) - (9 / 3)) = 1 + 50 - 3 = 48
is operationally equivalent to the above. However,
((1 + 10) * (5 - 9)) / 3 = (11 * (-4))/3 = -44/3 = -14
is quite dierent.
Summarizing, the order of precedence is
Order
0
1
2
3
Symbol
Operation
Comments
()
parentheses
inner ones evaluated rst
(unary) negation (unary minus)
RL evaluation
*/%
product, division, modulus
LR evaluation
+addition, subtraction
LR evaluation
The statement
cout << "5 + 2 = " << iResult << "\n";
prints the characters 5 + 2 = to the screen then the value of iResult and then sends
the newline escape sequence. There are more exible ways of printing things to the screen
and we shall encounter these shortly. Note how we can cascade the << operator for cout.
An equivalent way of accomplishing the same thing would be:
cout << "5 + 2 = ";
cout << iResult;
cout << "\n";
4.5. FLOATING POINT MATH IN C++
59
which is a little harder to read. Break up your output statement like this only for neatness
and readability.
4.5
Floating point math in C++
Consider the following C++ program that does mathematical operations with two oats:
//File: floatMath.cpp
#include <iostream>
int main(void)
{ cout << "Input the 1st floating point number: ";
float f1; // Declare the 1st float
cin >> f1; //and read it in
cout << "Input the 2nd floating point number: ";
float f2;
cin >> f2;
float fResult = f1 + f2; // sum
cout << f1 << " + " << f2 << " = " << fResult << "\n";
fResult = f1 - f2; // difference
cout << f1 << " - " << f2 << " = " << fResult << "\n";
fResult = f1 * f2; // product
cout << f1 << " * " << f2 << " = " << fResult
<< "\n";
if (0 == f2)
{ cout << "Division by zero not permitted!\n";
return(1);
}
else
{ fResult = f1 / f2; // division
cout << f1 << " / " << f2 << " = " << fResult << "\n";
return(0);
}
}
60
There are some new features in the above code.

float f1;
declares the existence of a oating-point number identied as f1. A oat is a representation
of numbers like 1.0, 2.5, 0.3333... in (usually) 32-bit representation. How a computer does
this we will leave as a mystery for the time being. However, the representation of most
oating point numbers is inexact, about 1 part in 107 for oats and about 2 parts in 1016 for
a double, a 64-bit representation of real numbers. Float and doubles also have maximum
and minimum values dened and this is architecture dependent, that is, on the computer
you are using. Common dynamic ranges for oats are between about 1038 and 1038 and
about 10308 and 10308 for doubles. Well have a lot more to say about oats as we get
deeper into the course. Floats and doubles are pretty much central to the issue of computing
in engineering and we will discuss the concepts as we require them.
Note that in C++ we can declare variables anywhere. Many programmers like to declare
them just before using them for the rst time.
The statement:
cin >> f1;
inputs the oat f1 from the keyboard.
4.6
The if/else if/else construct
The if/else if/else construct is one of the most important programming features of C++
and common to many computer languages. It has a single-choice form called if, a doublechoice form called if/else and multiple-choice forms if/else if and if/else if/else.
The general form of the if/else if/else is:
if (logical expression 1 )
{
STATEMENT BODY 1
}
else if (logical expression 2 )
{
STATEMENT BODY 2
}
else if (logical expression 3 )
{
STATEMENT BODY 3
4.6. THE IF/ELSE IF/ELSE CONSTRUCT
61
}
.
.
.
else if (logical expression N + 1 )
{
STATEMENT BODY N + 1
}
else
{
STATEMENT BODY N + 2
}
There are many forms of the if/else if/else construct:
The else if (logical expression) { STATEMENT BODY } is optional
The else { STATEMENT BODY } is also optional, independent of the above
If the STATEMENT BODY consists of one and only one statement, then the surrounding
{}s are optional. (It can help with clarity of programming to keep them, however.
Their use is highly recommended.)
Note the use of indenting, optional but highly recommended
At most one and only one of the STATEMENT BODY s can execute within an if/else
if/else construct can be executed
Processing control passes to the statement immediately following the if/else if/else
construct after execution of one of the STATEMENT BODY s.
If the if/else if or if form is used, it may happen that none of the logical conditions is satised. In this case processing control passes to the statement following the
construct.
if/else if/else constructs can be nested within each other
This is the ow chart for the if construct:
62
?
F
This is the ow chart for the if/else if construct:
F
?
?
T
T
4.7. LOGICAL EXPRESSIONS
63
This is the ow chart for the if/else if/else construct:
F
?
?
T
T
4.7
Logical expressions
So, what is a logical expression???

A logical expression is something that evaluates to TRUE or FALSE. FALSE has a numeric
value of the integer 0. TRUE has a numeric value of integer non-zero.
For example:
if (1)
cout << "This would always print,\n";
else if (0)
cout << "whereas this would NEVER print\n";
Here are some of the logical operators and their order of precedence
64

Order
0
1
2
2
3
Symbol
Operation
Comments
()
parentheses
inner ones evaluated rst
(unary) !
negation (unary NOT)
RL evaluation
<=,<
less than or equal, less than
LR evaluation
>,>=
greater than, greater than or equal
LR evaluation
==,!=
equivalent to, not equivalent to
LR evaluation
Lets return to the example of solving the quadratic equation, Ax2 + Bx + C = 0 that we
studied in owchart and pseudocode form in Lecture 3. If we look at the pseudocode on
page 7 of Lecture 3 we see that the ifs are nested and we would probably, at rst try, write
a code that looked something like the following, using only ifs and elses:
//File: quadratic.cpp
#include <iostream>
#include <cmath>
int main(void)
{ float a,b,c; // Declare the constants...
cout << "\n"
"We will solve for x in Ax*x + B*x + C = 0\n"
"=========================================\n"
"\n";
cout << "Input A,B,C: ";
cin >> a >> b >> c; //...and read them in
cout <<
"Calculating roots (x) of the equation\n"
<< a << "*x*x + " < b*b - 4*a*c) //b*b - 4*a*c is less than 0
{ cout << "No solution since b*b-4*a*c < 0\n";
return(0);
}
else // b*b - 4*a*c is >= 0
{ if (0 != a)
{
cout <<
"Soln 1 = " << (-b + sqrt(b*b - 4*a*c))/(2*a) << ", "
65
"Soln 2 = " << (-b - sqrt(b*b - 4*a*c))/(2*a) << "\n";

return(0);
}
else // b*b-4*a*c is >= 0 and a = 0
{ if (0 != b) // b*b - 4*a*c is >= 0, a = 0, b != 0
{ cout << "One soln since a = 0. Soln = " << -c/b << "\n";
return(0);
}
else // b*b - 4*a*c is >= 0, a = 0, b = 0
{ cout << "No solns since a and b = 0.\n";
return(0);
}
}
}
}
However, by using else ifs we can make the code look less nested and complicated. The
nesting is really still there and we have to remember that when we pass into an else if
that the if condition above it is still false! However, else ifs make the coding look neater
and easier to read.
//File: quadratic2.cpp
#include <iostream>
#include <cmath>
int main(void)
{ float a,b,c; // Declare the constants...
cout << "\n"
"We will solve for x in Ax*x + B*x + C = 0\n"
"=========================================\n"
"\n";
cout << "Input A,B,C: ";
cin >> a >> b >> c; //...and read them in
cout <<
"Calculating roots (x) of the equation\n"
<< a << "*x*x + " < b*b - 4*a*c) //b*b - 4*a*c is less than 0
{ cout << "No solution since b*b-4*a*c < 0\n";
return(0);
}
else if (0 != a) // b*b - 4*a*c is >= 0, a != 0
{ cout <<
"Soln 1 = " << (-b + sqrt(b*b - 4*a*c))/(2*a)
"Soln 2 = " << (-b - sqrt(b*b - 4*a*c))/(2*a)
return(0);
}
else if (0 != b) // b*b - 4*a*c is >= 0, a = 0, b !=
{ cout << "One soln since a = 0. Soln = " << -c/b
return(0);
}
else // b*b - 4*a*c is >= 0, a = 0, b = 0
{ cout << "No solns since a and b = 0.\n";
return(0);
}
<< ", "

<< "\n";
0
<< "\n";
4.7.1
Logical expressions with AND or OR
Here are more of the logical operators and their order of precedence:
Precedence Symbol
Operation
0 (rst)
(,)
parentheses, paired
1
(unary) !
negation (unary NOT)
2
<=,<
2
>,>=
3
==,!=
4
&&
logical AND
5 (last)
||
logical OR
Comments
LR evaluation
RL evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
The AND and OR operators work as follows. Imagine that two logical expressions evaluates
to either T (TRUE) or F (FALSE). This is the result of the && and ||s operating on them.
67
(T
(T
(F
(F
(T
(T
(F
(F
&&
&&
&&
&&
||
||
||
||
T)
F)
T)
F)
T)
F)
T)
F)
==
==
==
==
==
==
==
==
T
F
F
F
T
T
T
F
Sometimes these tables are dicult to remember, at least as abstract concepts. This is
helped somewhat if we substitute numerical values for T (1) and F (0).
(1
(1
(0
(0
(1
(1
(0
(0
&&
&&
&&
&&
||
||
||
||
1)
0)
1)
0)
1)
0)
1)
0)
==
==
==
==
==
==
==
==
1
0
0
0
1
1
1
0
Once we do this we realize that the result of (A && B) is the lower value of either A or B. The
result of (A || B) is the higher value of either A or B. It is no more complex than that! And
this is, in fact, how the electronic circuitry works in a computer during the interpretation of
logical expressions.
Some examples:
int i = 5;
if (0 i)
cout < i || 10 <= i)
cout << i << " is either negative, or has more than one digit\n";
4.7.2
Mixed arithmetic and logical expressions
It is possible to mix arithmetic and logical expressions. Usually it is done where a logical
expression is expected, such as within an if/if else/else construct. It is also possible to
mix these operators elsewhere, where a regular statement is expected. Dont do it unless
you really, really have to for some purpose!
68
When logical operators and arithmetic operators are mixed in a single expression, the order
of precedence is as follows. Above all, the contents of parentheses are evaluated starting with
the innermost. Then the arithmetic operators get interpreted in the order already discussed.
Finally, the logical operators get interpreted in the order already discussed. Here is the table
for precedence for both arithmetic and logical expressions.
Precedence
0 (rst)
1
2
3
4
4
4
5
6
7 (last)
Symbol
( )
(unary) - !
* / %
+ <=,<
>,>=
==,!=
&&
||
=
Operation
parentheses, paired
negation
multiply, divide, modulus/remainder
add, subtract
logical AND
logical OR
assignment
Comments
LR evaluation, innermost rst
RL evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
LR evaluation
RL evaluation
When in doubt, use parentheses. When not in doubt, use parentheses anyway. The
person reading your code may have forgotten all but the rst rule of precedence, parentheses
are interpreted rst. This will make your code more readable and understandable.
In fact, that is the convention I have adopted in almost all of my programming, put parentheses around every logical sub-expression in any logical expression that contains more than
one logical operators. I also put parentheses around any arithmetical sub-expression in a
mixed arithmetic/logical expression.
Heres an example of a mixed arithmetic and logical expression. Use the rules of precedence
stated above to see how the following statement is evaluated:
if (i*j == i*i + j*j || i - j <= i*i - j*j && i + j <= i*i + j*j) ...;
The ordering of && and || is also quite important. For example, using the rules of precedence
the above could collapse to 1 || 1 && 0 which is 1 or TRUE. However, if you mistakenly
thought that the || is evaluated rst, you would guess that the answer should be 0 or FALSE,
and that is the wrong answer in this case.
4.8. PROBLEMS
4.8
69
Problems
1. An essay-type question? In a computer1 class!

Answer the following questions in your own words...
(a) What is an algorithm?
(b) Not counting the ability to START and STOP, what are the 3 functional building blocks that are used to construct any algorithm.
(c) Give one example each of these 3 functional building blocks
(d) Name 3 ways in which algorithms may be described.
(e) Give one example each of the above 3 ways in which algorithms may be described.
2. Basic Arithmetic
Write down the value that is assigned to the variable on the left-hand side of each of
these expressions. The variables i and j are ints; x and y are floats.
i
x
j
y
y
x
=
=
=
=
=
=
1 + 2*3/4+5;
1.0 + 2.0 + 3.0/4.0 + 5.0;
(1 + 2)*3/4 + 5;
(1.0 + 2.0)*3.0/4.0 + 5;
1 + 2 + 3/4 + 5;
1 - (static_cast<int>(6*5/4.0*3))/2;
3. Basic Logical Operations

Do the following logical expressions evaluate to TRUE (1) or FALSE (0)?
i is an int that can take on any value.
(1 || 1 && 0 || 0)
(0 > i && 10 < i)
(0 > i || -10 < i)
4. Mixed Logical and Mathematical Operations
Do the following mixed mathematical-logical expressions evaluate to TRUE (1) or
FALSE (0)?
i is an int that can take on any value. (Hint: No matter what i is, 2*i + 1 is an odd
number and 2*i is an even number.)
1
Hey! Unless youve been on the moon for the last two weeks...This is a course about thinking and
making things work! Writing computer program is just one way (and a good way for Engineers!) to teach
this.
70

(2*i + 1)
(4*i%2 && 1)
(2*i + 1 || i/2)
5. Mind Your Ps, Qs and Semi-Colons
Which of the following are valid statements?
int i, j, k = 0;
int i, double j;
int i0 = 4, f = -1, k=0;
int i; float _0j;
int i j k l m;
int The, Wolverines, Are, Rated, Number, 3;
float f, F, _000000, Wolverine;
int i,j ; i == 1 && j || 3%2;
int static_cast<float>(N);
i = -i + -j;
i + j = f;
x = x*x + 3;
y = x^2 + 3;
y = a x + b;
z = x + y123;
z = x + 123y;
for(i = 0,i <= 10,i = i + 1) cout << "OK?\n";
while(thisIsTrue) cout << "OK?\n";
if (((x < 0) && (y > 3) || (z == 2)) cout << "OK?\n";
if (a != b && c > d && e < f && g == h) cout << "OK?\n";
6. Predict the Output
The examples of code given below all compile correctly. Predict the output that the
code would produce if you compiled and ran it.
int N =
if (N =
{ cout
}
else
{ cout
}
0;
0)
<< "The conditional expression was TRUE\n";
<< "The conditional expression was FALSE\n";
int i = 0;
if (0 != i)
4.8. PROBLEMS
71
cout << "Statement 1\n";

int j = 0;
if (0 < j);
{ j = j + 1;
}
cout << "j = " << j << \n";
7. This question is iy, REALLY iy
Dreaming up your own examples, write syntactically correct C++ code for the following. I have provided the answer to the rst question to give you an idea of what I am
looking for. You do not have to declare or initialize any of the variables you use.
(a) An example of the use of the if. If the condition is TRUE then one and only
one statement is executed.
A correct answer is: if (i < N) f = f*i;
(b) An example of the use of the if. If the condition is TRUE then exactly two
statements are executed.
(c) An example of the use of the if and the else if. If the if condition is TRUE
then exactly two statements are executed. If the else if condition is TRUE
then one and only one statement is executed.
(d) An example of the use of the if and two else ifs.
(e) An example of the use of the if, two else ifs and the else.
8. Location, location, location
Write a program that prompts its user for two ints, the rst representing an xcoordinate and the second representing a y-coordinate. Then the program prints out
where the (x, y) point is located, based on the following scheme.
x>0
x>0
x<0
x<0
x=0
x=0
x>0
x<0
x=0
y
y
y
y
y
y
y
y
y
> 0 upper right quadrant (URQ)

< 0 lower right quadrant (LRQ)
> 0 upper left quadrant (ULQ)
<0
lower left quadrant (LLQ)
>0
upper half (UH)
<0
lower half (LH)
=0
right half (RH)
=0
left half (LH)
=0
in the center (C)
72

Feel free to use abbreviations, like URQ, for upper right quadrant.
Heres an example of how it works:
% a.out
Enter two ints, representing x and y: 1 1
Upper right quadrant (URQ)
%
4.9. PROJECTS
4.9
73
Projects
1. Your rst programming problem

You have to create an entire code that will compile correctly, accept inputs and provide
outputs and stops. As for coding style, follow Appendix A.
Your program will do the following:
(a) Ask the user to input two non-negative integers.
(b) If the user inputs a negative integer, the program issues an error message and
stops.
(c) If the two numbers are equal, output is provided to inform the user. Otherwise
the user is informed which number is larger and which is smaller.
(d) The program outputs the ratio of the larger number to the smaller number unless
the smaller number is zero. If the smaller number is zero, a message to this eect
is output.
(e) The program outputs the quadrature sum of the two numbers, that is,
firstInteger * firstInteger + secondInteger * secondInteger.
(f) If the smaller number is zero, the program stops. Otherwise, the program checks
if the smaller number is a perfect divisor of the larger number. This means there
is no remainder after division. For example, 2 is a perfect divisor of 8, but not 7.
Output is provided to inform the user whether or not a perfect divisor was found.
(g) Program stops.
The above set of instructions is really a coarse pseudocode for your program. Read the
instructions very carefully.
Here are some example uses of the program. Your program should work exactly the
same.
> a.out
Input the 1st integer: -1
Input must be non-negative. Try again!
> a.out
Input the 1st integer: 1
Input the 2nd integer: -1
Input must be non-negative. Try again!
> a.out
Input the 2nd integer: 1
74

The two numbers are the same!
The ratio is 1, the quadrature sum is 2
1 is a perfect divisor of 1
> a.out
The bigger number is 10, the smaller number is 0
The smaller number is 0, so no ratio is calculated, the quadrature sum is 100
> a.out
The smaller number is 0, so no ratio is calculated, the quadrature sum is 100
> a.out
4 is not a perfect divisor of 7
> a.out
4 is not a perfect divisor of 7
> a.out
> a.out
4.9. PROJECTS
75
2. Leap year calculation

The rules to calculate if it is a leap year are:
(a) A year divisible by 4 is a leap year (2004, 2008...), unless
(b) it is also divisible by 100 (2100, 2200...) in which case it is not a leap year.
(c) There is an exception. A year divisible by 400 is a leap year (2000, 2400...).
Write a program that will accept a positive int, the year to be tested, from a user.
If the user inputs a 0 or negative number, the program quits without doing anything.
Otherwise, it will print a message saying whether or not the year is a leap year. Here
is an example of how the program works.
unix-prompt> a.out
Input a number for the year (>= 1): -1
Sorry, your year must be greater than 1.
unix-prompt> a.out
Input a number for the year (>= 1): 2001
The year 2001 is not a leap year
unix-prompt> a.out
The year 2004 is a leap year
unix-prompt> a.out
The year 2100 is not a leap year
unix-prompt> a.out
The year 2400 is a leap year
3. When in Rome...
The following table shows how Roman numerals are formed:
Arabic Roman Arabic Roman Arabic Roman Arabic Roman
1
I
10
X
100
C 1000
M
2
II
20
XX
200
CC 2000
MM
3
III
30
XXX
300
CCC 3000
MMM
4
IV
40
XL
400
CD
5
V
50
L
500
D
6
VI
60
LX
600
DC
7
VII
70
LXX
700
DCC
8
VIII
80
LXXX
800
DCCC
9
IX
90
XC
900
CM
76

A Roman number is written in decades starting with the largest decade on the left
and then going to the right. For example, the number 1999 would be represented as
MCMXCIX. The largest number that can be represented by Roman numerals is 3999,
or MMMCMXCIX. Write a program that will accept a positive int, between 1 and
3999 inclusive. If the user inputs something outside this range, the program quits
without doing anything. Otherwise, it will print the conversion to Roman numerals.
Here is an example of how the program works.
unix-prompt> a.out
Input an integer between
Sorry, your integer must
unix-prompt> a.out
Sorry, your integer must
unix-prompt> a.out
MCCCXXXVIII
unix-prompt> a.out
MMMDCCCLXXXVIII
unix-prompt> a.out
LII
1 and 3999 inclusive: 0

be between 1 and 3999 inclusive.
be between 1 and 3999 inclusive.
4. This is a 5-bit calculation

Write a program that will accept 5 bits (0 or 1) in the following order: b0 , b1 , b2 , b3 , b4
representing a 5-bit bit pattern b4 b3 b2 b1 b0 . (Note the ordering.) Write a program that:
(a) Determines if the 5-bit bit pattern b4 b3 b2 b1 b0 represents an odd or even unsigned
integer.
(b) Determines if the bit pattern represents a negative or positive signed integer in
5-bit twos complement representation.
(c) Prints the unsigned integer representation.
(d) Prints the bit patterns bit complement b4b3b2b1b0 (1s become 0s and 0s become
1s).
(e) Prints the signed integer representation assuming 5-bit twos complement representation.
Here is an example of how the program works.
4.9. PROJECTS
unix-prompt> a.out
Bit b0: 0
Bit b1: 1
Bit b2: 0
Bit b3: 1
Bit b4: 0
5-bit bit pattern is: 0 1 0 1 0
Bit pattern represent an even unsigned integer
Bit pattern represent a positive signed integer
Unsigned integer representation is: 10
Complement 5-bit bit pattern is: 1 0 1 0 1
Signed integer representation is: 10
unix-prompt> a.out
Bit b0: 1
Bit b1: 1
Bit b2: 1
Bit b3: 1
Bit b4: 0
Bit pattern represent an odd unsigned integer
Bit pattern represent a positive signed integer
Signed integer representation is: 15
unix-prompt> a.out
Bit b0: 1
Bit b1: 1
Bit b2: 1
Bit b3: 0
Bit b4: 1
Bit pattern represent a negative signed integer
Signed integer representation is: -9
unix-prompt> a.out
Bit b0: 1
Bit b1: 0
77
78

Bit b2: 0
Bit b3: 0
Bit b4: 1
Bit pattern represent a negative signed integer
Signed integer representation is: -15
5. Is binary arithmetic iy
We return to binary arithmetic of 5-bit bit patterns that was considered in one of
the problems in the previous chapter. We imagine a machine that can only represent
integers in 5-bit bit strings. However, this time you will write a program that emulates
5-bit binary arithmetic using actual ints in C. Remember that there is no 6th or
higher bit. The result of any calculation that would carry over into the 6th bit is
thrown away.
The program you will write will do two things, depending on the choice of the person
using your program:
(a) multiply a 5-bit bit pattern by a 2-bit bit pattern, for example, 10101 11, or
(b) add two 5-bit bit patterns, for example, 10101 + 10011.
The user of your program will provide the bit patterns that are to be added or multiplied. Your code will output the 5-bit bit pattern that is the result of the multiplication
or addition.
Examples of use:
Here is an example of running the code:
> a.out
What do you want to do?
Type a 0 if you want to add two 5-bit strings
Type a 1 if you want to multiply a 5-bit string by a 2-bit string
(0 or 1): 1
Enter the 5-bit binary number to be multiplied: 1 0 1 0 1
Enter the 2-bit multiplier: 1 1
4.9. PROJECTS
79
10101
x
11
-----...is the same as the addition...
10101
+01010
-----11111
>
Here is another example of running the code:
> a.out
What do you want to do?
Type a 0 if you want to add two 5-bit strings
Type a 1 if you want to multiply a 5-bit string by a 2-bit string
(0 or 1): 0
Enter the 1st 5-bit binary number to be added: 1 0 1 0 1
Enter the 2nd 5-bit binary number to be added: 1 0 0 1 1
10101
+10011
-----01000
>
6. Change for the good
You will write a program that will accept an input, a real number between 0.00 and
19.99 inclusive. This represents the change in dollars from the purchase of an item
using a $20 bill. You will write a C++ program that will calculate the most ecient
form that the change may be given in terms of $10, $5, and $1 bills plus quarters,
dimes, nickels and pennies.
Heres an example of how it should work. Note that the program was run twice. The
rst time the input was 19.94, the second time 16.26. Note how the program correctly
uses singular (One penny) and plural (Four $1 bills).
80

unix-prompt> a.out
Input a dollar amount between 0.00 and 19.99 inclusive: 19.94
Change of $19.94 is best given as:
One $10 bill
One $5 bill
Four $1 bills
Three quarters
One dime
One nickel
Four pennies
unix-prompt> a.out
Input a dollar amount between 0.00 and 19.99 inclusive: 16.26
Change of $16.26 is best given as:
One $10 bill
One $5 bill
One $1 bill
One quarter
One penny
unix-prompt>
7. Base 10 to binary conversion
Write a C++ program that will accept any integer between 0 and 127, inclusive, and
convert it to binary. You may not use loops of any kind and your solution must NOT
resemble the following:
if (decimal == 0) cout << "0";
else if (decimal == 1) cout << "1";
.
.
.
Heres an example of how it should work for an input of 126:

unix-prompt> a.out
Input a positive int between 0 and 127 inclusive: 126
4.9. PROJECTS
81
Decimal 126 converted to base 2 is 1111110

unix-prompt>
Hint: Depending on your algorithm, you may nd it useful to use the escape sequence
\b which causes the cursor to move backwards one space.
8. Something loopy to think about
This is a discussion question, not a programming assignment, because an elegant solution to this requires the syntax introduced in the next chapter. It is meant to make
you think about loops, the topic of the next chapter. How would you write a C++
program that could convert any unsigned integer to any base between 2 and 9? Using
loops, this actually easier to do than the solution to the previous problem!
unix-prompt> a.out
Input the base to convert to between 2 <= b <= 9: 7
Input any positive unsigned int: 126
unix-prompt>
82
Chapter 5
Loops
Loops are the third and nal programming element we need to write any algorithm. The
C++ language provides three easy ways to do this, the while, do/while and for loops.
5.1
The while loop
The general form of the while loop is:

while (logical expression)
{
STATEMENT BODY 1
}
Some features of the while loop:
Note the use of indenting.
Processing control passes to the statement immediately following the while construct
after the condition tested by the logical expression fails.
while loops (in fact, any repetition loop) can be nested within each other
The logical expression can mix arithmetic and logical operators to make for compact
coding. (This is not really recommended for novice programmers and often makes code
hard to read.)
If the logical expression is false (even the rst time), the STATEMENT BODY within the
while loop is never executed.
If the condition which is tested is TRUE and neither the STATEMENT BODY nor the
logical expression modies it so that it would become false, then it will run forever.
This is called a forever loop, while(1){}. Forever loops are bad, anti-social things.
83
84
CHAPTER 5. LOOPS
It puts your computer in a state known formally as hung, as in, Jeepers, I think
my computer is hung. It is NOT a compliment! Hung computers are put out of their
misery by a variety of desperate measures, increasing in severity. At the very worst,
the power has to be shut down, a really nasty, last resort for a hung computer.
This is the ow chart for the while loop:
F
T
STATEMENT BODY
Heres an example that tests Gausss summation formula:

//File: gauss.cpp
#include <iostream>
int main(void)
{ cout << "Input N: ";
int N;
cin >> N;
int i = 0, Sum = 0;
while (i < N)
{ i = i + 1;
5.1. THE WHILE LOOP
85
Sum = Sum + i;
}
cout <<
"Sum of " << N << " digits is " << Sum
<< "...according to Gauss: " << N*(N+1)/2 << "\n";
return(0);
}
Heres a more compact (but dangerous way) to do it:
//File: gaussAgain.cpp
#include <iostream>
int main(void)
{ cout << "Input N: ";
int N;
cin >> N;
int i = 0, Sum = 0;
while (i = i + 1 <= N) Sum = Sum + i; //NO, NO, NO!
cout <<
"Sum of " << N << " digits is " << Sum
<< "...according to Gauss: " << N*(N+1)/2 << "\n";
return(0);
}
This is dangerous because if you forgot the parentheses around the i = i + 1 part of the
conditional, you would generate an innite loop! Dont believe me? Try it! Um, wait, this
is a forever loop! OK, try it, but <CNTL>-C out of it after a few seconds.
The point istry to make your coding as easy to interpret as possible. Dont be too clever.
Dont save space just for the purpose of saving space at the expense of clarity. In the old days
of computing, memory and disk space were very expensive and precious. Programmers prided
themselves at making things very compact. Those days are, happily, long gone. However,
there still are people who do this. Maybe they think that an electron is as expensive as a
Higgs boson1 .
1
Electrons are essentially free. There are countless numbers of them in everyday things, like water. The
86
CHAPTER 5. LOOPS
5.2
The do/while loop
The general form of the do/while loop is:

do
{
STATEMENT BODY 1
}while (logical expression);
Some features of the do/while loop:
Do not forget the ; after the condition!

Processing control passes to the statement immediately following the do/while construct after condition tested by the logical expression fails.
do/while loops (in fact, any repetition loop) can be nested within each other.
The logical expression can mix arithmetic and logical operators to make for compact
coding. (This is not really recommended for novice programmers and often makes code
hard to read.)
The STATEMENT BODY is executed at least once even if the logical expression is false
initially.
If the condition which is tested is true and neither the STATEMENT BODY nor the logical
expression modies it so that it would become false, then it will run forever. This is
called a forever loop, do{}while(1);.
This is the ow chart for the do/while loop:
Higgs boson is very expensive. Physicists are spending billions to nd it. It has not been found yet, but
almost! The rst person to nd it will win the Nobel Prize for physics.
5.2. THE DO/WHILE LOOP
87
STATEMENT BODY
T
F
The only thing that really distinguishes a do/while loop from a while loop is that the
do/while loop evaluates the conditional at the end of the STATEMENT BODY and the while
loop evaluates it at the beginning. while loops are often used when incrementing a counter
and using (and usually updating) the counter within the statement body. do/while loops
are useful if you always want to process the STATEMENT BODY at least once. Because of this,
loops that accept input from the keyboard (or a le) make good use of the do/while loop
especially when the input is terminated using a sentinel.
Here is an example of sentinel controlled output for calculating the average of student grades.
The sentinel for stopping is when the input for a grade is less than zero. Even the cruelest
professor does not hand out a negative grades (even though it is legal)! So, this is a good
sentinel to use for this application.
//File: grades.cpp
#include <iostream>
88
CHAPTER 5. LOOPS
int main(void)
{ int nStudents = 0, grade, sumGrades = 0;
do
{ cout << "Input a grade: ";
cin >> grade;
if (0 > grade) cout << "Grade < 0, end of input.\n";
else
{
nStudents = nStudents + 1;
sumGrades = sumGrades + grade;
}
}while (0 <= grade);
if (0 < nStudents)
cout << nStudents
<< " grades were read in. "
<< "The class average is: "
<< sumGrades/nStudents
<< "\n";
return(0);
}
Note that the above calculates the average as sumGrades/nStudents, the result of an integer
division. This may not be completely fair as an average of, say, 79.9 should not really be
reported as 79.
The way to x this is to make the following changes:
1. Add the following #include in the pre-processor area of the code:
#include <iomanip>
2. Modify the cout statement as follows:
cout <<
<<
<<
<<
<<
<<
<<
nStudents
" grades were read in. "
"The class average is: "
setw(5) //Set field width to 5
setprecision(3) //3 digits of precision
static_cast<float>(sumGrades)/nStudents
"\n";
5.3. THE FOR LOOP
89
The code in its entirety can be found in the lectures area on the web. It is called grades1.cpp.
There are two new features in the cout statement. One is the introduction to the unary
cast operator static_cast<float>(expression) which converts sumGrades temporarily
(and internally) to a oat and the resulting calculation takes place as if it were a divide using
oats alone. It is not necessary to make the same conversion for nStudents. The division
of a oat by an integer or an integer by a oat always results in a oating point calculation
as the integer participating is promoted to a oat.
The general form of the cast operator is
static_cast<type_name>(expression) //Dont forget the parentheses!
to change expression to the type type_name. So, oats can be changed to ints and viceversa. It is possible to change among other types of variables that we will encounter later.
The other new feature is the use of a eld width and precision to modify the output of
the oat. The function setw(m) sets aside m spaces to print. The function setprecision(n)
prints n signicant digits. These functions are dened in the iomanip (I/O stream manipulator) library le. Hence, we had to include this le, else the compiler would have complained.
Note how these functions are employed. setprecision() only has to be set once. The
stream handlers remember it. However, setw() has to be set for every item!
There is another commonly used stream manipulator called fixed. We will not use it much.
5.3
The for loop
The for loop is similar to the while loop except that the syntax makes explicit the initialization, the looping condition and the loop counter modication. This form is generally
preferred if the loop is to be performed in a systematic fashion, to a well-dened termination
criterion with a regularly-spaced or easy-to-calculate loop counter indexing.
The general form of a for loop is:
for (expression 1; logical expression; expression 2)
{
STATEMENT BODY
}
Some features of the for loop:
90
CHAPTER 5. LOOPS
Processing control passes to the statement immediately following the for construct
after condition tested by the logical expression fails.
for loops (in fact, any repetition loop) can be nested within each other.
The logical expression is intended to provide the condition for which the execution
of the for loop terminates. This logical expression can mix arithmetic and logical
operators to make for compact coding. (This is not really recommended for novice
programmers and often makes code hard to read.)
The STATEMENT BODY will not be executed if the logical expression is false initially.
expression 1 is intended to initialize the loop counter (or index). Its use is optional.
However, in this case the initialization of any counters (should they be used) should
be done beforehand.
The use of the logical expression is also optional. If not provided, the STATEMENT
BODY is always entered. Another way of exiting the loop has to be provided otherwise
the loop is innite.
expression 2 is intended to increment the counter. The statement (or statements)

are executed within the loop after the STATEMENT BODY. Its use is also optional. If not
specied, the increments (or decrements) of any counters should be done within the
STATEMENT BODY.
The two ;s in the denition of the for loop are NOT optional. Leaving one or both
out will result in a compilation error.
This is the ow chart for the for loop:
5.3. THE FOR LOOP
91
EXPRESSION 1
F
T
STATEMENT BODY
EXPRESSION 2
Here is an example that executes a count down and a blast o.

//File: blastOff.cpp
#include <iostream>
int main(void)
{ cout << "Estimate number of counter loops/second: ";
int loopsPerSecond;
cin >> loopsPerSecond;
cout << "How many characters wide is your screen? ";
int screenWidth;
cin >> screenWidth;
cout << "How many characters high is your screen? ";
92
CHAPTER 5. LOOPS
int screenHeight;
cin >> screenHeight;
for (int i = 10;
{
// Next line
for (int j =
cout <= 0 ; i = i - 1)
should waste about 1 second
0; j <= loopsPerSecond; j = j + 1) ;
"\n";
cout << "\a.\n.\n.\n.\nWe Have ignition!\a\n.\n.\n.\n";

//"Draw" a rocket
cout << " /\\\n"
<< " ||\n"
<< " ||\n"
<< " ||\n"
<< " ||\n"
<< "/||\\\n";
//Start of the vapor trail
cout << " **\n";
for (int j = 0; j <= loopsPerSecond/2; j = j + 1) ;
cout << "*****\n";
for (int j = 0; j <= loopsPerSecond/5; j = j + 1) ;
//Vapor trail expands
for (int i = 6; i <= screenWidth; i = i + 1)
{ for (int j = 1; j <= i; j = j + 1) cout << "*";
for (int j = 0; j <= loopsPerSecond/i; j = j + 1) ;
cout << "\n";
}
//Vapor trail at maximum
for (int i = 1; i <= screenHeight; i = i + 1)
{ for (int j = 1; j <= screenWidth; j = j + 1) cout << "*";
for (int j = 0; j <= loopsPerSecond/screenWidth; j = j + 1) ;
cout << "\n";
}
//Vapor trail contracts
for (int i = screenWidth; i > 0; i = i - 1)
{ for (int j = 1; j <= i; j = j + 1) cout << "*";
5.3. THE FOR LOOP

for (int j = 0; j <= loopsPerSecond/screenWidth; j = j + 1) ;
cout << "\n";
}
//Clear the screen
for (int i = 1; i <= screenHeight; i = i + 1)
{ for (int j = 0; j <= loopsPerSecond/screenWidth; j = j + 1) ;
cout << "\n";
}
return(0);
}
93
94
CHAPTER 5. LOOPS
5.4
Problems
1. Flowcharts revisited
Draw the owchart representation (using lines, arrows, rectangles, diamond-shapes,
and words) for the following looping constructs:
(a) the for-loop:
for (i = 10; i > 0; i = i - 1){Sum = Sum + i;}
(b) the do/while loop:
do{cin >> oddNumber; odd = oddNumber%2}while(!odd);
(c) the while loop:
while (different){z = x; y = x; x = temp; different = x - y;}
2. Loop tests
(a) How many times is the following loop body executed?
int i = 0;
do
{ i = i + 1;
} while (i < 0);
(b) How many times is the following loop body executed?
int i = 0;
while(i < 0)
{ i = i + 1;
}
(c) How many times is the following loop executed?
int i = 3;
while(i)
{ i = i - 1;
}
(d) How many times is the following loop body executed?
int i = 0;
while(i <= 0)
{ i = i - 1;
}
(e) How many times is the following loop executed?
5.4. PROBLEMS
95
int j = 1;
do
{ j = -2*j + 1;
}while(j != 0);
(f) How many times is the following loop body executed?
for(int i = 0; i < 10; i = i + 2)
{ cout << "i = " << i << "\n";
}
3. Predict the output
The examples of code given below all compile correctly. Predict the output that the
code would produce if you compiled and ran it.
(a) int N =
if (N =
cout
else
cout
0;
0)
<< "The conditional expression was TRUE\n";
<< "The conditional expression was FALSE\n";
(b) int f;
for(f = 0; f <= 10; f = f + 2)
cout << f%10;
cout << "\n";
(c) int i = 0, Sum = 0, N = 3;
do
{ i = i + 1;
Sum = Sum + i;
}while(i <= N);
cout << "Sum = " << Sum << "\n";
(d) int i = 0;
if (0 != i)
(e) int j = 0;
if (0 < j);
{ j = j + 1;
}
cout << "j = " << j << "\n";
(f) int i = 2, j = -2;
while (j <= i)
96
CHAPTER 5. LOOPS
{
if (j < 0)
cout << "i is " << i << ".";
cout << " j is " << j << ".\n";
i = i - 1;
j = j + 1;
}
4. Loop drills
(a) Convert the following code containing a do/while loop to a code containing a
while loop in such a way that the execution of the code is identical.
int j = 0;
do
{ j = j + 1;
cout << j << "\n";
}while (j <= 10);
cout << j << "\n";
(b) Convert the following code containing a while loop to a code containing a do/while
loop in such a way that the execution of the code is identical.
int j = 10;
while (j > 0)
{ j = j - 2;
cout << j << "\n";
}
cout << j << "\n";
(c) Consider the following code:
#include <iostream>
int main(void)
{ int x = 1;
for (int i = 1; i <= 10 ; i = i + 1)
{ x = x*2;
cout << x << " ";
}
cout << endl;
return 0;
}
5.4. PROBLEMS
What does the above code print to the screen when you run it?
Draw the ow chart for the for loop in the above program.
(d) Consider the following code:
#include <iostream>
int main(void)
{ int i = 5;
do
{ i = i - 1;
cout <
int main(void)
{ int i = 13;
while(i%2)
{ i = i/3;
cout <
97
98
CHAPTER 5. LOOPS
int main(void)
{ for (int i = 1; i <= 2; i = i + 1)
for (int j = 1; j <= 3; j = j + 1)
cout <
#include <cmath>
int main(void)
{ float x;
cout << "Input x (must be positive): ";
cin >> x;
int i = 0;
const int NMax = 1000;
const float SMALL = 0.0001;
float xPower = 1.0, lastTerm, Sum = 0.0;
do
{ i = i + 1;
xPower = x*xPower;
lastTerm = xPower/i;
Sum = Sum + lastTerm;
}while (i < NMax && lastTerm > SMALL);
if (i == NMax)
cout << "Series did
else
{
cout << "i
=
<< "x
=
<< "lastTerm =
<< "Sum
=
}
not converge\n";
"
"
"
"
<<
<<
<<
<<
i << "\n"
x << "\n"
lastTerm << "\n"
Sum << "\n";
5.4. PROBLEMS
99
return 0;
}
(a) What series is being summed by this code? Sum = x + (you provide the rest)
(b) If the user inputs -1.0 for x, what will be printed out by this code?
(c) If the user inputs 0.0 for x, what will be printed out by this code?
(d) If the user inputs 1.0 for x, what will be printed out by this code?
(e) If the user inputs 0.0001 for x, what will be printed out by this code?
(f) If the user inputs 0.001 for x, what will be printed out by this code?
(g) If the user inputs 0.01 for x, what will be printed out by this code?
(h) If the user inputs 0.1 for x, what will be value of i that is printed out by this
code? (A hand calculator may help.)
6. Think Like a Computer Again
Consider the following C++ code.
#include <iostream>
#include <cmath>
int main(void)
{ float x;
cout << "Enter x: ";
cin >> x;
int n = 0;
float lastTerm;
const float EPSILON = 0.001;
float sign = 1;
float denominator = 1;
float total = 1;
while(n <= 1 || lastTerm > EPSILON)
{ n = n + 1;
if (n/2*2 == n)
{ denominator = denominator*n*(n - 1);
lastTerm = pow(x,n)/denominator;
sign = -sign;
total = total + sign*lastTerm;
cout << "Term " << n << " is " << lastTerm << "\n";
100
CHAPTER 5. LOOPS
}
}
cout << "The total is " << total << "\n";
return 0;
}
(a) If you input 0.01 for x, what will be printed out?
(b) If you input 1.0 for x, what will be printed out? (A hand calculator may help.)
(c) What is another condition for the if statement that gives the same result?
(d) What is the formula this program is implementing?
7. Predict and convert

(a) Prediction part: If you compiled and ran this program, exactly what would the
output look like?
#include <iostream>
int main(void)
{ const int maxTerms = 8;
const int x = 2;
int sign = 1;
int term = 1;
for (int i = 1; i <= maxTerms; i = i + 1)
{ sign = -sign;
term = term*x;
if (0 == i%2 || 2 <= i && 4 >= i)
{ cout < sign) cout << " - ";
else cout << " + ";
cout << term << endl;
}
}
return 0;
}
5.4. PROBLEMS
101
(b) Conversion part: Convert the program in the rst part of this question to use
a do-while loop rather than a for loop. Write your solution in the space below.
8. Predict and convert again
(a) Prediction part: If you compiled and ran this program, exactly what would the
output look like?
#include <iostream>
int main(void)
{ const int maxTerms = 8;
const int x = 2;
int term = 1;
int sum = 1;
cout << "Sum = 1";
for (int i = 1; i <= maxTerms; i = i + 1)
{ term = -term*x;
if (0 != i%3 && 0 != i%4 )
{ if (0 > term) cout << " - " << -term;
else cout << " + " << term;
sum = sum + term;
}
}
cout << " = " << sum << "\n";
return 0;
}
(b) Conversion part: Convert the program in the rst part of this question to use
a while loop rather than a for loop. Write your solution in the space below.
9. Fix the bugs
Consider the following C++ code. It is an attempt to do the sum:
S = 1 x2 + x4 x6 + x8 x10
The code has errors that the compiler detects as indicated in the comment lines. Fix
them. The code also has conceptual errors in some lines, also indicated by comments.
Conceptual errors are those that compile without error but do not provide correct
answers upon execution. Fix these as well.
102
CHAPTER 5. LOOPS
#include <iostream>
int main(void)
{ float x;
cout << "Input x: ";
cin >> x;
float Sum, p; // Conceptual errors on this line
int N = 10;
do
{ i = i + 2; // Compiler detects error on this line
p = p*x; // Conceptual errors on this line
Sum = Sum + p;
}while(i <= N); // Compiler detects error on this line
cout << "Sum = " << Sum << "\n";
return 0;
}
10. More bugs to x

Consider the following C++ code. It is an attempt to do the sum:
x4 x6 x8 x10
S = 1x +
2!
3!
4!
5!
2
The code has errors that the compiler detects as indicated in the comment lines. Fix
them. The code also has design aws in some lines, also indicated by comments.
Design aws are those that compile without error but do not provide correct answers
upon execution. Fix these as well. The corrected code should be able to compile and
calculate the above sum.

int main(void)
{ cout >> "Input x:"; //Compiler detects error(s) on this line
cin << x; //Compiler error(s) on this line
double xPower = x; //Design flaw(s)
double sum;
//Design flaw(s)
for(int term = 0; term <= 12; term = term + 1); //Design flaw(s)
xPower = xPower + x*term; //Design flaw(s)
5.4. PROBLEMS
103
sum = sum * xPower; //Design flaw(s)

cout << "Sum = " << Sum << "\n"; //Compiler error(s) on this line
return Sum; //Compiler detects error(s) on this line
}
11. Dealing cards
Complete the C++ code started below. The outer for loop deals 5 cards. The
statement:int rank = rand()%13 + 1; chooses a random number between 1 and 13
inclusive. So, a 1 is an ace, a 2 is a deuce, and so on up to 10, 11 is a jack, 12 is a
queen and 13 is a king. Then you must randomly pick the suit from one of 4: spades,
clubs, diamonds and hearts. Your code must print out the results as in the example.
Do not worry about having two cards the same as the deck your are dealing from is a
6-deck superdeck, six full decks of 52 cards each, so dealing identical cards is possible,
but not very likely.
Here is an example hand that was dealt by the program.:
Card
Card
Card
Card
Card
1
2
3
4
5
is
is
is
is
is
a queen of spades.
a 3 of spades.
an ace of diamonds.
a king of diamonds.
a 7 of clubs.
#include <iostream>
#include <cstdlib>
#include <ctime>
int main(void)
{ srand(time(NULL));
for(int nCards = 1; nCards <= 5; nCards = nCards + 1)
{ int rank = rand()%13 + 1;
// You fill in the rest
.
.
.
.
} // End of the for loop over the 5 cards
return 0;
} // End of the main routine
104
CHAPTER 5. LOOPS
12. Sentinel controlled input loops

Write a complete, standalone (if typed in, it would compile and run) C++ code that
prompts a user for an integer but only accepts it if it is a positive, odd integer less than
100. If the user provides an integer that does not satisfy this, an explicit explanation
is given and the user is prompted again. Your code has to provide output exactly as
follows. Computer output is in type font, user input is in italics.
User! Type in an integer: 0
The integer must be > 0. Try again.
The integer must be < 100. Try again.
The integer must be odd. Try again.
Successful input!
13. Count the factors of 2
Write a C++ code that inputs a positive integer and prints out the number of factors
of two in it. For example, an odd number has no factors of two, 2 has one factor of
2, 4 = 2 2 has 2 factors of 2, 6 = 3 2 has 1 factor of two, 8 = 2 2 2 has 3,
10 = 5 2 has 1 factor of 2, 12 = 3 2 2 has 2 factors of 2 and so on.
14. List the common factors
Write a C++ code that counts all the least common multipliers of a positive integer
including 1 and the number itself. For example 8 = 1 2 2 2, 17 = 1 17,
123456789 = 1 3 3 3607 3803. Here are sample uses of the program:
Input any positive int : 8
8 = 1 x 2 x 2 x 2
17 = 1 x 17
123456789 = 1 x 3 x 3 x 3607 x 3803
15. Compound interest

What happens if you deposit $1000.00 in a bank account that accumulates interest at
5% per year, compounded annually? After the rst year you have $1000.00 1.05 =
$1050.00 After the second year you will have $1050.00 1.05 = $1102.50. Write a
5.4. PROBLEMS
105
program that will accept a positive oat for the starting balance and output how much
money you will have at the end of each year. The program stops when your money
has at least doubled.
Heres an example of how it works:
% a.out
Input the starting balance: 1000
After
After
After
After
After
After
After
After
After
After
After
After
After
After
After
%
year
year
year
year
year
year
year
year
year
year
year
year
year
year
year
1, balance = 1050
2, balance = 1102.5
3, balance = 1157.62
4, balance = 1215.51
5, balance = 1276.28
6, balance = 1340.1
7, balance = 1407.1
8, balance = 1477.46
9, balance = 1551.33
10, balance = 1628.89
11, balance = 1710.34
12, balance = 1795.86
13, balance = 1885.65
14, balance = 1979.93
15, balance = 2078.93
16. Oyster-Shuckers Savings and Loan

Oyster-Shuckers Savings and Loan has a great deal for you! This banks pays 1%
interest per month, but only in months that do not have the letter r in its name (a
rather strange but lucrative deal). Thus, for each month with an r, the total amount
of money is multiplied by 1.01. For months without an r, the total is unchanged.
Write a program that:
Reads in an int representing the rst month the money is put in the bank,
Reads in an int representing the number of months the money is in the bank,
Reads in a float for the amount of money deposited,
Loops over the number of months that the money is in the bank, and,
If the current month has an r, multiplies the total by 1.01,
And nally, prints out the nal amount of money.
You may nd the following useful: The statement
106
CHAPTER 5. LOOPS
currentMonth = (startMonth + i) % 12;
calculates the month, given a starting month and a number of months elapsed. For
example, if you start in January (Month 1), 13 months later it is February (Month 2).
17. Theres Gold in Them Thar Ratios

In the Fibonacci series
1, 1, 2, 3, 5, 8, . . .
each number (after the rst two) is equal to the sum of the two before it. So, for
example, the next number after 8 is 8 + 5 = 13. One of the properties of the series is
that the ratio of a number in the series to the one immediately preceding it converges
(when the Fibonacci get very large) to the so-called Golden Ratio.
Write a program to approximate the Golden Ratio. Your program should compute
terms in the Fibonacci series, taking the ratio of the most recent term to the one
before it. Once the ratio is not appreciably dierent from the last time it was computed,
terminate the loop and print out the result.
18. The Mersenne primes
A prime number is a positive whole number that can only be divided by itself and 1
to produce another positive whole number. The rst few primes are 1,2,3,5,7,11... In
C++ language the test:
if (0 == N%i)
would never be true for a candidate prime N if i were greater than 1 and less than N.
A Mersenne prime is a prime number that also has the form 2p 1 where p is a positive
number greater than 1.
Write a C++ code that calculates candidate Mersenne primes 2p 1 for 2 p 31.
(This step is coded for you below.) Then test the number to see if it is prime. (Youll
need a loop to do this.) If you detect that the number is prime, print out the number
and the value of p.
19. Nest your fors, inside and out
Write a C++ program that will:
(a) Prompt the user for a positive odd integer within a do-while construct. If the
integer is 0, negative or even, prompt the user again. The algorithm goes to the
next step when an odd positive integer is received. Call this odd integer N.
(b) Output an inverted triangle of asterisks that looks like the following example.
The rst row has N *s. Each following row has two *s less than the one above it
and is centered with respect to the row above.
Heres an example of how the program should work:
5.4. PROBLEMS
red% a.out
Input an odd
Input an odd
Input an odd
Input an odd
***********
*********
*******
*****
***
*
red%
107
positive
positive
positive
positive
int:
int:
int:
int:
-1
0
2
11
You must use a do{}while(); loop to obtain N.

You must use for-loops to make the drawing.
You may not use arrays.
You must only use cout statements that have only single character strings, for example, cout << "*";. Something like cout << "**"; or cout << " *"; is not
permitted. Spaces are counted as a character.
Your program must work for arbitrary odd, positive N.
20. Nest your fors, left to right
Write a C++ program that will output the following:
****************************************
***********
*
* **********
*
* **********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
108
CHAPTER 5. LOOPS
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
**********
*
*
********** *
*
********** *
*
***********
*
**********
*
*********
*
********
*
*******
*
******
*
*****
*
****
*
***
*
**
****************************************
You ...
must use for-loops, some nested, some not.
may not use arrays. (Ignore this statement of you do not know what an array is.)
must only use cout statements that have only single character strings, for example,
cout << "*";. Something like cout << "**"; or cout << " *"; is not permitted. Spaces are counted as a character. For example, the top line would made
correctly as follows:
const int width = 4;
for (int i = 1; i <= width; i = i + 1) cout << "*";
cout << endl;
rather than the following which is not allowed,
cout << "**************************************************";
The width and length of the drawing are exactly 40 characters.
5.4. PROBLEMS
109
21. For success, nest your fors

In the output that follows, a C++ program prompted a user for two inputs, the height
and width of a box and the user responded with 30 for both. Write the C++ program
that did this:
Height of box: 30
Width of box: 30
* * * * * * * * * * * * * * *
** ** ** ** ** ** ** ** ** **
*** *** *** *** *** *** *** **
**** **** **** **** **** ****
***** ***** ***** ***** *****
****** ****** ****** ****** **
******* ******* ******* ******
******** ******** ******** ***
********* ********* *********
********** ********** ********
*********** *********** ******
************ ************ ****
************* ************* **
************** **************
*************** **************
**************** *************
***************** ************
****************** ***********
******************* **********
******************** *********
********************* ********
********************** *******
*********************** ******
************************ *****
************************* ****
************************** ***
*************************** **
**************************** *
*****************************
You must use nested for-loops to solve this problem.
You may not use arrays. (Ignore this statement if you do not know what an array
is.)
110
CHAPTER 5. LOOPS
You must only use cout statements that have only single character, for example,
cout << "*";. Something like cout << "**"; or cout << " *"; is not permitted. Spaces are counted as a character.
22. Base Pythagorean Triples

A base Pythagorean triple is dened as three integers, i, j, k such that i2 + j 2 = k 2 and
i, j, k have no common divisors other than 1. For example 3, 4, 5 is a base Pythagorean
triple. However, 6, 8, 10, although 62 + 82 = 102 , is not a base Pythagorean triple since
6, 8, 10 are all divisible by 2. Write a program that nds all the base Pythagorean
triples such that i, j, k 1000.
Here is an example of running the code:
Maximum integer to
Triple # 1: 3*3 +
Triple # 2: 5*5 +
Triple # 3: 8*8 +
Triple # 4: 7*7 +
Triple # 5: 20*20
Triple # 6: 12*12
Triple # 7: 9*9 +
Triple # 8: 28*28
Triple # 9: 11*11
Triple # 10: 16*16
Triple # 11: 33*33
Triple # 12: 48*48
Triple # 13: 13*13
Triple # 14: 36*36
Triple # 15: 39*39
Triple # 16: 65*65
Triple # 17: 20*20
.
.
.
test = 1000
4*4 = 5*5 = 25
12*12 = 13*13 = 169
15*15 = 17*17 = 289
24*24 = 25*25 = 625
+ 21*21 = 29*29 = 841
+ 35*35 = 37*37 = 1369
40*40 = 41*41 = 1681
+ 45*45 = 53*53 = 2809
+ 60*60 = 61*61 = 3721
+ 63*63 = 65*65 = 4225
+ 56*56 = 65*65 = 4225
+ 55*55 = 73*73 = 5329
+ 84*84 = 85*85 = 7225
+ 77*77 = 85*85 = 7225
+ 80*80 = 89*89 = 7921
+ 72*72 = 97*97 = 9409
+ 99*99 = 101*101 = 10201
5.5. PROJECTS
5.5
111
Projects
1. Something loopy to do
Write a C++ program that converts any unsigned integer to any base between 2 and
9.
unix-prompt> a.out
Input the base to convert to between 2 <= b <= 9: 7
Input any positive unsigned int: 126
unix-prompt>
2. Going nowhere fast

In this problem the program starts up assuming that it is following an object on a
2-dimensional plane as it takes many steps to perform a sort-of random walk. The
object starts o at the origin (x = 0 and y = 0). You will input two real numbers
that will represent the next coordinates of the object. The computer will calculate
the straight-line distance that it has to travel to get there and position the object at
that point. Then you will input two more real numbers representing a move from that
location, then another position (another pair of points) and another position (pair of
points) and so on. To stop the program you have to input the identical new coordinates
as the previous stepthat is, there is no distance to be traveled for that nal step.
This is a sentinel indicating that the loop should terminate. This last zero-step
should not be counted in the number of steps. When the program exits the loop at
this point make the program print out the total distance traveled, the total number of
moves and the average distance per move.
Here is some pseudocode for this problem:
(a) START: Declare N = 0 (Counter for the number of steps), xi , yi (New coordinates), x0 = 0, y0 = 0 (Current coordinates), s (Distance of a single step) and
t = 0 (Total distance).
(b) Start of the input/stepping loop.
i. Read in the the new coordinates: xi and yi.
ii. Test to see if xi and yi are valid input. If they are not, go back to (b-i.),
otherwise continue on to the next step (b-iii.).
112
CHAPTER 5. LOOPS
iii. Test to see if xi and yi are identical to x0 and y0 . If they are not you will
continue with step (b-iv.) Otherwise you will want to transfer control to the
the rst statement following the input/stepping loop, step (c).
iv. Increment the step counter N.
v.
vi.
vii.
viii.
Calculate the distance of the step s = (xi x0 )2 + (yi y0 )2 .

Accumulate s in t, the total distance.
Important! Update the current position x0 = xi , y0 = yi .
Try to take another step, that is return control to the top of the input/stepping
loop, step (b-i.).
(c) Report the total distance traveled.

(d) Report the total number of steps.
(e) Report the average distance traveled per step. Important: make sure that the
case of zero steps and zero distance traveled does not cause problems.
(f) END:
5.5. PROJECTS
113
3. The exponential function

(a) Computation of the factorial
Consider the following C++ program:
#include <iostream>
int main(void)
{ cout << "Input N: "; int N; cin >> N;
if (0 > N)
cout << "N < 0. Stopping.\n";
else if (0 == N)
cout << "0! = 1\n";
else
{ int factorial = 1;
for (int i = 1 ; i <= N; i = i +1)
factorial = factorial*i;
cout << N << "! = " << factorial << "\n";
}
return 0;
}
This program computes the factorial of the integer N, input by the user. The
factorial of N, represented by the mathematical symbol N!, is calculated by:
N! = N (N 1) (N 2) (N 3) 3 2 1
By denition, 0! = 1. Other factorials are 1! = 1, 2! = 2, 3! = 6, 4! = 24, 5! = 120
and so on. Factorials are used a lot in many branches of mathematics.
Your problem is to nd out for what values of N input by the user, does the
program above not give correct answers. (Heres a hint: The maximum signed
integer has a value 231 1 or 2147483647.) The way you determine this I leave
up to you. You can do it mathematically, or you can run the above program
for dierent values of N until you detect that something is wrong. Turn in a
description of how you found the answer.
(b) An improved computation of the factorial
Convert the above program so that the variable called factorial is a float.
(Turn in your code.) Now, try again to nd out for what values of N input by
the user, does the program above not give correct answers. (Heres another hint:
The maximum float has a value 3.40282347 1038 .) You may determine this
114
CHAPTER 5. LOOPS
mathematically, or you can run your program for dierent values of N until you
detect that something is wrong. How do you know that something has gone
wrong? Turn in a description of how you found the answer.
(c) Calculating the exponential of a number
The exponential function, called ex is one of the most common functions in all of
mathematics. One way of calculating it is to use the following:
1
x x2 x3 x4 x5
xn
e = + +
+
+
+
+
0! 1! 2!
3!
4!
5!
n!
x
an innite series, one that never stops. Mathematically, the above expression for
ex is exact for any x from minus innity to plus innity. However, to get an
accurate calculation from a computer requires some work. Here is what you have
to do:
i. Write a C++ program to compute ex . Your program should try to compute
it by summing the nite series
ex =
1
x
x2 x3 x4 x5
xN
+ +
+
+
+
+
0! 1! 2!
3!
4!
5!
N!
where N is some integer that you, the programmer, will set. Be careful
that you do not set N too high, otherwise, as indicated in the previous two
problems, your answer may not make sense. N should also be great enough
so that your answer is as accurate as it can be. Turn in your code.
ii. Discuss over what range of x you expect your program to provide a reasonably
accurate answer. Turn in your discussion in such a way that pleases your Lab
Instructor.
You can check your results with a calculator. Else, if you read ahead, you can
check your program using the C++ math function exp(x).
5.5. PROJECTS
115
4. Primal loop therapy

This question is all about loops. A solution for this may involve all of the loops we
have encountered: the do{}while() loop, the for(){} loop, and the while(){} loop.
A solution also facilitated with the use of the modulus operator (%).
A prime number is a positive whole number that can be divided perfectly only by itself
and one. The rst 26 prime numbers are 1, 2, 3, 5, 7, 11, 13, 17, 19, 23, 29, 31, 37,
41, 43, 47, 53, 59, 61, 67, 71, 73, 79, 83, 89, and 97. You will write a C++ code that
will accept two integers from its user. You will restrict the range of inputs to between
4 and 1,000,000 (one million). The user does not know this, he or she only knows to
input integers and so you will keep prompting the user until he or she gives you proper
input. (Sentinel-controlled do{}while() loops are best for this job.) The two integers
can be accepted in any order, bigger one rst or smaller one rst. You can assume that
the user is not hostile. The user will only input integers, not oats nor characters
nor anything else that cin might nd disturbing when asking for an int.
Once you have two integers you can work with, you will nd all the prime numbers
between those two numbers, including the starting and ending numbers themselves.
You will output the prime numbers as you nd them to the screen, and as a last step,
print out how many of them you found.
Hint 1: Make sure that the primes your program gives agree with the ones in the list
above.
Hint 2: There are 78,496 prime numbers between 4 and one million.
Hint 3: If you are using Hint 2 as a way of checking your program (please do!) and
you nd that it takes a very, very long time to get your program to complete its task,
stop a moment and think! Think about how many potential divisors you have to test
to see if a number is prime or not. It should not take more than about a minute or
so to nd all the primes between 4 and one million, depending on what machine you
are using. On a 700 MHz PIII machine a solution takes about 70 seconds if it outputs
every number it nds to the screen and only about 10 seconds if it outputs only the
number of primes found. To be sure, there are even faster ways of doing it!
Finding prime numbers is an exercise that has intrigued mathematicians for centuries.
As soon as computers were invented, this was one of the early problems that mathematicians programmed them for. We know that integers have only 32-bit representations.
This is just food for thought, not part of the question: How you would nd out if a
number bigger than 232 is prime? Now thats a tough computer problem. Here is an
example of output from correct solution:
> g++ primes.cpp
> a.out
Input the first integer: 1000001
116
CHAPTER 5. LOOPS
Input must be > 4 or < 1000000
Keep trying!
Input the first integer: 1000000
Input the second integer: 1
Input must be > 4 or < 1000000
Keep trying!
Input the second integer: 4
Looking for all the prime numbers between 4 and 1000000 inclusive
5 is prime
7 is prime
11 is prime
13 is prime
17 is prime
19 is prime
23 is prime
29 is prime
31 is prime
37 is prime
41 is prime
43 is prime
47 is prime
53 is prime
59 is prime
61 is prime
67 is prime
71 is prime
73 is prime
79 is prime
83 is prime
89 is prime
97 is prime
.
.
.
999953 is prime
999959 is prime
999961 is prime
999979 is prime
999983 is prime
There were 78496 prime numbers between 4 and 1000000 inclusive
5.5. PROJECTS
117
5. Wheres Waldo?
It is a little known fact that Waldo Schtinklebaum, the character of the infamous
Wheres Waldo? game was a student here at the University Michigan. He started
o his career in Civil Engineering. He was notorious for 1) being a lousy dresser, 2)
having no sense of direction (why he was always getting lost), 3) and being indecisive.
In fact, after partying all night at the Frats, Waldo would often lose his way back
to his dorm. Hed leave the party, forget which way his dorm was, take a step in a
random direction, change his mind and his direction and take another step. This is not
a bad approach to getting where you want to go, except that it can take a very long
time to reach your destination. Waldo sometimes ended up walking around randomly
until the morning when someone would take pity on him and lead him back home. In
fact, Waldos meanderings led to his second career. Shortly after graduation, Waldo
participated in the infamous Spring running event that takes place in the late evening
on the last day of Winter term. On this day, Waldo changed his manner of dress
permanently. Despite his motivation to run in the correct direction, Waldo ended up
in the Kresge Eye Center were his vision was corrected and he donned the hospital
greens of a medical doctor. Waldo is now an optometrist working in Livonia under an
assumed name. His specialty is treating children who have gone cross-eyed staring at
the puzzles that bears his former name.
Your project is to write a program that imitates Waldos after-party mission to get
home. A pseudocode description of the solution is given below.
Pseudocode for Waldos mission
Waldo starts his journey at location (x, y) = (0, 0), the front door of the Frat
house, the center of the universe that Waldo is in. Waldos universe is a 2dimensional plane that measures 2020, that is 10 x 10 and 10 y 10.
Waldos home is a 2 2 square in the upper right hand corner, 8 x 10 and
8 y 10.
Declare a variable called stillWalking and give it an initial value of 1 (true),
which means that Waldo is still trying to nd his way home on foot.
Declare another variable called atHome and give it an initial value of 0 (false)
which means that Waldo has not yet found his way home.
Declare another variable called stepsToTry and give it an initial xed value of
1000. Waldo hopes to get home before 1000 steps. If he doesnt, you will either
tell him to keep going and try another 1000 steps, or to give up and sleep on the
nearest park bench.
WHILE (stillWalking)
FOR-LOOP until 1000 steps are taken and while Waldo has not yet reached
home.
DO (Loop for choosing the direction of the step)
118
CHAPTER 5. LOOPS
Take a proposed step of unit length in a random direction. (See note
on the next page.)
WHILE (endpoint of the step is outside of Waldos universe)
Take the step.
If Waldos new location is within the square dened as his home, set
atHome to true.
END FOR-LOOP
IF (atHome), print out the total number of steps that Waldo has taken and
stop the program.
ELSE, read in a new value for stillWalking. If the user inputs a non-zero
value for stillWalking, Waldo will try to take another 1000 steps, otherwise,
if the user inputs a zero value for stillWalking the logical condition of the
outer WHILE loop will be false.
END WHILE
End the program.
Note 1: Taking a step in a random direction
To use random numbers, you must
#include <cstdlib>
in the preprocessor area for the code. Then, an expression of the form:
rand()
will produce a random int between 0 and RAND MAX inclusive, where RAND MAX is a big
positive number dened in cstdlib. This random int must be converted to a random
double that lies between 0 and 2. This can be done via the following statement:
double angle = 2*PI*rand()/RAND_MAX;
where PI can be declared and initialized (earlier in the program) by
const double PI = 4*atan(1.0);
Using the above atan() function requires the use of
#include <cmath>
in the preprocessor area. Finally, x and y displacements can be generated by the
statements:
5.5. PROJECTS
119
dx = cos(angle);
dy = sin(angle);
where dx and dy are doubles. This is one of Waldos steps. You should use exactly
this order of displacement statements in your solution.
Example use of the program
red% a.out
Will Waldo make it home? Press <RETURN> to continue.
Waldo is still lost after taking 1000 steps
Continue playing? (Input "0" to stop): 1
Waldo is still lost after taking 2000 steps
Continue playing? (Input "0" to stop): 1
Waldo made it home after taking 2552 steps!
Note 2: Making it really sort of random (optional topic)
If you have followed the above instructions carefully, you will notice that every execution of the program is exactly the same. Moreover, the solution of your colleagues,
at least those who followed the instructions carefully, will also get Waldo home in the
same number of steps (assuming you all are using the same type of computer). In this
case, it is a desirable thing because, it allows your GSIs to grade your assignment
quickly. This makes your GSIs happy because most GSIs would volunteer to undergo
a frontal lobotomy rather than grade. (I used to be a GSI and I know that I would.)
Happy GSIs give out better grades, its a fact of life. Im digressing again...
If you want to make each play of the game dierent, do the following. In the preprocessor area of your code,
#include <ctime>
Then, in your code before your rst use of rand(), put in the following expression:
srand(time(NULL));
Thats how I generated the dierent solutions that are shown in graphical form on the
next page. The expression time(NULL) gives the number of seconds since the start of
the epoch, January 1, 1970, 12:00am GMT. This is considered to be the start of the
modern computer age.
Note that you can also seed the random number generator rand() using the following
statement:
120
CHAPTER 5. LOOPS
srand(someInt);
where someInt is an int that you can choose. If you choose the same someInt every
time, the game will be the same. Dierent someInts result in dierent games. Try it!
Some things to think about (optional)
These are some things to think about. Dont hand answers to these to your GSIs.
However, feel free to speak to them or me (the Prof) about it.
Why did I conne Waldo to a 20 20 universe?
Is Waldo always guaranteed to make it home if he tried long enough?
If I took away the walls of the universe, on average how far away would Waldo be
from his starting point after a certain number of steps.
If you took away the walls, would Waldo always guaranteed to make it home if
he tried long enough?
Graphical output of Waldos meanderings on next page
Waldo is there. Can you see him?
5.5. PROJECTS
121
Wandering Waldo
First example
Waldo
gets
inspired!
Home
Waldo starts here

Waldos tortuous path
Wandering Waldo
Second example
122
CHAPTER 5. LOOPS
Chapter 6
Early AbstractionFunctions
6.1
Motivation for functions
Let us consider an implementation of the al-Kashi algorithm. This algorithm computes the
X N where N is any integer, positive, negative or non-zero. Here is what the C++-code
looks like (This code is called alKashi.cpp on the lecture distribution area on the course
website):
//File: alKashi.cpp
#include <iostream>
int main(void)
{ cout << "A calculation of X (float) to the power of N (int) \n";
float X; // The float to be raised to a power
int
N; // Power to be raised to
cout << "Input X (float) and N (int): ";
cin >> X >> N;
float result; // The result of the calculation
{
//Open a new "scope" or "code" block

//An implementation of the al-Kashi algorithm
int
n = N; //An int used internally in this code block
float x = X; //A float used internally in this code block
123
124
CHAPTER 6. EARLY ABSTRACTIONFUNCTIONS
if (0 > N)
{ //This "trick" allows us to calculate for negative powers!
n = -N;
x = 1/X;
}
result = 1; // Initial value
while(1 <= n)
{ if (0 == n % 2)
{ n = n/2;
x = x*x;
}
else
{ n = n - 1;
result = result*x;
}
}
} // End of the scope block
cout << X << " to the power " << N << " = " << result << "\n";
return(0);
}
The above example introduces a new concept in C++-programming, that of a block, or
code block. A block is any set of statements enclosed by a set of curly brackets {...}s.
In the above example,
{ //Open a new "scope" or "code" block
.
.
.
delimit a code blockthis one associated with the Al-Kashi algorithm. Variables can be
dened within a code block. Those that are dened within them (in this example, float x
and int n) are not known (or dened) outside of the the block. If you tried to do something
with x or n outside of their block, you would get a compilation error, complaining that these
variables were not dened. At least, that will be our understanding of code blocks until we
get to the section called scope rules a little later in the course. We have actually seen
code blocks before. The main routine has one, as well as the if/else if/else construct,
the while and do/while constructs as well as the for loop.
6.1. MOTIVATION FOR FUNCTIONS
125
Now, suppose that we did not know about the al-Kashi algorithm and coded the algorithm in
a more straightforward way (This code is called dumbPower.cpp on the lecture distribution
area on the course website.):
//File: dumbPower.cpp
#include <iostream>
int main(void)
int
cin >> X >> N;
{
//Open a new "scope" or "code" block

//An implementation of the dumbPower algorithm
int
n = N;
float x = X;
if (0 > N)
{ n = -N;
x = 1/X;
}
result = 1; // Initial value
for (int i = 1 ; i <= n ; i = i + 1)

{ result = result*x;
}
return(0);
}
In these two examples, we note that there is a lot of repetition in the code outside of the
scope block. Repetitious coding is BAD! It means that someone has wasted a lot of time!
126
Indeed, as far as the main routine is concerned, it does not really care about the contents of
the code block! All it really wants is the answer. It can think of the algorithm to compute
the power in a very abstract way. All main really needs to do is to give the scope block the
variables X and N.
What we really, REALLY want is a routine that is independent of the specic calculational
technique that is used to obtain X N . In fact there is. There is a function in the C++
standard math library that does this. It is called pow. If we use this information we can
write a much more compact code (This code is called power0.cpp on the lecture distribution
area on the course website.):
//File: power0.cpp
#include <iostream>
#include <cmath>
int main(void)
int
cin >> X >> N;
result = pow(X,N);
return(0);
}
Note the use of
#include <cmath>
that helps with the interpretation of the statement:
result = pow(X,N);
6.1. MOTIVATION FOR FUNCTIONS
127
This is a much more abstract and powerful implementation of the previous two examples.
It allows the main routine to be concerned with only the input and output of the various
parameters, X, N, and the answer. The main routine in this case is much more general.
Let us pretend that we are now in the infancy of the C++-language, and that the pow function
does not exist. However, we require this functionality repeatedly in our applications. But,
we do know about al-Kashis algorithm. Wanting to be ecient, we would code the al-Kashi
algorithm into a C++-function and employ it as follows (This code is called power1.cpp on
the classcodes distribution area):
//File: power1.cpp
#include <iostream>
float pow(float x, int n)
{ float result; // The result of the calculation
if (0 > n)
{ n = -n;
x = 1/x;
}
result = 1;
while(1 <= n)
{ if (0 == n % 2)
{ n = n/2;
x = x*x;
}
else
{ n = n - 1;
result = result*x;
}
}
return result;
}
int main(void)
128

int
cin >> X >> N;
result = pow(X,N);
return(0);
Now we note a few important things. There are a lot of things to remember about functions
and we are only going to scratch the surface.
In our use of functions, the function denitions and the main routine go in one le.
(They can be put into dierent les but it becomes more complicated.)
The function denitions go at the top of the le, after the pre-processor directives but
before the denition of the main routine. (They can be put after the main routine but
it becomes more complicated.)
The functions denitions should go before they are used in any other function, including
main. (Again, we can use another ordering, but it becomes more complicated.)
Functions, the way we have used them so far, can return only one thing via the statement
return expression.
The expression in the return line must evaluate to a data type that is specied in
the denition of the function, in this example, float pow()...
The parameters in the function denition, in this example, (float x, int n), must
be declared.
The variables in the function call, in this example, X,N from mains function call
result = pow(X,N);
are not changed by the function. (There is another way to call functions where the
parameters can change. This will be taught later.)
6.2. USER-DEFINED FUNCTIONS
129
Once the function returns to its calling routine, main in this case, the variables in
the functions parameter list and any other variables that it declares, are lost. The
only thing that main has is the return value. (To be absolutely truthful, there are
ways that the calling routine can look at the memory where the function stored things
temporarily, but...its complicated! Moreover, it is almost never done.)
Functions can call other functions.
main is just another function.
If a function does not return anything, its type is void. Example,
void returnNothingFunction(void){}
If a function has no parameters, put void where the parameter list goes, as in the
above example.
Finally, functions are really, really, REALLY important. So it is important to understand them and use them well. We will spend a lot of eort in this course with
functions.
6.2
User-dened functions
In this section, we examine the syntax associated with functions.

There are three things that are associated with the denition of a function in ANSI C++:
The function prototype: For small applications, where all the source code is found in
one le, the function prototype is usually written in the pre-processor area, the area
before any function denition (see below). In small applications, where all the source
code is found in one le, the function prototype can be omitted if the function denition is located in the le before any function call (see below) is made to it. For large
applications, or applications that have more than one source le, function prototypes
are usually gathered into an include le, named, for example, myIncludes.h and accessed via an include statement #include <myIncludes.h> placed in the preprocessor
area.
The function denition: For small applications, where all the source code is found in
one le, the function denition is written after the function prototype and outside
of the main routine. For large applications, or applications that have more than one
source le, it is conventional to have one function denition per source le.
130
The function call: The call to the function can be inside of main or another function.
In fact, a function can even call itself. This is called recursion and is discussed in a
later chapter.
Understanding the syntax of these three things is crucial to the successful use of functions.
The general use of a user-dened function is as follows:
/* Opening comments */
.
.
.
// Start of pre-processor directives
.
.
.
// function prototype (if required)
(type) myFunction(variable type list separated by ,s);
.
.
.
// end of pre-processor directives
.
.
.
/* function definition */
(type) myFunction(variable type list and variables separated by ,s)
{
.
.
.
/* Function body with declarations and
executable statements
*/
.
.
.
}
.
.
.
int main (void)
{ // Start of main routine
131
.
.
.
// function usage
...myFunction(list of values separated by ,s)...;
.
.
.
} // End of main routine
General comments on functions:

The function prototype is the declaration of a function. It species what the function
is expected to provide (return value, see below), what is in the parameter list (inside
the parentheses) and its identier (name). You do not need a function prototype if
all your source code is located in one le and if the rst time the compiler encounters
your function within a source code le is the function denition itself. The compiler
will form its own prototype from the denition! This is why we can get away with
not using prototypes in most cases when everything is in the same le, the function
denitions are in the proper order and main comes last. There are exceptions to this
rule but we will probably not encounter them in this course. If the compiler encounters
the function for the rst time in the function call, it will form its own prototype and
assume that the return type is an int and can potentially mess up on the argument
list as well. Avoid this situation at all costs!
The (type) in the function prototype and function denition inform the compiler
that the function called myFunction will return a variable of a certain type. (type)
could be float, e.g.
float myFunction();
which means that myFunction is expected to return a oat.
Other possibilities for (type) could be int or any other variable type in C++ (that
have not been discussed yet in this course).
If (type) is not dened, then the compiler assumes that it is an int.
The other possibility for (type) is void. void means that the function will not return
a value. (It will perform a function on some data but not return any kind of value.)
(type) in the function prototype and function denition must agree, else there
will be a syntax error.
132
Forgetting to return a value when (type) indicates that one should be returned results
in unpredictable behavior.
Returning a value when (type) is void causes a syntax error.
Note that
int main (void)
{
.
.
.
}
is a function denition! Its (type) is int. We could have said
main ()
{ .
.
.
}
because functions are assumed to be (type) int unless we state otherwise. main is a
special function in C++ that the operating system knows is the place where to start
processing. Only main has this special status. I always (type) every function that I
dene and you should do the same.
The rules for naming a function, e.g. myFunction, are the same rules as for variables,
i.e. the normal identier naming rules ( 31 alphanumeric characters and the underscore character and NOT starting with a number.)
The argument list in the function prototype is a comma separated list of variable types,
e.g.
void myFunction(float, int, float);
which indicates myFunction will be called and dened with a float, an int, and a
float in that order. Heres another example:
void realQuadraticRoots(float, float, float);
A variable type is required for any non-int. Therefore, declaring a variable position as
an int is optional. However, it is recommended that you declare the variable type for
all members in the parameter list.
133
The parameter list can also be void.

The parameter list in the function denition follows the same rules except that the
variables must be named. e.g.
void realQuadraticRoots(float a, float b, float c){/* Function definition */}
The function call (within the body of the main routine or within the body of another
function) as shown here:
int main (void)
{
.
.
.
realQuadraticRoots(a, b, c);
.
.
.
}
transmits the values of the variables in the parameter list to the function. This is
termed call by value. There is another way of doing this termed call by reference
which we will discuss later.
The variables dened in the function denition (as well as any other variables that the
function denes) are internal to that function, even if the name (identier) is identical
to a name elsewhere in the program.
Functions must all be dened in a separate code segment. You can not dene a function
within another function.
Function prototypes outside of the body of any function apply to all function calls in
that source le.
Function prototypes inside the body of any function apply only to the associated calls
within that function.
A function can return in 3 ways:
1. by running into its closing brace } (this will work only for void functions)
2. running into the following statement anywhere in the function body
return;
134

This will work only for void functions.
3. running into the following statement anywhere in the function body
return (an expression);
This is the way non-void functions must return. The parentheses around an
expression are optional. The expression should evaluate to the same variable
type that the function is expected to return.
Functions can be created independently by dierent programmers. The only information that has to be agreed upon is the return type, the function names and the
parameter lists. Otherwise, each programmer can do his or her job however he or she
pleases. In a team setting, prototypes are usually decided upon at the design stage.
An example follows.
6.3
Example: A problem tackled with teamwork
(Completed version posted on the ENG101 web site in the lectures area called quadraticRoots.cpp)
A team of 4 people, Prof (the boss) and his graduate 3 students, Bill, Jane and Eugene (the
math expert) are writing a program together.
Prof to Team: Team: We will write a code to solve the quadratic equation! Well patent
it and Ill get rich!
Team to Prof: Yeah, right!
Prof to Bill: Bill, I want you to write a bool function called realQuadraticRoots that
will return an integer, and accept the three oats a, b and c in that order. If there are
no real roots your function has to return the value false. If there are, it has to return
the value true.
Bill to Prof: Um, Prof, whats a bool??
Prof to Bill: Its a variable type that can only have two values, true or false. It is
declared and assigned value as follows:
bool truthOrDare = false;
Bill to Prof: Gotcha, Im on it. By the way, whats my cut?
Prof to Jane: Jane, I want you to write a function called numberQuadraticRoots that
will return an integer, and accept the three oats a, b and c in that order. I will only
call your routine if Bills routine tells me that there are real roots. The integer you
return should contain the number of real roots, either 1 or 2.
6.3. EXAMPLE: A PROBLEM TACKLED WITH TEAMWORK
135
Jane to Prof: OK, its clear what I have to do.

Prof to Eugene: Eugene, you do the math! I want you to...
Eugene to Prof: You probably want me to write a function called something like
twoQuadraticRoots that will return no value, and accept the three oats a, b and
c in that order and print a statement telling you what the two roots are. You will only
call your routine if Janes routine tells you that there are two real roots. Right?
Prof to Eugene: Eugene, stop interrupting! I want you to write a function called
twoQuadraticRoots that will return no value, and accept the three oats a, b and c
in that order and print a statement telling me what the two roots are. I will only call
your routine if Janes routine tells me that there are two real roots.
Prof to Team: Get to work!
Team to Prof: Whats in it for us?
Prof to Team: Stop kvetching already! Heres the main routine with function stubs.
You guys ll in the blanks. My code compiles so its absolutely perfect.
Heres the Profs main routine with stub functions:
//File: quadraticRootsStubs.cpp
#include <iostream>
//Do I need anything else, fellow programmers?
//Function stubs
bool realQuadraticRoots(float a, float b, float c)
{
}
int numberQuadraticRoots(float a,float b,float c)
{
}
void twoQuadraticRoots(float A, float B, float C)
{
}
int main(void)
{ cout << "a, b, c: ";
float a, b, c;
136

cin >> a >> b >> c;
if (realQuadraticRoots(a,b,c) && 2 == numberQuadraticRoots(a,b,c))
twoQuadraticRoots(a, b, c);
return 0;
}
Here is what the team produced.
Heres Bills code. Note that the Prof did not fully specify his task and so he let him have
it. He had to nd a way to terminate the program from a subroutine and so he had to go to
the library and do some research. (The Prof never answers e-mail from graduate students!)
Bill learned how to do this with the exit function and other denitions given in cstdlib.
bool realQuadraticRoots(float a, float b, float c)
/*********************************************************************
Purpose:
See if there are real roots
Receives:
Quadratic constants a,b,c in a*x*x + b*x + c = 0
Returns:
true if real roots, false if none
Programmer:
Bill Boole (with attitude)
Start Date:
09/17/00
Remark1:
Needs <iostream>, <cstdlib>
Remark2:
Prof needs to rethink this thing
*********************************************************************/
{ if (0 > (b*b - 4*a*c)) return false;
else if (0 == a && 0 == b)
{ cout << "Prof, you numbskull!\n"
<< "You didnt tell me what to do if a and b are both 0!\n"
<< "Go back and redesign the code!\n";
exit(EXIT_FAILURE);
}
else return true;
}
Heres Janes code. Her job was made easier by Bills code letting her know that there is at
least one solution.
int numberQuadraticRoots(float a,float b,float c)
/*********************************************************************
Purpose:
6.3. EXAMPLE: A PROBLEM TACKLED WITH TEAMWORK

Receives:
Returns:
Programmer:
Start Date:
137

1 if one real root, 2 if two real roots
Jane Justintime
09/17/00
Remarks:
Enters this routine knowing that there are roots
*********************************************************************/
{ if (0 == a || (0 == (b*b - 4*a*c))) return 1;
else return 2;
}
Eugene, who thinks hes a genius and a super-geek, likes to do things his own way. He
likes to store all the values his function receives in uppercase variables. Eugene has his own
coding style. Eugene also thinks he is smarter than the compiler and so he played with the
quadratic equation to put it into a form that he thinks will execute faster. (You can verify
for yourselves that what Eugene did was correct mathematically.)
void twoQuadraticRoots(float A, float B, float C)

/*********************************************************************
Purpose:
Receives:
Returns:
The 2 two real roots of the quadratic equation
Programmer:
Eugene Dweebowski (the genius)
Start Date:
09/17/00
Remark1:
Needs <cmath>
Remark2:
Enters knowing that there are two real roots
Remark3:
I dare ya to find something faster and better!
*********************************************************************/
{ float twoA = 2*A, twoC = 2*C, radical = sqrt(B*B - twoA*twoC) - B;
cout << "Solution 1: " << radical/twoA << "\n";
cout << "Solution 2: " << twoC/radical << "\n";
}
Some comments are in order for the Profs code. It is incomplete. It does not do anything
when there is only one solution. His code needs a little renement before it can be considered
complete. And, he has to decide what to do if all the constants are zero.
As a postscript to this story, function stubs are a good alternative to function prototypes
for small projects. Note that the Profs stub code will compile so that he can at least check
his programming. However, for larger projects, prototypes are the way to go.
138
6.4
Call by value, call by reference, reference parameters
In our present usage of function calls, the parameter lists have contained only values that
are transmitted to the function and stored in internal variables known only to that function.
This is given the name call by value. A function that is called by value operates on the
data it is given (plus whatever else can be transmitted to it by virtue of the rules scope) and
can return to its calling routine at most one value, the return value, in a return statement
of the form:
return returnValue;
This is unsatisfactory and restrictive if we need to return more than one value. An alternate
way of calling functions, a way that can return more than one thing is called call by reference.
This discussion must start, however, with a discussion of the addresses of variables, that is,
where they are stored in a computers (virtual) memory.
6.4.1
The address of a variable
We introduce the unary operator called the address operator, &. To see its usage, consider
the following program :
//File: addressLocal.cpp
#include <iostream>
int main(void)
{ float x = 1.0f;
double y = 3.0;
int
z = 2;
cout <<
"xs
value:
"ys
value:
"zs
value:
return 0;
}
// Declare and initialize a float

// Declare and initialize a double
// Declare and initialize an int
" << x << "
" << y << "
" << z << "
xs address: " << &x << "\n"

ys address: " << &y << "\n"
zs address: " << &z << "\n";
New stu:
The address operator & applied to a variable, for example, &y should be read as the
address of the double y.
6.4. CALL BY VALUE, CALL BY REFERENCE, REFERENCE PARAMETERS
139
Compiling and running it results in the following output:

xs
ys
zs
value: 1
value: 3
value: 2
xs address: 0xbffff734
ys address: 0xbffff72c
zs address: 0xbffff728
When cout prints an address, it prints it out as 0x... which is the hexadecimal (hex)
representation. The leading 0x is there just to tell you that what follows is in hex.
Note the the address stack grows from a large address towards smaller addresses.
Note the the operating system (a Linux installation in this case) puts the rst variable it
encounters (in this case a 32-bit float) at address 0xbffff734. (This address, once converted to an integer number, is something in excess of 3 billion.) This is the address in bytes,
one byte being 8 bits. A 32-bit machine has a virtual address space of 232 , or 4,294,967,296
bytes, often abbreviated as 4GB. Above this address, starting at 0xbffff738 in this case
(This boundary is machine and O/S dependent) is where the program environment resides (well discuss that later) and above that still is where the operating system resides (the
kernel) occupying the upper reaches of address space up to the 4GB limit.
The next variable, a double is 64-bits or 8 bytes in length. Note that its address is smaller
(by 8 bytes). The address space that is provided for local variables is called the stack frame
and it grows downwards. Note where the next variable is located.
Each function creates its own stack frame for putting its local variables into memory, releasing
it when the function terminates. This is what is meant by the stack getting destroyed when
a function terminates. The next function call will use this space for its own stack frame.
The executable program, the load module is at the bottom of the address space, starting at
byte 0. This area is called the text segment. Above the text segment are stored initialized
and uninitialized global variables. Above this is the heap space, the address space where
dynamic memory (memory allocated by the program) is located. It grows upwards. If your
programs and data are so big that your stack frames and heap run into each other, you
either have to redesign your program to use less space or purchase a more expensive 64-bit
computer, one with 264 bytes of array space. (In a few years, most computers will be 64-bit
computers.)
Here another example program that demonstrates that global variables are located at a
dierent place in memory:
//File: addressGlobal.cpp
#include <iostream>
140
double y = 3.0; // Declare and initialize a global double

int
z = 2;
// Declare and initialize a global int
int main(void)
{ float x = 1.0f;
cout <<
"xs
value:
"ys
value:
"zs
value:
return 0;
}
// Declare and initialize a float

" << x << "
" << y << "
" << z << "
xs address: " << &x << "\n"

ys address: " << &y << "\n"
zs address: " << &z << "\n";

xs
ys
zs
value: 1
value: 3
value: 2
xs address: 0xbffff734
ys address: 0x8049a70
zs address: 0x8049a78
Note how the global variables, y and z in this case, are located starting at a lower address
0x8049a70 (something like 1 billion). Note how the global stack grows up. Note also how
the double, y, occupies 8 bytes.
6.4.2
Call-by-value vs. call-by-reference
Imagine for a moment that the pow() does not exist and we wish to write a program that
computes the square and cube of an integer. Moreover, we wish to write a single function
that returns both the square and cube. Using what we have learned so far, we are frustrated
to discover that we can not do it! Instead we have to separate the task into two functions.
Our code would look something like:
//File: returnTwice.cpp
#include <iostream>
int square(int j)
{ cout << "In function square(), j is located at " << &j << "\n";
return j * j;
}

int cube(int k)
{ cout << "In function
return k * k * k;
}
141
cube(), k is located at " << &k << "\n";
int main(void)
{ int i = 3, iSquared, iCubed;
cout
<< "\n\nIn main():\n"
<< "
i is located at " << &i
<< "\n"
<< "iSquared is located at " << &iSquared << "\n"
<< " iCubed is located at " << &iCubed
<< "\n\n"
<< "Before the call to square and cube:\n"
<< "
i = " << i << "\n"
<< "iSquared = " << iSquared << "\n"
<< " iCubed = " << iCubed << "\n\n";
iSquared = square(i);
iCubed
= cube(i);
cout
<< "\n After
<< "
i
<< "iSquared
<< " iCubed
return 0;
the
= "
= "
= "
call to square and cube:\n"

<< i << "\n"
<< iSquared << "\n"
<< iCubed << "\n\n";
}
where I have put in a lot of cout statements to let me know where the variables are stored.
In main():
i is located at 0xbffff734
iSquared is located at 0xbffff730
iCubed is located at 0xbffff72c
Before the
i =
iSquared =
iCubed =
call to square and cube:

3
134520020
134514715
In function square(), j is located at 0xbffff728
142
In function
After the
i =
iSquared =
iCubed =

cube(), k is located at 0xbffff728
call to square and cube:
3
9
27
Note that main()s local variables i, iSquared, iCubed reside in main()s stack frame.
Also note that iSquared and iCubed are not initialized by main() and so the memory
locations of these variables contain nonsense bits, junk left over from the previous use of
those memory locations, whatever process that happened to be. The rst function called
is square(). It has one local variable called j which it locates in its own stack frame,
starting at address 0xbffff728. square() does its work happily and returns the result of
its computation to main() which then called the function cube(). cube() has one local
variable called k which it locates in its own stack frame, starting at address 0xbffff728,
the same address used by square()! Once square() returned to main(), its stack frame
was destroyed and cube() did the most ecient thing, locating its stack frame at the next
available location, the location just vacated by square().
We now introduce the method of call-by-reference. This is best done by example, so, here
is the above program converted to a single function call that has both the usual value
parameters and also something newreference parameters in the functions parameter list.
//File: returnTwoThings.cpp
#include <iostream>
void squareCube(int j, int& j2, int& j3)
{ cout << "In function squareCube():\n"
<< "j is located at " << &j << "\n"
<< "j2 is located at " << &j2 << "\n"
<< "j3 is located at " << &j3 << "\n";
j2 = j * j;
j3 = j2 * j;
return;
}
int main(void)
{ int i = 3, iSquared, iCubed;
cout
<< "\n\nIn main():\n"
<< "
i is located at " << &i
<< "\n"

<<
<<
<<
<<
<<
<<
143
"iSquared is located at " << &iSquared << "\n"

" iCubed is located at " << &iCubed
<< "\n\n"
"Before the call to squareCube:\n"
"
i = " << i << "\n"
"iSquared = " << iSquared << "\n"
" iCubed = " << iCubed << "\n\n";
squareCube(i, iSquared, iCubed);

cout
<< "\nAfter the call to squareCube:\n"
<< "
i = " << i << "\n"
<< "iSquared = " << iSquared << "\n"
<< " iCubed = " << iCubed << "\n\n";
return 0;
}
In the above example, in the parameter denition list (int j, int& j2, int& j3), the
variable j is a value parameter receiving the value of the rst parameter in the call statement,
the i in squareCube(i, iSquared, iCubed). In squareCube(), the variable j will be
placed in squareCube()s stack frame. The other two parameters in the parameter denition
list, int& j2 and int& j3, are reference parameters. What this means is that the int
variables called j2 and j3 within the function squareCube() refer to the variables located
at the addresses of the variables given to it in the function call. When the compiler calls
the function squareCube() and sees that the parameter list in the function denition is
expecting addresses, it says, Aha, squareCube() is expecting addresses for iSquared and
iCubed and so the addresses are transferred to squareCube(). squareCube() puts these
addresses in its stack frame. When the code inside squareCube() refers to the variables j2
and j3, all it is really doing is renaming temporarily the memory locations referred to in
main() as iSquared and iCubed.
Compiling and running the example results in the following output:
In main():
i is located at 0xbffff734
iSquared is located at 0xbffff730
iCubed is located at 0xbffff72c
Before the
i =
iSquared =
iCubed =
call to squareCube:
3
134519980
134514715
144
In
j
j2
j3
function squareCube():
is located at 0xbffff720
is located at 0xbffff730
is located at 0xbffff72c
After the call to squareCube:

i = 3
iSquared = 9
iCubed = 27
Note the memory locations of the variables local to main() and the fact that squareCube()
thinks that j2 and j3 live in the same place. Note also that in squareCube(), the local
variable j is located at 0xbffff720, 12 bytes below the end of main()s stack frame. Yet,
j, an integer is only 4 bytes long. Can you guess what use squareCube() is making of those
8 bytes of memory?
Finally, note that value parameters and reference parameters can be mixed in any order in the
parameter list. Just make sure, in the function denition, that all the reference parameters
names are preceded by the ampersand, & symbol.
About the only real-life example that I can come up with that relates to this is my relationship
with my stockbroker. I am the main() routine and he, the incompetent bum, is a function
called stockBroker(). Here is his function denition. Ill let you gure out how Im doing
on the stock market.
//File: stockBrokerFunction.cpp
void stockBroker(double myFee, double& alexMoney)
{ double transactionCost = 200.;
while (0 < alexMoney && 0 < myFee)
{ myFee = myFee - transactionCost; //Deduct usual transaction charge
alexMoney = alexMoney * 0.95; //Make a stupid investment
}
return;
}
However, I do want you to notice that I provide him with a fee, myFee which is his to keep.
However, he is playing with my money, alexMoney, a reference parameter, the value of which,
I am unhappy to report, is diminishing rapidly. It is still MY money, but I have given the
stockbroker the ability to change it.
6.5. THE RULES OF SCOPE
6.5
145
The rules of scope
One of the most powerful features of C++ is its implementation of the rules of scope. The
philosophy is that information (data) should be hidden from operational code unless it really
needs to know about the data. This avoids the danger of a program unit changing data that
it is not supposed toeven if it calls the variables by the same name! This is how dierent
people can develop dierent parts of a program and as long as each programmer stays within
his or her assigned duties, data does not get corrupted. C++ has also introduced the concept
of namespace to keep things separate. For this course we will always use: using namespace
std; although it is possible to use a dierent namespace. We will only use the standard
namespace in this course.
There are several hierarchies of scope:
le scope An identier declared outside of any function denition has le scope. That
is, global variables, function denitions and function prototypes (if you use them!)
placed outside of any function apply for the entire le. It is perfectly ne to dene
an int, say, outside of any function or main. This variable can then be used by all
functions and main. If separate les are linked together to form a program, the other
les will not know about le scope identiers outside of their own les.
function scope The only identier that is dened for the entire function is a label (an
identier followed by a colon :), e.g.
ThisIsALabel:
A transfer to a label can be done with the dreaded goto or within the switch statement,
neither of which will be taught in this course.
Labels are dened for the whole function and undened outside of it.
block scope A block is anything within braces, {...}. For example:
{
}
// This is a block
{
}
// This is another separate block
// This is a "yet another block"

{ // This is still another block nested
// within "yet another" block
}
}
Note the use of optional spacing to delineate the separate blocks in the above.
146

We have seen many examples of blocksfunction blocks, if/else if/else blocks,
while/do while/for blocks. We can also dene our own blocks by surrounding statements with braces, {...} to dene a segment of code that has its own scope.
Nested scope is treated in a very special way. Variables dened outside a scope block
will carry into a nested scope block unless the nested scope block declares a variable
with the identical identier. In this case the variable from the outer block is hidden
until the inner block terminates. We will see examples of this.
function-prototype-scope One way of dening a function prototype is to include variable

identiers. For example:
int myFunction(float a, float b, float c);
is a function prototype that names three identiers a, b, and c. These identiers are
only known within the function prototype. Often programmers name the variables they
are expecting in a function prototype to remember which variables go in which order.
These variable names are ignored by the compiler outside of the function prototype
statement.
This is all very confusing unless we do some examples. So...lets do some examples!
Consider the following code called scope0.cpp:
//File: scope0.cpp
#include <iostream>
void myFn(float x, float y) // MyFn declares floats x and y
{ float z; // Internal to myFn
cout << "myFn:
pre-op x = " << x << ", y = " << y << "\n";
z = x; x = y; y = z; // Variable switch
cout << "myFn:
post-op x = " << x << ", y = " << y << "\n";
}
int main(void)
{ float x = 1.0, y = 2.0; // Internal
cout << "main: pre-call x = " << x
myFn(x, y); // Main gives values to
cout << "main: post-call x = " << x
return 0;
}
to main
<< ", y = " << y << "\n";
myFn
<< ", y = " << y << "\n";
147
Compiling and running scope0.cpp results in the following output:

main: pre-call
myFn:
pre-op
myFn:
post-op
main: post-call
x
x
x
x
=
=
=
=
1,
1,
2,
1,
y
y
y
y
=
=
=
=
2
2
1
2
main declares two oats, x and y and gives myFn their values. myFn happens to call these
variables by the same name (they could have been dierent). myFns job is to switch the
values of these two variables. Within its own scope this has been accomplished but the
switch has not been eected in main. This is because the x and y in main and the x and y
in myFn are dierent.
//File: scope1.cpp
#include <iostream>
float x = 1.0, y = 2.0; // Global variables
void myFn(void) // No declarations
cout << "myFn:
pre-op x = " << x << ",
cout << "myFn:
post-op x = " << x << ",
}
int main(void)
{ cout << "main: pre-call x = " << x << ",
myFn(); // Main just calls myFn
cout << "main: post-call x = " << x << ",
return 0;
}
y = " << y << "\n";

y = " << y << "\n";
y = " << y << "\n";

y = " << y << "\n";

main: pre-call
myFn:
pre-op
myFn:
post-op
main: post-call
x
x
x
x
=
=
=
=
1,
1,
2,
2,
y
y
y
y
=
=
=
=
2
2
1
1
148
In this example x and y are dened globally (outside of any function). Both main and myFn
can access them and change them. myFns job is to switch the values of these two variables.
Within its own scope this has been accomplished and the switch has been eected in main
as well. This is because the x and y in main and the x and y in myFn are identical.
//File: scope2.cpp
#include <iostream>
float x = 1.0, y = 2.0; // Global variables
void myFn(float x, float y) // myFn declares x,y
cout << "myFn:
pre-op x = " << x << ", y = " << y << "\n";
cout << "myFn:
post-op x = " << x << ", y = " << y << "\n";
}
int main(void)
{ cout << "main: pre-call x = " << x << ", y = " << y << "\n";
myFn(x,y); // Main transfers x and y values
cout << "main: post-call x = " << x << ", y = " << y << "\n";
return 0;
}
main: pre-call
myFn:
pre-op
myFn:
post-op
main: post-call
x
x
x
x
=
=
=
=
1,
1,
2,
1,
y
y
y
y
=
=
=
=
2
2
1
2
In this example x and y are dened globally (outside of any function). main can access them
and change them. However, myFn denes its own x and y and hence the global ones are
hidden and myFn cannot access them and change them.
//File: scope3.cpp
#include <iostream>
149

int main(void)
{ float x = 1.0, y = 2.0, z = 0.0; // Local to main
cout << " main: pre-call x, y, z = "
<< x << ", " << y << ", " << z << "\n";
{ // Start of block
float z = 42.0; // Local to {} block
cout << "block: pre-call x, y, z = "
<< x << ", " << y << ", " << z << "\n";
cout << "block: post-call x, y, z = "
<< x << ", " << y << ", " << z << "\n";
} // End of block
cout << " main: post-call x, y, z = "
<< x << ", " << y << ", " << z << "\n";
return 0;
}
main: pre-call x,
block: pre-call x,
block: post-call x,
main: post-call x,
y,
y,
y,
y,
z
z
z
z
=
=
=
=
1,
1,
2,
2,
2,
2,
1,
1,
0
42
1
0
In this example x, y, and z are dened local to main. Within main there is a nested block
that denes its own scope. Since this block does not dene x and y, these variables pass
into the block. However, z is dened within the nested block and the external one is hidden.
However, the nested block has changed the values of x and y when control is passed back to
main.
//File: scope4.cpp
#include <iostream>
void myFn(void)
{ float z = 0; // Internal to myFn
cout << "myFn:
pre-op z = " << z << "\n";
150

z = 1; // Reset z
cout << "myFn:
post-op z = " << z << "\n";
}
int main(void)
{ cout << "First
myFn(); // Main
cout << "Second
myFn(); // Main
return 0;
}
call to myFn\n";
calls myFn
call to myFn\n";
calls myFn again

First call to myFn
myFn:
pre-op z =
myFn:
post-op z =
Second call to myFn
myFn:
pre-op z =
myFn:
post-op z =
0
1
0
1
In this example z is dened locally local to myFn which also initialized it to zero. Every time
myFn is called, z is reset to zero and the executable statements are processed.
//File: scope5.cpp
#include <iostream>
void myFn(void)
{ static float z = 0; // Internal to myFn
cout << "myFn:
pre-op z = " << z << "\n";
z = z + 1; // Add to z
cout << "myFn:
post-op z = " << z << "\n";
}
int main(void)
{ cout << "First
myFn(); // Main
cout << "Second
myFn(); // Main
call to myFn\n";
calls myFn
call to myFn\n";
calls myFn again
151
return 0;
}
First call to myFn
myFn:
pre-op z =
myFn:
post-op z =
Second call to myFn
myFn:
pre-op z =
myFn:
post-op z =
0
1
1
2
In this example z is dened locally local to myFn which also initializes it to zero but holds it
static. The specier static indicates that this variable is to retain its value when it leaves
a routine, preserving its value for the next entry into the routine. The initialization to zero
is only eective the rst time the routine is entered.
static is called a storage class and is a qualier on the variable type that means that its
value is to be saved, not destroyed when the function ends execution.
Consider the following code called scope6.cpp
//File: scope6.cpp
#include <iostream>
float max(float u, float v) // function max declares u,v
{ float temp = u;
u = v;
v = temp;
cout << "max: u = " < v)
{ return u;
}
else
{ return v;
}
}
int main(void){
float x = 1.0, y = 2.0, maxxy; // Internal to main
cout << "main: x = " << x << " y = " << y << "\n";
maxxy = max(x, y);
152

cout << "main: max of x and y = " << maxxy << "\n";
return 0;
}
main: x = 1 y = 2
max: u = 2 v = 1
main: max of x and y = 2
main declares two oats, x and y and gives myFn their values. myFn happens to call these
variables by dierent names (they could have been the same as in some of the previous
examples). myFns job is to switch the values of these two variables and determine which
is greater. Within its own scope this has been accomplished but the switch has not been
eected in main. This is because the x and y in main and the u and v in myFn are dierent.
6.6. PROBLEMS
6.6
153
Problems
1. Another essay question

(a) In your own words numbering 50 or less, describe when you would want to make
a function call-by-reference, rather than call-by-value.
(b) Then give an example of a function call-by-reference that accomplishes something
that you could not do using call-by-value.
(c) Give an example of
i. the function prototype,
ii. the function call, and
iii. the function denition (with an empty statement block) that you would write
for your example.
2. Simple functions
Type in the following main routine exactly as shown and add the denitions of two
functions, cubeit0 and cubeit1, each that provides the cube of the argument x. You
may not use the pow() math library function.
#include <iostream>
//Function cubeit0 goes here
//Function cubeit1 goes here
int main(void)
{ cout << "Input a number to be cubed: ";
double x;
cin >> x;
cout << cubeit0(x) << endl;
double result;
cubeit1(x,result);
cout << result << endl;
return 0;
}
Here is an example of how the program is intended to work:
154

% a.out
Input a number to be cubed: 3
27
27
%
3. Fix the bug(s)

(a) Why wont the following code compile and/or work as intended? Fix the bug(s).
#include <iostream>
void sumInts(int a, int b)
{ int i;
i = a + b;
return i;
}
int main() {
int i = 1, j = 2;
cout << "The sum of ints i = " <
#include <cmath>
float trigs(float y, &f1, &f2)
{ f1 = sin(y); f2 = cos(y);
return 0;
}
int main(void)
{ float x = 0.0, cosx, sinx;
int returnValue = trigs(x, sinx, cosx);
cout << "sin(" << x << ") = " << sinx << ", "
<< "cos(" << x << ") = " << cosx << "\n";
6.6. PROBLEMS
155
return 0;
}
4. Predict the output
What does this code print?
#include <iostream>
void cyclic(int i, int& j, int k, int& l)
{ int m;
m = i;
i = j;
j = k;
k = l;
l = m;
}
int main(void)
{ int i = 1, j = 2, k = 3, l = 4;
cyclic(i, j, k, l);
cout << "i,j,k,l : " << i << j << k << l << "\n";
return 0;
}
5. Fix the design error(s)
(a) This code attempts to dene a function that returns the maximum of two oats
and orders the two oats in descending order (bigger rst, smaller second). This
code only works some of the time. There is one design mistake that occasionally
leads to incorrect results. Assume that the scanf statement is always executed
correctly. You may not x this using a library function that determines maximum.
#include <iostream>
float fMax(float f1, float f2)
{ if (f2 > f1)
{ float temp = f2; // f2 is maximum
f2 = f1; f1 = temp; // switch them
return f1;
156

}
else return f1; // f1 must be maximum, no need to switch
}
int main(void)
{ cout << "Input 2 floats: ";
float a, b;
cin >> a >> b;
cout << "Maximum is: " << fMax(a,b) << "\n";
cout << "Descending order is: " << a << ", " <
void sort3(int i, int
{ int t;
if (i > j) {t = i;
if (j > k) {t = j;
if (i > j) {t = i;
}
j, int k)
i = j; j = t;}
j = k; k = t;}
i = j; j = t;}
int main(void)
{ int i = 5, j = 3, k = 1;
sort3(i,j,k);
cout << "In ascending order (smallest->largest) "
<< "the three ints are "
<
float squareCube(float x, float xx)
{ xx = x*x;
return xx*x;
6.6. PROBLEMS
157
}
int main(void)
{ float x = 2.0, xx, xxx;
xxx = squareCube(x,xx);
cout << "Square of x: " << xx << ", Cube of x is " << xxx << "\n";
return 0;
}
6. Roots of the quadratic equation
Write a function that is compatible with the following function call:
bool twoRealRoots = twoRoots(a,b,c,root1,root2);
The function returns true if b2 4ac > 0 and a = 0. Otherwise, it returns false. If
the function returns true, then root1 and root2 contain the roots of the quadratic
equation ax2 + bx + c = 0, which are
b b2 4ac
x1,2 =
2a
7. Write your own
function
The following algorithm may be used to calculate x =
a:
(a) Start with an initial guess: x = a

(b) Rene the guess: x = (x + a/x)/2
(c) If x and x are the same, x = a, otherwise, let x = x and go to step 2.
Write a function that calculates x = a by the above method.

8. Polar coordinates
Write a function that is compatible with the following main routine:
#include <iostream>
#include <cmath>
#include <string>
//Function definition goes here
int main(void)
158

{
float x, y, r, t;
cout << "x, y: ";
cin >> x >> y;
cout << "x = " << x << " y = " << y << " is in the " << polar(x,y,r,t)
<< ". " << "r = " << r << " t = " << t << " degrees\n";
return 0;
}
The function, polar(), called in the second cout statement above returns a string
that indicates
upper right quadrant if x 0 and y 0
upper left quadrant if x < 0 and y 0
lower left quadrant if x < 0 and y < 0
lower right quadrant otherwise
Afterthe function returns, the float called r contains the radius computed as
r = x2 + y 2. Hint. r = sqrt(x*x + y*y); After the function returns, the float
called t contains the angle in degrees computed as
= 180 tan1 (y/x)/. Hint. Use t = 45*atan2(y,x)/atan(1.0); to calculate this.
Here is an example of how it works:
red% a.out
x, y: 1 1 //User types in "1 1" in response to the "x, y: " prompt
x = 1 y = 1 is in the upper right quadrant. r = 1.41421 t = 45 degrees
9. How many bits to an unsigned int?
The following algorithm may be used to determine the number of bits that are used to
represent an unsigned int:
(a) Starting with an unsigned int initialized to 1, keep multiplying it by 2 until it
equals 0.
(b) The number of times you multiply it by 2 is related to the number of bits used to
represent the unsigned ints.
Write a function that determines the number of bits that are used to represent an
unsigned int on the computer that executes the program. Your function has to be
compatible with the following main routine.
6.6. PROBLEMS
#include <iostream>
//Function definition goes here
int main(void)
{ cout << "The unsigned int size is " << intSize()
<< " bits on this computer\n";
return 0;
}
Here is how it works:
red% a.out
The unsigned int size is 32 bits on this computer
159
160
6.7
Projects
1. Math library functions

The following table lists some of the commonly used functions in C++s math library
which can be accessed by employing #include <cmath> in the preprocessor area of a
C++ le:
Math function
log(x), log e (x), ln(x)
exp(x), ex
log10 (x)
sin(x)
cos(x)
x
|x|
Purpose
natural logarithm (base e)
exponential
logarithm (base 10)
sine of x
cosine of x
square root of x
absolute value of x
Example
double x
double x
double x
double x
double x
double x
double x
usage
= 1.;
= 1.;
= 1.;
= 1.;
= 1.;
= 1.;
= 1.;
double
double
double
double
double
double
double
y
y
y
y
y
y
y
=
=
=
=
=
=
=
log(x);
exp(x);
log10(x);
sin(x);
cos(x);
sqrt(x);
fabs(x);
Note that the argument to the trigonometric functions above must be in terms of
radians.
360 (degrees) = 2 (radians).
Write a main program that generates the table on the following page similar to that
shown. In other words, you will have to have a for-loop that somehow generates
the doubles -1.0, -0.9, ..., 1.0, calculate and output x, log(x), exp(x), log10(x),
sin(PI*x), cos(PI*x), sqrt(x), and fabs(x).
Some notes:
Note the argument of the sin() and cos(). You were given the method to
generate a good value for in the Wheres Waldo? project in the previous
chapter.
You will have to use the setw() and setprecision() functions as described in
Chapter 5.
You need one other thing to allow the printing of the redundant trailing zeros.
Put in the statement
cout.setf(ios::fixed);
before your rst use of cout.
Note that the outputs -Infinity and NaN are provided automatically by cout
when the argument to the math library functions is out of range of its denition.
6.7. PROJECTS
x
----1.0
-0.9
-0.8
-0.7
-0.6
-0.5
-0.4
-0.3
-0.2
-0.1
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
log(x)
---------Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-2.302585
-1.609438
-1.203973
-0.916291
-0.693147
-0.510826
-0.356675
-0.223144
-0.105361
0.000000
161
exp(x)
-------0.367879
0.406570
0.449329
0.496585
0.548812
0.606531
0.670320
0.740818
0.818731
0.904837
1.000000
1.105171
1.221403
1.349859
1.491825
1.648721
1.822119
2.013753
2.225541
2.459603
2.718282
log10(x)
---------Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-Infinity
-1.000000
-0.698970
-0.522879
-0.397940
-0.301030
-0.221849
-0.154902
-0.096910
-0.045757
0.000000
sin(PI*x)
---------0.000000
-0.309017
-0.587785
-0.809017
-0.951057
-1.000000
-0.951057
-0.809017
-0.587785
-0.309017
0.000000
0.309017
0.587785
0.809017
0.951057
1.000000
0.951057
0.809017
0.587785
0.309017
0.000000
cos(PI*x)
---------1.000000
-0.951057
-0.809017
-0.587785
-0.309017
0.000000
0.309017
0.587785
0.809017
0.951057
1.000000
0.951057
0.809017
0.587785
0.309017
0.000000
-0.309017
-0.587785
-0.809017
-0.951057
-1.000000
sqrt(x) fabs(x)
-------- ------NaN
1.0
NaN
0.9
NaN
0.8
NaN
0.7
NaN
0.6
NaN
0.5
NaN
0.4
NaN
0.3
NaN
0.2
NaN
0.1
0.000000
0.0
0.316228
0.1
0.447214
0.2
0.547723
0.3
0.632456
0.4
0.707107
0.5
0.774597
0.6
0.836660
0.7
0.894427
0.8
0.948683
0.9
1.000000
1.0
162
2. Strive for Perfection

A perfect number is a positive whole number that is the sum of its proper divisors.
The set of proper divisors of number N does not include N itself, but does include
1. For example, the proper divisors of 6 are 1,2,3 and 1 + 2 + 3 = 6. So, 6 is a
perfect number. In fact, it is the rst perfect number. The second perfect number
is 28 = 1 + 2 + 4 + 7 + 14. The rst 4 perfect numbers were known to the ancient
Greek mathematicians. In fact, only 37 of them are known today! A lot is known
about perfect numbers but much more remains a mystery. For example, the following
is known.
(a) The biggest known perfect number, the last discovered, can be written
23021376 (23021377 1), a number that contains 1,819,050 decimal digits.
(b) All known perfect numbers are even.
(c) All known perfect numbers are given by Euclids formula: N = 2k1(2k 1) where
k is some positive integer and 2k 1 also happens to be a prime number. This is
a special kind of prime number known as a Mersenne prime. It is a really good
idea to make use of this fact in designing your computer program, described
below.
A lot is not known about perfect numbers. It is not known whether or not
(a) there is an innite number of perfect numbers.
(b) all perfect numbers are even.
(c) all perfect numbers are given by Euclids formula.
Your job is to write a program that will:
(a) nd the rst 5 perfect numbers and print out their proper divisors.
(b) be composed of at least 2 functions additional to main, one that returns true if
its integer argument is a perfect number and a second function that prints out
the proper divisors of a perfect number. You may dene more than 2 additional
functions if you wish.
(c) complete execution in less than one minute of CPU time on a run-of-the-mill,
modern computer.
(d) have output that looks similar to the following:
% a.out
6 is a perfect number
6 = 1 + 2 + 3
6.7. PROJECTS
28 is a perfect number
28 = 1 + 2 + 4 + 7 + 14
.
.
.
163
164
3. Random integration
(a) Youll need this later!

Write a function that accepts 3 doubles (x, x0 and x1 below) and returns the
double (f (x, x0 , x1 ) below) for the function:
f (x, x0 , x1 ) =
x sin(x)
1+x
x0 x x1
If x is outside of the range of denition, a Nan must be returned.

Hint 0: A NaN is IEEEs way of saying Not a Number. NaNs can be produced
when the result of calculation using one of the oating-point representations is
undened. For example, the coding:
double zero = 0; cout << zero/zero;
produces a NaN. Try it!
Hint 1: Heres a main routine you can use to test your function:
#include <iostream>
#include <cmath>
// The function should go here...
int main(void)
{ cout << "Input x, x0, x1: ";
double x, x0, x1;
cin >> x >> x0 >> x1;
cout << "Evaluating the function at x = " << x
<< " between the limits x0 = " << x0
<< " and " << x1 << "\n";
cout << "f(" << x << ") = " << myFunction(x, x0, x1) << "\n";
return 0;
}
(b) A random way to nd an area
i. Write a function that will nd the position, xmax , which is the position where
the maximum of a function f (x) occurs. This maximum lies between two
limits x0 and x1 where x1 > x0 . Your value for xmax has to be within a
tolerance of 1012 /|x1 x0 |. You can assume that the function has only one
local maximum between x0 and x1 , that the function f (x) 0 between x0
and x1 and that the function is not a constant.
6.7. PROJECTS
165
Hint 0: Use doubles, not floats to work on your solution.

Hint 1: Life is made easy because there is only one local maximum between
x0 and x1 . That means that you can start with any x such that x0 x x1
and move a little in one direction, say, x , where is some small positive
number. If the function is smaller there, that is, f (x ) < f (x), you are
testing for the maximum in the wrong direction, so try x + .
Hint 2: Life is made dicult because the tolerance is so small. It is not
feasible to set your initial delta to 1012 /|x1 x0 |. You will have to start
with something much larger, say, 101 /|x1 x0 |. Once you have found the
position of the maximum within this tolerance, lower it to 102 /|x1 x0 | and
so on. This incremental renement technique is reasonably ecient. This is
only a suggestion. You can probably think of a much more ecient way to
do it.
ii. Write another function that determines the area under f (x) between the two
limits x0 and x1 using a technique called random sampling. Random sampling
will be discussed in class. You will need to know the position of the function
maximum before you can determine the area under the function.
iii. Complete the problem by using the following function:
f (x) =
x sin(x)
1+x
for 0 x , most of which was coded in the last problem. Use 100,000
samples to estimate the area.
HOW TO WORK ON THIS PROBLEM:
i. Try a few dierent examples for f (x) for which you know the answer (without
knowing calculus!). Good candidates would be:
A. f (x) = x, for 0 x 1.
B. f (x) = 1 x, for 0 x 1.
C. f (x) = x, for 0 x 1/2, f (x) = 1 x, for 1/2 x 1.
ii. Once you are condent that your algorithm works, then try the assigned
function:
x sin(x)
f (x) =
1+x
for 0 x
iii. Write out a pseudocode to help you design your program.
166
Chapter 7
More Variable Types, Data
Abstraction
7.1
Representation of oating-point numbers
We have seen that ints are represented by 32 bit words and that they can only represent
a nite range of integral numbers. A signed integer covers the range 231 i (231 1).
Floating numbers have their own restrictions. One bit is used for the sign of the number, 23
bits represent the mantissa1 (the precision or granularity of the number) and 8 bits are used
to dene the exponent. Consequently, floats have a range of about 1045 to about 10+38
and a resolution of about 1 part in 107 . (The particular oating-point representation can be
dierent on dierent operating systems.)
doubles are represented by 64 bits. 52 for the mantissa, one for the sign and 11 for the
exponent. The granularity is about 2 parts in 1016 and ranges from about 10324 to about
10+308 . (The particular oating-point representation can be dierent on dierent operating
systems.)
Some CPU manufacturers, for example Intel (which provides the CPUs for your laboratory
computers and, if you own them, your home computer or laptop), have adopted an IEEE
standard that represents oating point numbers with 80 bits, doing all the internal calculations at precisions higher than both floats and doubles. This variable type is called
the long double. 64 bits are used for the mantissa, one for the sign and 15 for the exponent. The granularity is about 1 part in 1019 and ranges from about 104965 to about
10+4932 . Other manufacturers have larger representations. For example, the Sun machines
maintained by CAEN have 128-bit long doubles.
1
Actually, it is assumed that the leading bit is always 1, so the representation uses 23 physical bits and
one hidden or virtual bit. All the oating-point numbers have a hidden bit.
167
168
CHAPTER 7. MORE VARIABLE TYPES, DATA ABSTRACTION
This is a representation of the dierence between float and double:
FLOAT 32-BITS
EXPONENT
MATISSA 23 BITS
8 BITS
1 BIT
SIGN
DOUBLE 64-BITS
EXPONENT
11 BITS
MATISSA 52 BITS
1 BIT
SIGN
Consider the following code called precisionFloat.cpp:

//File: precisionFloat.cpp
#include <iostream>
float precision(void)
{ float one = 1.0f, e = 1.0f, onePlus;
int counter = 0;
do
{ counter = counter + 1;
e = e/2.0f;
onePlus = one + e;
}while(onePlus != one);
cout << "Converged after " << counter << " iterations\n";
return e;
}
int main(void)
{ cout << "Float resolution = " << precision() << "\n";
return 0;
}
7.1. REPRESENTATION OF FLOATING-POINT NUMBERS
169
New stu:
1. The statement: float one = 1.0f contains the qualier "f". This means that 1.0 is
a float, typically a 32 bit constant. If the "f" were omitted, the constant would be a
double, typically a 64 bit constant.
Compiling and running precisionFloat.cpp results in the following output:
Converged after 24 iterations
Float resolution = 5.96046e-08
Consider the following code called precisionDouble.cpp:
//File: precisionDouble.cpp
#include <iostream>
double precision(void)
{ double one = 1.0, e = 1.0, onePlus;
int counter = 0;
do
e = e/2.0;
onePlus = one + e;
return e;
}
int main(void)
{ cout << "Double resolution = " << precision() << "\n";
return 0;
}
New stu:
1. The statement
double one = 1.0, e = 1.0;
declares and initializes two doubles.
170
Compiling and running precisionDouble.cpp results in the following output:

Double resolution = 1.11022e-16
Consider the following code called precisionLongDouble.cpp:
//File: precisionLongDouble.cpp
#include <iostream>
long double precision(void)
{ long double one = 1.0L, e = 1.0L, onePlus;
int counter = 0;
do
e = e/2.0L;
onePlus = one + e;
return e;
}
int main(void)
{ cout << "Long double resolution = " << precision() << "\n";
return 0;
}
New stu:
1. The statement
long double one = 1.0L, e = 1.0L;
declares and initializes two long doubles. The qualiers "L" specify that the constant
1.0 is to be considered a long double constant, an 80-bit number on my computer.
Compiling and running precisionLongDouble.cpp results in the following output:
Long double resolution = 5.42101e-20
171
Note that what is called Float resolution, Double resolution, or Long double resolution
in the above example outputs from the programs is the number that can NOT be resolved
with respect to 1.
To test the range of a float consider the following code called rangeFloat.cpp:
//File: rangeFloat.cpp
#include <iostream>
#include <iomanip>
int main(void)
{ float x = 1.0f, x0;
float y = 1.0f, y0;
int counter = 0;
do
x0 = x; x = x*2.0f; // Multiply by 2
y0 = y; y = y/2.0f; // Divide by 2
cout << setw(4) << counter << ": "
<< setw(12) << x << " :: "
<< setw(12) << y << "\n";
}while(x0 != x || y0 != y);
return 0;
}
Compiling and running rangeFloat.cpp results in the following output:
1:
2:
3:
4:
5:
2
4
8
16
32
::
::
::
::
::
0.5
0.25
0.125
0.0625
0.03125
.
.
.
125:
126:
127:
128:
129:
130:
4.25353e+37
8.50706e+37
1.70141e+38
Inf
Inf
Inf
::
::
::
::
::
::
2.35099e-38
1.17549e-38
5.87747e-39
2.93874e-39
1.46937e-39
7.34684e-40
172
.
.
.
147:
148:
149:
150:
Inf
Inf
Inf
Inf
::
::
::
::
5.60519e-45
2.8026e-45
1.4013e-45
0
New stu:
1. The output Inf is the IEEEs way of representing a number that is too large to be
represented. Inf has its own distinctive bit pattern.
2. Note that the floats appear to have more range for small numbers than large
numbers. IEEE allows for a gradual decrease to zero. These numbers below about
6e-39 are called sub-normal, with fewer bits of precision. The leading bits of the
mantissa are padded with zeros for sub-normal numbers.
To test the range of a double consider the following code called rangeDouble.cpp:
//File: rangeDouble.cpp
#include <iostream>
#include <iomanip>
int main(void)
{ double x = 1.0, x0;
double y = 1.0, y0;
int counter = 0;
do
x0 = x; x = x*2.0; // Multiply by 2
y0 = y; y = y/2.0; // Divide by 2
<< setw(12) << x << " :: "
<< setw(12) << y << "\n";
}while(x0 != x || y0 != y);
return 0;
}
Compiling and running rangeDouble.cpp results in the following output:

1:
2:
3:
4:
5:
2
4
8
16
32
::
::
::
::
::
173
0.5
0.25
0.125
0.0625
0.03125
.
.
.
1021: 2.24712e+307 :: 4.45015e-308
1022: 4.49423e+307 :: 2.22507e-308
1023: 8.98847e+307 :: 1.11254e-308
1024:
Inf :: 5.56268e-309
1025:
Inf :: 2.78134e-309
.
.
.
1073:
Inf :: 9.88131e-324
1074:
Inf :: 4.94066e-324
1075:
Inf ::
0
To test the range of a long double (at least an IEEE 80-bit one as implemented on my
machine) consider the following code called rangeLongDouble.cpp:
//File: rangeLongDouble.cpp
#include <iostream>
#include <iomanip>
int main(void)
{ long double x = 1.0L, x0;
long double y = 1.0L, y0;
int counter = 0;
do
x0 = x; x = x*2.0L; // Multiply by 2
y0 = y; y = y/2.0L; // Divide by 2
<< setw(12) << x << " :: "
<< setw(12) << y << "\n";
}while(x0 != x || y0 != y);
return 0;
174
}
Compiling and running rangeLongDouble.cpp results in the following output:
1:
2:
3:
4:
5:
2
4
8
16
32
::
::
::
::
::
0.5
0.25
0.125
0.0625
0.03125
.
.
.
16382: 2.97433e+4931 :: 3.3621e-4932
16383: 5.94866e+4931 :: 1.68105e-4932
16384:
Inf :: 8.40526e-4933
16385:
Inf :: 4.20263e-4933
.
.
.
16443:
Inf :: 1.45808e-4950
16444:
Inf :: 7.2904e-4951
16445:
Inf :: 3.6452e-4951
16446:
Inf ::
0
Now thats a lot of numbers. Try these example codes on your own machine. If you get
output that is dierent, particularly for the long doubles, Id like to know about it!
7.1.1
When is one not one? Floating point anomalies.
According to mathematics, a very strange way of writing one would be

N

1
Is this true for computers? No!

The following code demonstrates the eect.
//File: oneNotOne.cpp
#include <iostream>
#include <iomanip>
1
N
7.2. CHAR: THE CHARACTER VARIABLE
175

int main(void)
{ cout << "N? ";
double N;
cin >> N;
float fraction = 1.0/N;
cout << "Will compute " << fraction << " * " << N << " = ";
float sum = 0;
for (int i = 1; i <= N; i = i + 1) sum = sum + fraction;
cout << setprecision(19) << sum << "\n";
return 0;
}
Try running this code for small values of N and big values of N, say N = 107 or 108 . You
will get answers near 1 for small N, but something entirely dierent for big N! Why? How
could you x this?
Recall that the representation of oating-point numbers is inexact in computers. If you are
using floats with a precision of about 1 part in 107 , you will start to notice anomalies if
you try to add two things together that dier in size by 6 or 7 orders of magnitude. These
anomalies can be readily observed with floats. Converting to doubles or long doubles
treats the symptom but does not cure the disease! It is always there but often not noticeable
with higher precision numbers. So why not use doubles or long doubles all the time?
Higher precision costs in terms of storage. Sometimes memory requirements dictate the
use of lower precision. There are speed/performance issues to consider as well. This is yet
another one of the compromises one has to make in designing computer programs.
7.2
char: the character variable
A single character can be dened as the character type: char

char is a data type used to encode single characters
The following line declares a character called ch
char ch;
The following assigns the character a to the char variable. A literal character is a
character between single quotes.
176

ch = a;
The following line declares a character called ch and assigns it the character a
char ch = a;
The following statement contains an error. What is the error?
ch = "a";
Some special characters:
\n is a newline
\t is a horizontal tab
\0 is a null. It means nothing! We have to make a special character that
means nothing!
There are lots of others.
The inner secret of chars? They are really integers!
But, they are really integers but only one byte (8-bits) long.
Thus there can be at most 28 = 256 of them.
The following are all valid statements:
char c;
int i;
i = 65;
c = i;
c = 99;
Consider the program :
//File: chars.cpp
#include <iostream>
int main(void)
{ for(int i = 0; i < 256 ; i = i + 1)
{ char c = i;
if (i <= 127)
7.2. CHAR: THE CHARACTER VARIABLE
177
cout << "Int " <" << c << " in ASCII\n";
else
cout << "Int " <" << c << " beyond ASCII\n";
}
return 0;
}
Try running this program and see what it gives you!
The characters corresponding to integers 0127 are the ASCII standard character
set. (ASCII = American Standard Code for Information Interchange) which is universally adopted in English-speaking nations. The printing characters are represented by
ASCII codes 32126. Some of the ASCII codes do special things or represent special
characters. For example, 7 is the alert or bell, 8 is the backspace, 9 is the horizontal tab, 10 is a new line, 11 is a vertical tab, 13 is a carriage return, 127 is
the delete. Note the eect of some of these when running chars.cpp.
The extended ASCII character set corresponding to integers 0255 is internationally adopted and permits many of the characters from some foreign languages to be
represented.
Movement is afoot to accept the UNICODE standard which codes characters in 2
bytes (16 bits) with 216 = 65536 possibilities. This is enough to code not only every
language that exists, but even every language that has ever existed. Even Egyptian
hieroglyphics has a UNICODE representation. There is probably no need to extend
beyond 16-bit representation until we start making contact with extra-terrestrial life
forms, well, at least ones that want to communicate with us for whatever reason.
7.2.1
Character strings; the string class
The string class requires the use of #include <string>

You have seen them already!
cout << "This is a string!\n";
Declaring a string called str and assigning it a value:
string str = "This is a string!\n";
A character string is really kind of a vector in disguise.
str[0] is "T"
str[1] is "h"
etc.
178
Strings can be read in with cin and printed with cout. The cin operation continues
on the string until it sees a space or a new-line character (hitting the Return key).
Here is an example that will be discussed in class:
//File: string1.cpp
#include <iostream>
#include <string>
int main(void)
{ string response;
do
{ cin >> response;
cout << response + "\n";
}while (. != response[0]); //Note vector notation
return 0;
}
It is important to understand that the space character is considered by the cin function
to signal the end of a string.
The fact that a space character is considered by the cin operation to signal the end
of a string can sometimes be a nuisance if you want to input spaces. For example, you
might want an entire sentence to be considered a single string. The getline() string
class member function (as declared by <string>) can be used for this purpose. Here
is an example that will be discussed in class:
//File: string2.cpp
#include <iostream>
#include <string>
int main(void)
{ string sentence;
do
{ getline(cin,sentence);
cout << sentence + "\n";
}while (. != sentence[0]);
return 0;
7.3. THE VECTOR CLASS
179
The length() member function returns the length of a string, the number of characters
that it contains. The substr() member function extracts a sub-strings from a string.
Note that strings can be concatenated with the addition operator, although at least
one of the operands must be a string variable. Here is an example that demonstrates
these three new things.
//File: string3.cpp
#include <iostream>
#include <string>
int main(void)
{ cout << "Type in a string: ";
string r1;
cin >> r1;
cout << "Your response was " << r1.length() << " chars long\n";
cout << "Type in another string: ";
string r2;
cin >> r2;
cout << "Your response was " << r2.length() << " chars long\n";
cout << "Putting them together: " << r1 + " " + r2 << "\n";
cout << "Slicing and splicing them up: "
<< r1.substr(0,2) + r2.substr(4,7) << "\n";
return 0;
}
7.3
7.3.1
The vector class

Introduction to vectors
So far we have been dealing with individual floats and ints, declaring them individually,
naming them individually, manipulating them one at a time, individually. Individuality is
nice, but sometimes it is more ecient to refer to a group of similar individuals in a collective
manner. In many applications we wish to use data, many of the same type, in some sort of
systematic or regular way.
180
For example, a function of x is called f (x). An operation on f (x), such a multiplication

by a constant factor operates on all the elements of f (x), eectively an innite number of
operations! A more tangible example is; say a kind-hearted professor wanted to raise all the
grades in the class by 5%. If he had a collection of the grades in an ordered list, it would be
a lot easier (and more likely to happen) if he could multiply the whole list at once by 1.05,
rather than each grade on an individual basis.
To carry this example a little further, our class list is made up of 217 individuals (as in the
case of ENG101/200). Each of these individuals is assigned a dierent grade at the end of the
course. Lets say we wanted to calculate the class average. A computation of a class average
would involve accessing all this data in some regular fashion using a computer program.
C++ provides a way of doing this by allowing you to dene a vectora symbol that is
associated with many objects.
7.3.2
Declaring and using vectors
Suppose I wanted to declare variables associated with the individual grades in this class. I
could do the following:
int
int
int
.
.
.
int
gradeForStudent1 = 98;
This is a really dumb, repetitive way to do it! Anything that seems really repetitive for
a human can probably be done in a better way by a machine. C++ allows you to dene
everything at once using the following syntax:
vector<int> gradeForStudent(217);
which declares, in one fell swoop, 217 ints. (We could also have made them floats or
doubles. We can make vectors out of any variable type.) So, we can refer to the collection
of grades by a single identier, gradeForStudent, in a collective fashion.
But we still have work to do. We still have to enter the numbers somehow into the computer
program. One way is through assignment statements:
gradeForStudent[0] = 98; /* 1st student */
181
gradeForStudent[1] = 79; /* 2nd student */

gradeForStudent[2] = 88; /* 3rd student */
.
.
.
gradeForStudent[216] = 91; /* 217th student */
which also seems a little dumb, but indicates that we can also refer to the individuals by a
numerical index. (For example, 7of9 in the Borg collective. If you are not a Trekkie, please
ignore the last statement. However, for some strange reason, 7of9 is the only individual Borg
that I can name.)
In fact, for data of this nature, there is always some boring work to be donetyping in all
the grades. Unless there is an automatic way to generate the components of the vector, some
person has to do the boring job of typing it all in. This is called data entry.
The above example, however, shows that vector indices (the integer inside the []s) start with zero.
This leads to o-by-one errors that plague all novice (and some experienced) programmers
in C++. So, just remember that the nth element in a vector is referred to by the index n
- 1.
Another way of entering the data to the vector is by using an input loop. For example:
#include <iostream>
#include <vector>
.
.
.
const int NumberOfStudents = 217;
vector<int> gradeForStudent(NumberOfStudents);
for(i = 0; i < gradeForStudent.size(); i = i + 1)
{ cout << "Input grade for student number <> gradeForStudent[i];
}
gets the input done in another way, although someone still has to do all the typing.
There is some new stu here that needs a little explanation. The vectors that we have just
introduced are really a class in C++. A class is dened as data and functions that operate on
the data within the class. The function size() is one of the vector class member functions.
The syntax to have a member function operate on a data variable in its class is: name the
data variable and connect it via the period, ., and then name the member function with
the appropriate argument list, in this case nothing. The size() member function returns
the number of elements in the vector, in this case, 217.
182
There is one other new thing in the above code snippet, the introduction of the const
qualier to a variable. All this means is that the variable is to be held constant after its
initialization. Any attempt to change it would result in an error. This is a safe way to keep
vector sizes safe from programmer error if they are known to be xed.
The best way to input the numbers is to input them in a separate le and read this le
in during the execution of the code. This would make the code more general and the code
itself would not have to be concerned with having to enter the data faithfully. And, another
program could make use of this data. We will learn how to read and write les later on in
the course.
7.3.3
Vector syntax and rules:
You need to #include <vector> to use vectors.

The name of a vector (e.g. gradeForStudent) follows the naming convention for identiers.
Vectors can be global or local, just like individual variables, depending on where
they are declared.
The index also called the subscript is the thing that goes inside the [ ]s.
What goes inside the [ ]s is an integer or an integer expression.
If you violate the bounds of a vector (write data either below it or above it), one of
two things can happen
Either your program will crash, or
The data occupying memory adjacent to the vector will be corrupted, the program
will carry on and operate on or with corrupted data.
At least in the rst situation you know something is wrong. In the second you dont
and that could be dangerous.
The ith element of vector someVector is someVector[i - 1]
someVector[someIntegerExpression] denotes a value. It can go on the left of an
assignment operator =
When a vector is declared all of its elements are set to zero.
Once declared, it is best to think of vectors as regular variables. For example,
vector<float> vector2 = vector1;
183
where vector1 was previously dened, creates a new vector called vector2 with the
same size as vector1 and copies all the elements of vector1 to vector2.
Example: Computing averages
Suppose we have a class of 10 students. I want to compute the average grade of a class test.
Here is a way to do it.
//File: sumGradesGood.cpp
#include <iostream>
#include <vector> //Need this to use the vector "class"
int main(void)
{ const int MaxNoStudents = 10;
vector<int> grade(MaxNoStudents);
int sum = 0;
for (int i = 0; i < grade.size(); i = i + 1)
{ cout << "Grade for student #" <> grade[i];
sum = sum + grade[i];
}
cout << "\nGrade report:\n\n";
cout << "Grade for student #" <(sum)/grade.size() << "\n";
return 0;
}
And here is a sample use of the code:
Grade
Grade
Grade
Grade
Grade
for
for
for
for
for
student
student
student
student
student
#1:
#2:
#3:
#4:
#5:
1
2
3
4
5
184
Grade
Grade
Grade
Grade
Grade
for
for
for
for
for
student
student
student
student
student
#6: 6
#7: 7
#8: 8
#9: 9
#10: 10
Grade report:
Grade
Grade
Grade
Grade
Grade
Grade
Grade
Grade
Grade
Grade
Class
for
for
for
for
for
for
for
for
for
for
avg
student #1:
student #2:
student #3:
student #4:
student #5:
student #6:
student #7:
student #8:
student #9:
student #10:
is : 5.5
1
2
3
4
5
6
7
8
9
10
Example: A more rened way of computing averages

The above example requires that the C++ code know in advance how many students there
are in the class. So, if the number of students change, the code has to be recompiled. Why
not ask the user of the program to input the number of students, then declare the vector
and then do the calculation? Why not? Here is a better way to do it.
//File: sumGradesBetter.cpp
#include <iostream>
int main(void)
{ cout << "Number of students writing the test: ";
int nStudents;
cin >> nStudents;
vector<int> grade(nStudents);
int sum = 0;

{
185
cout << "Grade for student #" <> grade[i];
sum = sum + grade[i];
}
cout << "Grade for student #" <(sum)/grade.size() << "\n";
return 0;
}
A example of its use is:
Number of
Grade for
Grade for
Grade for
students writing the test: 3

student #1: 6
student #2: 8
student #3: 9
Grade report:
Grade
Grade
Grade
Class
for
for
for
avg
student #1:
student #2:
student #3:
is : 7.66667
6
8
9
Example: An even better way

Im still dissatised with the above program. It requires me to count how many students
wrote the test. Why not have the vector simply grow in size as I enter more data? Why not?
Here is an even better way to do it. In this case, the input loop terminates with a negative
grade or a grade larger than 100.
//File: sumGradesBest.cpp
#include <iostream>
186
int main(void)
{ vector<int> grade;
double sum = 0, score;
do
{ cout << "Input a students grade: ";
cin >> score;
if (0. <= score && 100. >= score)
{ sum = sum + score;
grade.push_back(score);
}
}while(0. <= score && 100. >= score);
cout << "Grade for student #" << i + 1 << ": \t" << grade[i]
<< "\n";
cout << "Class avg is : " << sum/grade.size() << "\n";
return 0;
}
An example of its use is:
Input
Input
Input
Input
Input
a
a
a
a
a
students
students
students
students
students
grade:
grade:
grade:
grade:
grade:
79
60
99
57
-1
Grade report:
Grade
Grade
Grade
Grade
Class
for
for
for
for
avg
student #1:
student #2:
student #3:
student #4:
is : 73.75
79
60
99
57
The new features of the above program is a new member function of the vector class, seen
in the statements:
187
.
.
.
vector<int> grade;
.
.
.
grade.push_back(score);
.
.
.
Note that the rst statement above simply declares a vector without a size! We are free to do
this. All it does is bind the identier grade to a vector variable type. The member function
push back(score) increases the size of the vector by one and initializes the new location
with its argument, in this case, the value of score. There is another member function called
pop back(), with no argument, that reduces a vectors size by one.
7.3.4
Vectors and functions
Remember that it is usually useful to think of vectors as simple variables. This is especially
true for vectors and functions. Take all the rules you have learned about functions and
variables and apply them to functions and vectors. To demonstrate this, consider the following example that demonstrates numerical integration. All the algorithm does is integrate
a simple function in three dierent ways. First, it puts the function into a vector with a
certain number of points based on an evenly divided mesh, or grid. Then it forms rectangles
in 3 dierent ways, using for its height the left mesh point, the right mesh point and the
midpoint. Can you guess why the midpoint method is so much better?
//File: integrate.cpp
#include <iostream>
#include <cmath>
#include <vector>
double integrateLeft(vector<double> f, vector<double> x)
{ double sum = 0;
for (int i = 0; i < f.size() - 1; i = i + 1)
sum = sum + f[i]*(x[i + 1] - x[i]);
return sum;
188
}
double integrateRight(vector<double> f, vector<double> x)
{ double sum = 0;
sum = sum + f[i + 1]*(x[i + 1] - x[i]);
return sum;
}
double integrateMiddle(vector<double> f, vector<double> x)
{ double sum = 0;
sum = sum + 0.5*(f[i] + f[i + 1])*(x[i + 1] - x[i]);
return sum;
}
int main(void)
{ cout << "Number of points in the integration mesh: ";
int mesh;
cin >> mesh;
vector<double> f(mesh), x(mesh);
for(int i = 0; i < mesh; i = i + 1)
{ x[i] = static_cast<double>(i)/(mesh - 1);
f[i] = exp(x[i]);
}
double integral = integrateLeft(f, x);
cout <<
"left = \t\t" << integral << " cf " << exp(1.0) - 1.0
<< " diff = " << integral - (exp(1.0) - 1.0) << "\n";
integral = integrateRight(f, x);
cout <<
"right = \t" << integral << " cf " << exp(1.0) - 1.0
integral = integrateMiddle(f, x);
cout <<
"midpoint = \t" << integral << " cf " << exp(1.0) - 1.0
return 0;
7.4. PROBLEMS
189
7.4
Problems
1. The inner product

Complete the program started below. You must complete the function innerProduct()
which computes the inner product of two vectors, x and y. The inner product is the
sum x0 y0 + x1 y1 + ... + xn1 yn1 where xi is the i + 1th element of x and the number
of elements in the vector is n. Here is a function stub and a main routine that you
must use to test your function:
#include <iostream>
#include <vector>
#include <cstdlib>
float innerProduct(const vector<float>& a, const vector<float>& b)
{ //Why did the function use the const qualifiers in the parameter list
//and pass the vectors by reference?
//Answer the question and insert your code after this line.
}
int main(void)
{ int vectorLength = 10;
vector<float> x (vectorLength);
cout << " x: ";
for (int i = 0; i < vectorLength; i = i + 1)
{ x[i] = rand()%100;
cout << "\t" << x[i];
}
cout << "\n";
vector<float> y(vectorLength);
cout << " y: ";
{ y[i] = rand()%10;
cout << "\t" << y[i];
}
190

cout << "\n";
cout << "Inner product is: " << innerProduct(x,y) << "\n";
return 0;
}
In the comments of your function, answer the question: Why did the function use the
const qualiers in the parameter list and pass the vectors by reference?
2. Maximum and minimum

Complete the program started below. You must complete the function minMax() which
returns both the minimum and the maximum elements of the vector. Here is a function
stub and a main routine that you must use to test your function:
#include <iostream>
#include <vector>
#include <cstdlib>
void minMax(const vector<int>& c, int& min, int& max)
{ //Why were min and max given as "reference parameters"?
}
int main(void)
vector<int> a(vectorLength);
cout << " a: ";
{ a[i] = rand();
cout << "\t" << a[i];
}
cout << "\n";
int minimum, maximum;
minMax(a, minimum, maximum);
cout << "The minimum array element is: " << minimum << "\n";
cout << "The maximum array element is: " << maximum << "\n";
7.4. PROBLEMS
191
return 0;
}
In the comments of your function, answer the question: Why were min and max given
as reference parameters ?
3. Nothing negative
Complete the program started below. You must complete the function nothingNegative()
which accepts a vector called bothSigns containing positive and negative integers and
returns a vector of equal or lesser length that contains only the non-negative integers
of the vector bothSigns. Here is a function stub and a main routine that you must
use to test your function:
#include <iostream>
#include <vector>
#include <cstdlib>
vector<int> nothingNegative(const vector<int>& a)
{ //Why can this program, when compiled with just the function stub,
//go into what looks like an infinite loop?
}
int main(void)
vector<int> bothSigns(vectorLength);
cout << " Input vector: ";
{ bothSigns[i] = rand()%201 - 100;
cout << "\t" << bothSigns[i];
}
cout << "\n";
vector<int> noNegs = nothingNegative(bothSigns);
cout << " Output vector: ";
for (int i = 0; i < noNegs.size(); i = i + 1)
{ cout << "\t" << noNegs[i];
192

}
cout << "\n";
return 0;
}
In the comments of your function, answer the question: Why can this program, when
compiled with just the function stub, go into what looks like an innite loop?
4. Vector manipulations
Write a function that accepts one called-by-reference vector of ints and returns another
vector of ints. Upon entry to the function, the vector in the parameter list can be
any size and contains ints that can have any value. Upon return, the vector in the
parameter list contains only the positive ints (> 0) in the same order. Upon return,
the returned vector contains only the negative ints (< 0) in the same order as they
appeared in the original called-by-reference vector in the parameter list, when the
function was entered.
Vector in the parameter list supplied to the function:
0 -6 -8 3 4 -8 0 8 -9 -8 8
That vector after the function returns:
3 4 8 8
Returned vector after the function returns:
-6 -8 -8 -9 -8
5. Push and pop
This is a more challenging version of the previous question. Complete the program
started below. Write a void function that accepts two called-by-reference vectors of
ints. Upon entry to the function, the rst vector can be any size and contains ints
that can have any value. The second vector is empty, its size is 0. Upon return,
the rst vector contains only the positive ints (> 0) in the same order. Upon return, the second vector contains only the negative ints (< 0) in the same order as
they appeared in the original rst vector, when the function was entered. Important:
Do this without declaring any new vectors.
#include <iostream>
#include <vector>
#include <cstdlib>
7.4. PROBLEMS
193
#include <ctime>
//Function definition goes here...
int main(void)
int size = rand()%20 + 10; //size is a RN between 10 and 29
vector<int> in(size), out;
for (int i = 0; i < in.size(); i = i + 1)
in[i] = rand()%19 - 9; //RN between -9 and +9
cout << "Original vector is size " << in.size() << endl;
cout << in[i] << " ";
cout << "\n\n";
extractNegs(in, out);
cout << "Returned vector is size " << out.size() << endl;
for (int i = 0; i < out.size(); i = i + 1)
cout << out[i] << " ";
cout << "\n\n";
cout << "Modified vector is size " << in.size() << endl;
cout << in[i] << " ";
cout << "\n\n";
return 0;
}
6. More vector manipulations
Write a function that returns a vector<int> and accepts one called-by-reference
vector<int>. Upon entry to the function, the vector in the parameter list can be
any size and contains ints that are positive, negative and 0. Upon return, the vector
that is returned contains the indices of the original vector whose original values were
negative. Upon return, the original vector in the parameter list is modied so that
it only contains its original positive (> 0) ints in the same order that they appeared
in the original vector. Write your function below or on the next page. It should be
194

compatible with the main routine below..
//Precompiler code omitted for brevity
//Function goes here
int main(void)
int size = rand()%11 + 10; //size is random between 10 and 20
vector<int> in(size);
in[i] = rand()%19 - 9; //vector element is random between -9 and 9
cout << "Original vector is size " << in.size() << endl;
cout << "Original vector: ";
for (int i = 0; i < in.size(); i = i + 1) cout << in[i] << " ";
cout << endl;
vector<int> negs = removeNegs(in);
cout << "Modified vector is size " << in.size() << endl;
cout << "Modified vector: ";
for (int i = 0; i < in.size(); i = i + 1) cout << in[i] << " ";
cout << endl;
cout << "Returned vector is size " << negs.size() << endl;
cout << "Returned vector: ";
for (int i = 0; i < negs.size(); i = i + 1) cout << negs[i] << " ";
cout << endl;
return 0;
}
red% a.out
Original vector is size 13
Modified vector: -9 0 4 9 9 9 0 3 3 -1 -6 5 0
Modified vector is size 7
Modified vector: 4 9 9 9 3 3 5
Returned vector is size 3
Returned vector: 0 9 10
7.4. PROBLEMS
195
7. Perfect squares
Complete the program started below. Write a function that returns a vector<int>
and accepts one constant called-by-reference vector<int>. Upon entry to the function,
the vector in the parameter list can be any size and contains ints that are positive
(> 0). Upon return, the vector that is returned contains the indices of the rst vector
whose values are perfect squares.
#include
#include
#include
#include
#include
<iostream>
<vector>
<cstdlib>
<cmath>
<ctime>

//Function definition goes here...
int main(void)
int size = rand()%20 + 10; //size is a RN between 10 and 29
vector<int> in(size);
{ in[i] = rand()%90 + 10;
cout << in[i] << " ";
}
cout << endl;
vector<int> out = squareIndex(in);
cout << "Returned vector is size " << out.size() << endl;
for (int i = 0; i < out.size(); i = i + 1) cout << out[i] << " ";
cout << endl;
return 0;
}
8. Sort this out
Complete the program started below. You must complete the function sort() that
has the void return type. This function accepts a vector called unsorted containing
random ints. Here is a function stub and a main routine that you must use to test
your function:
196

#include <iostream>
#include <vector>
#include <cstdlib>
void sort(vector<int>& a){}
int main(void)
{ cout << "Input the minimum and maximum random int: ";
int minRange, maxRange;
cin >> minRange >> maxRange;
cout << "Input the length of the vector: ";
int vectorLength;
cin >> vectorLength;
vector<int> unsorted(vectorLength);
cout << " Vector beforehand: \t";
{ unsorted[i] = minRange + rand()%(maxRange - minRange + 1);
cout << unsorted[i] << " ";
}
cout << "\n";
sort(unsorted);
cout << " Vector afterhand: \t";
for (int i = 0; i < unsorted.size(); i = i + 1)
{ cout << unsorted[i] << " ";
}
cout << "\n";
return 0;
}
9. A prime example
Complete the program started below. You must complete the function getPrimes()
that has the vector<int> return type. This function accepts an int called NPrimes
that is the number of consecutive prime numbers starting with 2. The return vector
contains these prime numbers in order.
You must make your function ecient by using something similar to the following test
to eliminate numbers that are not prime:
7.4. PROBLEMS
197
if (0 == n%divisor) nIsPrime = false;

where N is the candidate prime and divisor is a prime number. The divisor must
not be non-prime.
#include <iostream>
#include <vector>
#include <cstdlib>
vector<int> getPrimes(const int NPrimes){}
int main(void)
{ cout << "Input the number of primes to find: ";
int NPrimes;
cin >> NPrimes;
vector<int> primes = getPrimes(NPrimes); //Get the primes
cout << "The first " << primes.size() << " prime numbers: ";
for (int i = 0; i < primes.size(); i = i + 1)
{ cout << primes[i] << " ";
}
cout << "\n";
return 0;
}
10. No repeats
Complete the program started below. You must complete the function noRepeats()
that has the void return type. This function accepts a vector called bothSigns containing random ints and eliminates any values in it that are repeated. Here is a
function stub and a main routine that you must use to test your function:
#include <iostream>
#include <vector>
#include <cstdlib>
void noRepeats(vector<int>& a){}
int main(void)
198

{
cout << "Input the minimum and maximum random int: ";
int minRange, maxRange;
cin >> minRange >> maxRange;
cout << "Input the length of the vector: ";
int vectorLength;
cin >> vectorLength;
vector<int> hasRepeats(vectorLength);
cout << " Vector beforehand: \t";
{ hasRepeats[i] = minRange + rand()%(maxRange - minRange + 1);
cout << hasRepeats[i] << " ";
}
cout << "\n";
noRepeats(hasRepeats);
cout << " Vector afterhand: \t";
for (int i = 0; i < hasRepeats.size(); i = i + 1)
{ cout << hasRepeats[i] << " ";
}
cout << "\n";
return 0;
Chapter 8
More Data Abstraction, Arrays,
Structures
8.1
Arrays
An array is an extension of the vector concept. An array can represent a list of objects with
a single index, for example, Ai , for i = 1, 2, 3...N. For numerical computation, so called
one-dimensional arrays are used to represent lists of objects (e.g. a list of grades). Two
one-dimensional arrays (e.g. x[i],f[x]) can represent a function f (x) by specifying a grid
of points along the x-axis, x[i], or xi , and the function evaluated at those grid points, f[i],
or fi = f (xi ). These are the most common uses of one-dimensional arrays.
Vectors, which we encountered in the previous lecture, are really one-dimensional arrays.
However, arrays are a lower level construct that comes from C. Simply stated, arrays are ordered collections of things you already know about, for example, ints, floats and doubles
with a few special rules related to indexing, initializing and communicating with functions.
A vector, on the other hand, is a standard class in C++ and it comes with powerful member
functions like size(), push back() and pop back(), that operate on the data in vectors. For
this reason, we will not investigate one-dimensional arrays in detail in this course, choosing
instead to focus on the more powerful vector class when dealing with 1-D applications.
The multi-dimensional array is an extension of the one-dimensional array concept. A twodimensional array can be thought of as a grid or table of values, a set of rows and columns,
each with its own value. There are many expressions of these. A two-dimensional array can
represent a checkerboard, a chessboard, the board in game of Battleship or Minesweeper, a
football eld, or the surface of the earth. (Yes, it is two-dimensional, Christopher; its just
not at!) We could think of many examples, there being no limit to human creativity. After
this lecture, you will have enough syntax tools and rules to program a game of chess. With
a little more understanding of atmospheric physics (not programming), after this lecture
199
200
CHAPTER 8. MORE DATA ABSTRACTION, ARRAYS, STRUCTURES
you could write a program that predicts the weather! After only about 20 lectures we have
gotten this far! Now this is getting interesting! You have the tools and the rules, now apply
your creative thinking to design algorithms which themselves are only composed of three
thingsthe three pillars of algorithm design: sequence, branch and loop. It really just boils
down to this. It really is that simple and wonderful.
In more abstract terms, a two-dimensional array can represent a table of objects with two
indices, for example, Ai,j . For numerical computation in science and engineering, twodimensional arrays are used to represent tables of objects as in the examples above. Twodimensional arrays (e.g. x[i][j],f[i][j]) can represent a function of two variables, f (x, y),
by specifying a grid of points in the xy-plane, x[i][j], or xi,j , and the function evaluated
at those grid points, f[i][j], or fi,j = f (xi,j ). However, the most interesting ones for scientists and engineers are square two-dimensional arrays, also called matrices. In this course,
we will restrict our discussion to 2-dimensional arrays. There is nothing new in three and
higher dimensions that we will not see in studying 2-dimensional ones. And, the rules for
one-dimensional arrays can be obtained easily from two-dimensional arrays.
Unfortunately, C++ has not dened a new class to describe these objects. So we will have
to grapple with the more primitive array inherited from C. Later on in the course, we will be
studying Matlab. Matlab is built on two-dimensional arrays as the the fundamental variable
type and has developed a lot of functionality for it, which we will discover later. Matlab is
a very sophisticated program (used by professional engineers worldwide). However, if you
can develop the application in C++, it will always run faster, sometimes much faster. So,
we spend some eort learning how to work with two-dimensional arrays in C++. Besides,
doing so will be an excellent exercise in algorithmic thinkingthe prime reason for teaching
this course.
The ANSI standard calls for arrays of at least 12 dimensions. Believe it or not, there are
actually applications for this!
OK, enough with the sermon, already. Lets dive into the syntax before we learn how to
exploit it for creative expression. Some of the syntax was already introduced above.
8.1.1
Declaring a 2-dimensional array
Heres an example of how to declare and initialize a two-dimensional array:

const int NROWS = 3;
const int NCOLUMNS = 2;
int a[NROWS][NCOLUMNS];
This is called a 3 by 2 array or more generally as an NROWS by NCOLUMNS array. It is useful
to visualize 2-dimensional arrays with the columns numbered 0 to NCOLUMNS - 1 going from
8.1. ARRAYS
201
left to right and the rows numbered 0 to NROWS - 1 going down the page. Of course, this
visualization is completely arbitrary but helps serve as a memory aid.
a[0][0]
a[1][0]
a[2][0]
a[0][1]
a[1][1]
a[2][1]
Another useful way to think about it is as a one-dimensional array of other one-dimensional

arrays. Indeed, the notation a[NROWS][NCOLUMNS] is very suggestive of this! We can visualize
this as vectors of length NROWS aligned vertically and columns of length NCOLUMNS aligned
horizontally. Use whatever visual aid works for you. The above two suggestions work for
me.
8.1.2
Initializing a 2-dimensional array
Initializing a 2-dimensional array in the declaration statement can be done in a variety of

ways. Here are some examples:
int a[NROWS][NCOLUMNS] = {0};
Sets all of the array elements to zero. That is:
0 0
0 0
0 0
This feature is important to remember as it saves a lot of typing, a particular passion of
mine.
The brace notation may also be used:
int a[NROWS][NCOLUMNS] = {{1,2}, {3,4}, {5,6}};
The result is:
202

1 2
3 4
5 6
The brace notation is nice because it may be used to initialize the array a little more suggestively. For example,
int a[NROWS][NCOLUMNS] =
{ {1,2},
{3,4},
{5,6}
};
with the same result.
Note the use of nested braces. If a brace is not lled completely, it is lled with zeros. For
example,
int a[NROWS][NCOLUMNS] = {{1}, {3}, {5}};
results in
1 0
3 0
5 0
Note that it is important, when initializing arrays in this fashion, to make the array size
declared with ints that are xed with the const qualier. Otherwise the compiler will
complain and not complete. Arrays can be declared anywhere in a program, for example:
const int nrows = 10;
const int ncolumns = 20;
float f[nrows][ncolumns];
However, the array will have to be initialized with executable statements. Also, in contrast
to vectors, arrays are not automatically initialized to zero. It must be done either on the declaration statement or explicitly initialized to zero within executable statements. Otherwise,
the array will contain bit patterns that were left in memory following previous use.
8.1. ARRAYS
203
Heres a small program that demonstrates some of the declaration, initialization and addressing concepts.
//File: whereItsAt.cpp
#include <iostream>
int main(void)
{ const int NROWS = 3;
int a[NROWS][NCOLUMNS] = {{1}, {3}, {5}};
cout << a[0][0] << ":" << a[0][1] << "\n";
cout << a[1][0] << ":" << a[1][1] << "\n";
cout << a[2][0] << ":" << a[2][1] << "\n";
cout
cout
cout
cout
cout
cout
<<
<<
<<
<<
<<
<<
"&a[2][1]:
"&a[2][0]:
"&a[1][1]:
"&a[1][0]:
"&a[0][1]:
"&a[0][0]:
"
"
"
"
"
"
<<
<<
<<
<<
<<
<<
&a[2][1]
&a[2][0]
&a[1][1]
&a[1][0]
&a[0][1]
&a[0][0]
<<
<<
<<
<<
<<
<<
"\n";
"\n";
"\n";
"\n";
"\n";
"\n";
cout << "a: " << a << "\n";

return 0;
}
which produces the following output:
1:0
3:0
5:0
&a[2][1]: 0xbffff72c
&a[2][0]: 0xbffff728
&a[1][1]: 0xbffff724
&a[1][0]: 0xbffff720
&a[0][1]: 0xbffff71c
&a[0][0]: 0xbffff718
a: 0xbffff718
Note how this multi-dimensional array is stored. Computer memory is organized as if it were
a one-dimensional array. Thus, a multi-dimensional array has to be unfolded into a line
204
and stored. Dierent languages do this in dierent ways! C++s convention is to store it
from the outer index to the inner index, reinforcing the vectors of vectors concept. In our
two-dimensional example this means that all the columns in the rst row are stored, then
all the columns of the second row and so forth.
It is an important performance issue when writing programs to be aware of this storage order
and use the arrays in the most ecient manner. For example, the following program :
//File: fastArray.cpp
int main(void)
{ const int SIZE = 1000;
float a[SIZE][SIZE];
for(int k = 1 ; k <= 100 ; k = k + 1)
for(int i = 0 ; i <= SIZE - 1 ; i = i + 1)
for(int j = 0 ; j <= SIZE - 1 ; j = j + 1)
a[i][j] = 1;
return 0;
}
executes almost twice as fast on my computer than slowArray.cpp which is only dierent
in the ordering of the two for loops:
//File: slowArray.cpp
int main(void)
{ const int SIZE = 1000;
float a[SIZE][SIZE];
for(int k = 1 ; k <= 100 ; k = k + 1)
for(int i = 0 ; i <= SIZE - 1 ; i = i + 1)
for(int j = 0 ; j <= SIZE - 1 ; j = j + 1)
a[j][i] = 1;
return 0;
}
8.1.3
Passing a two-dimensional array to a function
Here is where multi-dimensional arrays dier signicantly from one-dimensional arrays.

When a function is passed an array it has to be given enough information to know how
to move around in the array. Thus, the dimension of all the outer indices (not the innermost) must be specied. See how this is done in the following example.
8.1. ARRAYS
//File: 2d.cpp
#include <iostream>
#include <iomanip>
#include <cmath>
#include <vector>
const int SIZE = 24;
const int MAX_PRINT = 24;
void printArray(float a[][SIZE])
{ if (SIZE <= MAX_PRINT)
{ for(int i = 0; i < SIZE; i = i + 1)
{ for(int j = 0; j < SIZE; j = j + 1)
cout << setw(2) << static_cast<int>(a[i][j]) << " ";
cout << "\n";
}
}
else cout << "Array is too big to print.\n";
return;
}
void productArray(vector<float>& y, float O[][SIZE], vector<float> x)
{ for(int i = 0; i < SIZE; i = i + 1)
for(int j = 0; j < SIZE; j = j + 1)
y[i] = y[i] + O[i][j] * x[j];
return;
}
int main(void)
{ /* Initialize the D array */
float D[SIZE][SIZE] = {0};
for(int i = 0; i < SIZE - 1; i = i + 1)
{ D[i][i] = -1;
D[i][i + 1] = 1;
}
// Initialize the I array
float I[SIZE][SIZE] = {0};
for(int i = 1; i < SIZE; i = i + 1)
for(int j = 0; j <= i; j = j + 1)
205
206

I[i][j] = 1;
// Print out D and I arrays
cout << "The \"D\" array:\n"; printArray(D);
cout << "The \"I\" array:\n"; printArray(I);
// Initialize the f and x vectors
vector<float> f(SIZE), x(SIZE);
{ x[i] = -1 + 2.0*i/(SIZE - 1);
f[i] = 1.0/(1.01 - x[i]);
}
// Operate on f with D and normalize
vector<float> d(SIZE);
productArray(d,D,f);
for(int i = 0; i < SIZE - 1; i = i + 1)
d[i] = d[i]/(x[i+1] - x[i]);
// Operate on f with I and normalize
vector<float> F(SIZE);
productArray(F,I,f);
F[i] = F[i]*(x[i] - x[i-1]);
// Print results
cout
<< "
x
"
<< "
f
"
<< "
d
"
<< "
F\n";
for(int i = 0 ; i <= SIZE - 1 ; i = i + 1)
cout << setiosflags(ios::fixed)
<< setw(11) << x[i] << " "
<< setw(11) << f[i] << " "
<< setw(11) << d[i] << " "
<< setw(11) << F[i] << "\n";
return 0;
}
The output of this program is:
8.1. ARRAYS
The "D" array:
-1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1 0
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1 1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 -1
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
The "I" array:
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0 0
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 0 0 0 0
207
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
208
1
1
1
1
1

1
1
1
1
1
1 1
1 1
1 1
1 1
1 1
x
-1.000000
-0.913043
-0.826087
-0.739130
-0.652174
-0.565217
-0.478261
-0.391304
-0.304348
-0.217391
-0.130435
-0.043478
0.043478
0.130435
0.217391
0.304348
0.391304
0.478261
0.565217
0.652174
0.739130
0.826087
0.913043
1.000000
1
1
1
1
1
1
1
1
1
1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
1 1 1 1 1 1
f
d
0.497512
0.258711
0.520009
0.283216
0.544637
0.311375
0.571713
0.343954
0.601622
0.381930
0.634833
0.426560
0.671925
0.479500
0.713621
0.542947
0.760834
0.619879
0.814736
0.714408
0.876859
0.832347
0.949237
0.982116
1.034638
1.176305
1.136925
1.434410
1.261657
1.787930
1.417129
2.290509
1.616304
3.039655
1.880621
4.228182
2.248289
6.283189
2.794654
10.317343
3.691814
20.073689
5.437352
56.080322
10.313904 1031.390381
100.000000
0.000000
1
1
1
1
1
1
1
1
1
1
1 1
1 1
1 1
1 1
1 1
F
0.000000
0.088480
0.135840
0.185554
0.237869
0.293072
0.351500
0.413554
0.479714
0.550560
0.626809
0.709351
0.799320
0.898183
1.007892
1.131120
1.271669
1.435201
1.630704
1.873718
2.194744
2.667557
3.564421
12.260068
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
1
0
0
1
1
1
0
0
0
1
1
0
0
0
0
1
What does this program do?

Using the following formula for the product of a two-dimensional array with a vector
yi = Ai0 x0 + Ai1 x1 + Ai2 x2
we see that the D array dierentiates and the I array integrates the function f (x). Yes,
you can do calculus with arrays as long as the spacing of x is such that the function f (x) does
not change too much from element to element. This is the basis of most calculations that
solve dierential equations. (You wont be tested on the information in this last paragraph!)
This is what the result looks like:
8.2. CHARACTER ARRAYS
8.2
8.2.1
10
10
10
10
10
10
10
1.0
209
f
D*f
I*f
0.5
0.0
0.5
1.0
Character arrays
One dimensional character arrays
We have seen previously that the string class is really a vector in disguise. One
can also string together characters in a more basic way using one-dimensional arrays.
One should only use this form when it is inconvenient to use the string class, as
in the example we have just been discussing, because there are some special rules
to remember. Moreover, forgetting these rules can lead to problems related to overrunning the boundaries of arrays. Indeed, some of these subtle behaviors are exploited
by destructive computer hackers (the scourge of the computer industry) to break into
computers and cause damage.
So, a character string may be represented as a one-dimensional array of single characters
that terminates with a specic character sentinel, the null character, i.e. \0. See!
There is a use for something that represents nothing! So, we always have to remember
to include this in the one-character array. If we forget to do this and try to interpret a
one-dimensional array of characters as a string, then the code, as it is operating on the
210

string, will look for the end-of-string sentinel until it nds it, possibly reading (or more
dangerously changing) memory locations that were not intended to be accessed. (This
is a major destructive hacking tool! Dont do it!) The string class handles all this
automatically for you, so that is why it is preferable to use the string class whenever
possible.
The following declares and initializes a string:

char s[4];
s[0] = H;
s[1] = i;
s[2] = \0; /* null */
Despite the fact that we have allotted 4 bytes for the string, we have only used 3 of
them because thats where we put the \0 terminator for that string. After the above
statements s[3] is undened.
Note that the statements
s[2] = \0; /* null */
s[2] = 0;
/* the same thing */
are exactly the same thing.
Strings dene their own length with the \0 terminator which acts as a sentinel for
completion of the string.
A simpler way of initializing a character string at the declaration stage is:
char s[] = {H,i,\0};
which is more convenient than the previous one because there is no need to count the
number of array elements.
A simpler way still is:
char s[] = "Hi";
which uses the doubles quotes " to enclose what is called a string literal or a string
constant. The above array s[] has length 3. The terminating sentinel \0 is put in
automatically by the compiler.
It is a common mistake to forget about the terminating sentinel \0. For example,
say we dened a character string called mistake in the following:
8.3. STRUCTURES
211
char mistake[] = {b,o,o,b,o,o};

As far as C++ is concerned, when the character array mistake[] is accessed as a
string, this is not considered to be a string whose value is "booboo"! C++ considers
this to be a string "booboo........\0", that is, it keeps going until it runs into the
terminating sentinel \0, or it runs out of memory. This is an error that can easily
crash a program or even worse, cause it to behave erratically.
Can you guess how the following program works?
//File: charArray.cpp
#include<iostream>
#include<string>
int main(void)
{ char correct[12] =
{I, A, m, C, o,
char moreData[] = "Will this
char someData[5] = {100, 97,
char mistake[8] = {A, m,
r, r, e, c, t, !, \0};
print?";
114, 110, 0};
i, s, t, a, k, e};
cout << mistake << "\n" << correct << "\n";

return 0;
}
8.2.2
Two dimensional character arrays
One can make arrays out of chars. For example, a checkerboard could be represented by:
char checkerboard[8][8];
for (int i = 0; i < 8; i = i + 1)
for(int j = 0; j < 8; j = j + 1)
checkerboard[i][j] = ; //Initialize with spaces
checkerboard[0][0] = R; //Place a red checker on the board
checkerboard[7][0] = B; //Place a black checker on the board
8.3
Structures
So far we have been introduced to many forms of variables that contain only one item of
information, bools, chars, ints, floats and doubles. These are fundamental variable
212
types in C++, useful interpretations of bits and bytes that are pre-dened in the denition
of the C++ language.
We have also seen more complex objectsclasses like vectors and strings.
We have also seen arrays of one and two and more dimensions that are made up of only one
of the above fundamental variables types but contain many items referred to by indexing
within the array.
In many cases, these pre-dened variable types are insucient to elegantly formulate a
solution to an interesting problem. Fortunately, C++ gives the programmer the ability to
make new compound variable types made up of a collection of the existing variable types.
This compound variable type is called a structure.
For example, we could dene a structure representing a person and the elements of the
structure could be a string for the persons name, an int for the persons age, floats for
the persons height and weight, more strings for the persons address and occupation and
so on. Structures are the essential composite data type for databases, which represent a
relatively large component of the software industry.
Another example relates to the car racing application of Assignment 6. Lets say that we
wanted to introduce new, racing-like features into the game, like oil slicks that reduce traction
on the road surface, or the ability to draft behind the opponents car. Every grid point
representing a part of the race track would have to carry more information, the coecient
of friction on the surface, and the direction and number of seconds past the time that a car
passed through it (so that the wind speed could be calculated). While these examples could
be done by separating the information into dierent variables, why not put them together
in a compound variable? After all, they are all related to one other.
8.3.1
The structure denition
An example of a structure denition:

// Define a structure called Student
struct Student
{ string name;
int section;
float grade;
};
The keyword struct begins a structure denition. The identier Student is called the
structure tag. It is important to note that this is a denition and not a declaration. All we
are doing at this point is informing the compiler that we will be declaring variable types of
the form Student later on in the program.
8.3. STRUCTURES
8.3.2
213
Where can structures be dened?
structs can be dened everywhere although it is conventional to dene them outside of any
programming unit such as main or another function so that all programming units will have
access to this variable type.
8.3.3
The members of a structure
The variables within the braces {}s are the members of the structure called Student. In
this case the members dened, as suggested by the naming of the member types, are a string
that will contain a name of a student, an int that will contain the section number of the
student and a oat that will contain the students grade. The use of a struct seems quite
natural in the application because the various things associated with any given student can
be variables of various types.
8.3.4
Declaring structure variables
The structure tag is used with the keyword struct to declare (and optionally initialize)
variables of the structure type. For example, if we had dened the above structure type
Student, in the main routine or elsewhere we could declare and assign as follows:
struct Student
Dan = {"Smith, Dan", 210, 79.8},
Jan = {"Brown, Jan", 201, 89.3};
declares two structures of the Student type called Dan and Jan and assigns them names,
section numbers and grades.
8.3.5
Assigning values to structure members
Individual structure members can be assigned values by using the . notation of the form
structureName.structureMember. For example, if we had not made the assignment in the
declaration statement, we could have assigned values as follows:
Dan.name
Jan.name
= "Smith, Dan"; Dan.section

= "Brown, Jan"; Jan.section
= 210; Dan.grade = 79.8

= 201; Jan.grade = 89.3
214
8.3.6
Re-assigning values of structure members
Once dened and declared, we can treat these as regular variables, reassigning them, printing them, manipulating them in various ways. For example, we could make the following
redenitions:
Jan.name = "Smith, Jan"; // Jan and Dan get married!
Jan.section = 210;
// Jan moves over to Dans section (bad idea)
We can also treat a whole structure as a single variable. For example,
Dan = Jan;
reassigns the entire Dan structure, giving each member the values of the Jan structure.
Consult the complete working code called structure0.cpp that manipulates the Dan and
Jan structures.
8.3.7
Function call-by-value of a structure
Structures can be called by value. This means that the structures are copied locally in the
called function and the structures employed by the calling routine remain untouched. Here
is an example called structure1.cpp:
//File: structure1.cpp
#include<iostream>
#include<string>
struct Student{string name; int section; float grade;};
void switchStruct(struct Student a, struct Student b)
{ // I/O statements left out
struct Student temp = a; a = b; b = temp; // Switch a and b
// I/O statements left out
return;
}
int main(void)
{ struct Student
Dan = {"Smith, Dan", 210, 79.8},
Jan = {"Brown, Jan", 201, 89.3};
8.3. STRUCTURES
215

switchStruct(Dan,Jan);
return 0;
}
8.3.8
Function call-by-reference to a structure
The more conventional way is to use call by reference for structures, saving memory. Here is
the same example as structure1.cpp but using called by reference. It is called structure2.cpp
#include<iostream>
#include<string>
struct Student{string name; int section; float grade;};
void switchStruct(struct Student& a, struct Student& b)
{ // I/O statements left out
struct Student temp = a; a = b; b = temp; // Switch a and b
return;
}
int main(void)
{ struct Student
Dan = {"Smith, Dan", 210, 79.8},
Jan = {"Brown, Jan", 201, 89.3};
switchStruct(Dan,Jan);
return 0;
}
8.3.9
Structure arrays
One of the most common uses of structs is to make arrays of them. Lets now dene an
array of structs of the form Student and make each element of the array represent one
of the students of the ENG101 course. Here is how it would be done. The code is called
structure3.cpp. It creates a class of ctitious students, assigns section numbers and grades.
Then it sorts them by name, section or grade, whatever the user wants.
216
#include<iostream>
#include<iomanip>
#include<string>
#include<vector>
#include<cstdlib>
// Define a structure called Student
const int maxNameLength = 10;
struct Student
{ char name[maxNameLength];
int section;
float grade;
};
bool switchStruct(struct Student& a, struct Student& b)
{ struct Student temp = a; a = b; b = temp; // Switch a and b
return false;
}
void sortStruct(vector<struct Student>& list, bool n, bool s, bool g)
{ bool sorted;
if (n)
{ do
{ sorted = true;
for (int i = 0; i < list.size() - 1; i = i + 1)
{ string string1 = list[i].name;
string string2 = list[i+1].name;
if (string1 > string2)
sorted = switchStruct(list[i],list[i+1]);
}
}while (!sorted);
}
else if (s)
{ do
{ sorted = true;
if (list[i].section > list[i+1].section)
}while (!sorted);
}
else if (g)
8.3. STRUCTURES
{
do
{ sorted = true;
if (list[i].grade < list[i+1].grade)
}while (!sorted);
}
return;
}
int main(void)
{ cout << "How many students in the class? ";
int nStuds;
cin >> nStuds;
vector<struct Student> myClass(nStuds);
for (int i = 0; i < nStuds; i = i + 1)
{ myClass[i].section = 200 + rand()%10;
myClass[i].grade = 0.1*(rand()%1001);
int nameLength = 2 + rand()%(maxNameLength - 2);
for (int j = 0; j < nameLength; j = j + 1)
myClass[i].name[j] = 97 + rand()%26;
myClass[i].name[nameLength - 1] = \0;
myClass[i].name[0] = myClass[i].name[0] - 32;
cout << setw(maxNameLength) << myClass[i].name << ", "
<< myClass[i].section << ", "
<< myClass[i].grade << "\n";
}
bool sortByName = false;
bool sortBySection = false;
bool sortByGrade = false;
cout << "Sort by name (0 (no) or 1 (yes))? ";
cin >> sortByName;
if (!sortByName)
{ cout << "Sort by section (0 (no) or 1 (yes))? ";
cin >> sortBySection;
if (!sortBySection)
{ cout << "Sort by grade (0 (no) or 1 (yes))? ";
cin >> sortByGrade;
}
217
218

}
sortStruct(myClass, sortByName, sortBySection, sortByGrade);
for (int i = 0; i < nStuds; i = i + 1)
cout << setw(maxNameLength) << myClass[i].name << ", "
<< myClass[i].section << ", "
<< myClass[i].grade << "\n";
return 0;
8.3.10
Example using structures and arrays: Ion transport
The following code will be described in class:

//File: survivor.cpp
#include <iostream>
#include <fstream>
#include <iomanip>
#include <cstdlib>
const
const
const
const
const
int
int
int
int
int
NX = 20;
NY = 7;
STEPS = NY;
NCELL = 1000;
TOTAL = (NX*NY*NCELL);
//
//
//
//
//
x-grid dimensions
y-grid dimensions
# of time steps
# of pairs per cell
total # of pairs
struct N {int nE; int nP;}; // # electrons, # ions

ofstream resultsOnFile; // Open an output file stream
//-------------------------------------------------------------------//---------------------- Function definitions ----------------------//-------------------------------------------------------------------// This function initializes the number of e-s and ions in each cell
void init(struct N grid[][NY])
{ for (int i = 0; i < NX; i = i + 1)
{ for (int j = 0; j < NY; j = j + 1)
{ grid[i][j].nE = NCELL;
grid[i][j].nP = NCELL;
8.3. STRUCTURES
}
}
return;
}
//-------------------------------------------------------------------// This function counts the number of e-s
int countE(struct N grid[][NY])
{ int count = 0;
for (int i = 0; i < NX; i = i + 1)
for (int j = 0; j < NY; j = j + 1)
count = count + grid[i][j].nE;
return count;
}
//-------------------------------------------------------------------// This function counts the number of positive ions
int countP(struct N grid[][NY])
{ int count = 0;
for (int i = 0; i < NX; i = i + 1)
for (int j = 0; j < NY; j = j + 1)
count = count + grid[i][j].nP;
return count;
}
//-------------------------------------------------------------------// This function recombines the e-s and ions
void recombine(struct N grid[][NY], int step)
{ for (int i = 0; i < NX; i = i + 1) // Loop over all rows
{ for (int j = step; j < NY; j = j + 1) // Loop over all columns
{ int nE = grid[i][j].nE;
for (int k = 1; k <= nE; k = k + 1) // Loop over all es
{ for (int l = 1; l <= grid[i][j].nP; l = l + 1)
{ // Potential annihilation with every ion in the cell
if (0 == rand()%2000) // Small chance of annihilation
{ // If there is an annihilation, reduce numbers by 1
grid[i][j].nE = grid[i][j].nE - 1;
grid[i][j].nP = grid[i][j].nP - 1;
break; // Break out of loop, only one life to give!
}
}
}
219
220

}
}
return;
}
//-------------------------------------------------------------------/* This function shifts the electrons to the right and counts how
many are collected. Note that, depending on the step, we can start
at the y index where the number of e-s is non-zero.
*/
int tStep(struct N grid[][NY], int step)
{ int escaped = 0;
for (int i = 0; i < NX; i = i + 1)
{ escaped = escaped + grid[i][NY-1].nE;
for (int j = NY - 2; j >= step; j = j - 1)
grid[i][j+1].nE = grid[i][j].nE;
grid[i][step].nE = 0;
}
return escaped;
}
//-------------------------------------------------------------------// This function prints out the grid
void fGrid(struct N grid[][NY],int step){
resultsOnFile << "Step: " << step << "\n";
for (int i = 0; i <= NX - 1; i = i + 1)
{ for (int j = 0; j <= NY - 1; j = j + 1)
resultsOnFile << "|" << setw(5) << grid[i][j].nE
<< "," << setw(5) << grid[i][j].nP;
resultsOnFile << "|\n";
}
return;
}
//-------------------------------------------------------------------//-------------------------------------------------------------------//--------------------------- Main routine --------------------------//-------------------------------------------------------------------int main(void)
{ struct N grid[NX][NY];
// grid is an N struct
8.3. STRUCTURES
221
init(grid); // Initialize the grid

// Output starting
cout << "Step " <<
<< " : nE = "
<< " : nP = "
<< " : nEsc =
conditions
setw(2) << 0
<< setw(6) << countE(grid)
<< setw(6) << countP(grid)
" << setw(6) << 0 << "\n";
resultsOnFile.open("N.dat");
fGrid(grid, 0); // Write the grid to file
// Loop over time steps, set = to NY grid size
int tEsc = 0; // Total number of escapees
for (int step = 0; step < STEPS; step = step + 1)
{ recombine(grid, step); // Recombine e- & ions
int nEsc = tStep(grid, step); // Take a time step
cout << "Step " << setw(2) << step + 1
<< " : nE = " << setw(6) << countE(grid)
<< " : nP = " << setw(6) << countP(grid)
<< " : nEsc = " << setw(6) << nEsc << "\n";
fGrid(grid, step + 1); // Write grid, next step
tEsc = tEsc + nEsc; // Sum total escapes
}
resultsOnFile.close(); // Close the file
cout << "Chance of survival = " << setprecision(3)
<< static_cast<float>(tEsc)/TOTAL << "\n";
return 0;
}
//--------------------------------------------------------------------
222
8.4
Problems
1. Multiplying vectors by arrays

Complete the following program. The main routine declares an int array a[ROWS][COLS].
Your program should work for any values for ROWS and COLS, not just the ones provided. Read through main() to understand what it is doing. The function print()
prints the array to cout. The function randomFill() lls the array with random ints
between 0 and 9. Hint: Use rand()%10. The function mMult() returns a vector whose
value of its ith element is:
y[i] = a[i][0]*x[0] + a[i][1]*x[1] + a[i][2]*x[2] +
+ a[i][NCOLS - 1]*x[NCOLS - 1];
...
Hint: Use for loops to do this. Supply the parameter lists for the three functions.
Answer the following question:
Why is the return vector always full of zeros?
#include <iostream>
#include <vector>
#include <cstdlib>
const int ROWS = 5; // Set to 5 but can be anything
const int COLS = 4; // Set to 4 but can be anything
void print(/* Supply parameter list*/
{//Supply code for this function...
}
void randomFill(/* Supply parameter list*/
}
vector<int> mMult(/* Supply parameter list*/
8.4. PROBLEMS
223
}
int main(void)
{ int a[ROWS][COLS]; // Declare the array
randomFill(a); // Fills array with rand()%10 ints
print(a); // Prints the array
vector<int> x(COLS); // Declare a vector
// Multiply the vector x by the array a, returning y
vector<int> y = mMult(a,x);
// Print the result
for (int i = 0; i < ROWS; i = i + 1) cout << y[i] << endl;
return 0;
}
2. Making arrays of various types
Consider the following main routine:
int main(void)
{ int I[SIZE][SIZE]
int U[SIZE][SIZE]
int L[SIZE][SIZE]
int H[SIZE][SIZE]
=
=
=
=
{0};
{0};
{0};
{0};
ident(I);
upper(U);
lower(L);
hatch(H);
print(I);
print(U);
print(L);
print(H);
return 0;
}
The 5 functions called by main() produce the output shown following, when SIZE =
10. Write the 5 functions that complete this program. Your program has to work for
any SIZE greater than 0.
(Output from the completed program, when SIZE = 10)
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
224

0 0 0 0 0 0 0 0 1 0
0 0 0 0 0 0 0 0 0 1
1
0
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
1
1
1
1
1
0
0
0
0
0
1
1
1
1
1
1
0
0
0
0
1
1
1
1
1
1
1
0
0
0
1
1
1
1
1
1
1
1
0
0
1
1
1
1
1
1
1
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
1
1
1
1
1
1
0
0
1
1
1
1
1
1
1
1
0
0
0
1
1
1
1
1
1
1
0
0
0
0
1
1
1
1
1
1
0
0
0
0
0
1
1
1
1
1
0
0
0
0
0
0
1
1
1
1
0
0
0
0
0
0
0
1
1
1
0
0
0
0
0
0
0
0
1
1
0
0
0
0
0
0
0
0
0
1
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
0
1
1
1
1
1
1
1
1
1
1
1
0
1
0
1
0
1
0
1
0
3. Vectors and arrays

Complete the program started below.
The main routine declares an int array a[ROWS][COLS]. Your program should work for
any values for ROWS and COLS, not just the ones provided. Read through main() to
understand what it is doing. The function randomFill() lls the array with random
ints between 0 and 3 inclusive. Hint: Use rand()%4. The function sumRowsAndCols()
8.4. PROBLEMS
225
sums the rows and columns putting the results into two separate vectors, sumRows, a
vector ROWS in length containing the sum of each row, and sumCols, a vector COLS in
length containing the sum of each column: The function print() prints the array to
cout and the vectors as shown in the example below. Note, pretty formatting is not
important, but the numbers have to be correct. Supply the parameter lists for the
three functions.
Example use:
>
2
3
1
3
9
a.out
2 1 3
2 3 0
3 0 0
1 3 1
- - 8 7 4
3
2
1
2
8
:
:
:
:
11
10
5
10
#include <iostream>
#include <vector>
#include <cstdlib>
const int ROWS = 4; // Set to 4 but can be anything
const int COLS = 5; // Set to 5 but can be anything
int main(void)
{ int a[ROWS][COLS];
randomFill(a);
vector<int> sumRows(ROWS);
vector<int> sumCols(COLS);
sumRowsAndCols(a,sumRows,sumCols);
print(a,sumRows,sumCols);
return 0;
}
4. More vectors and arrays
Consider the following main routine:
const int ROWS = 10; //Constant defined as global
226

const int COLS = 8;
//Constant defined as global
int main(void)
{ int a[ROWS][COLS] = {0};
vector<int>
for(int i =
vector<int>
for(int i =
r(ROWS);
0; i < r.size(); i = i + 1) r[i] = rand()%4; // 0 <= r[i] <= 3
c(COLS);
0; i < c.size(); i = i + 1) c[i] = rand()%4; // 0 <= c[i] <= 3
outerProduct(a,r,c);
print(a,r,c);
return 0;
}
Write the two functions outerProduct() and print(). outerProduct() computes
the outer product of two vectors, as follows:
a[i][j] = r[i]*c[j];
where r[i] and c[j] are vectors. print() prints out the vectors and array as shown
below. You do not have to concern yourself with pretty formatting, but the numbers
should come out in the order shown. Your program has to work for any positive values
for ROWS and COLS.
Here is how it works:
red% a.out
v:
1 0 2 2 3 2 1 2
---------------r[0]: 3| 3 0 6 6 9 6 3 6
r[1]: 1| 1 0 2 2 3 2 1 2
r[2]: 0| 0 0 0 0 0 0 0 0
r[3]: 3| 3 0 6 6 9 6 3 6
r[4]: 1| 1 0 2 2 3 2 1 2
r[5]: 3| 3 0 6 6 9 6 3 6
r[6]: 3| 3 0 6 6 9 6 3 6
r[7]: 1| 1 0 2 2 3 2 1 2
r[8]: 1| 1 0 2 2 3 2 1 2
r[9]: 1| 1 0 2 2 3 2 1 2
5. Complex structs
This problem uses a struct dened as follows:
8.4. PROBLEMS
227
struct Complex {double real; double imag;};

We may use struct Complexs to dene complex numbers. A complex number is
really a pair of numbers, the real part and the imaginary part, as suggested by
the above struct denition. If a = (aR , aI ) and b = (bR , bI ) are two complex numbers,
their product is also a complex number ab = (aR bR aI bI , aR bI +aI bR ) Write a function
that accepts two called-by-reference vector<struct Complex>s, both of the same size
and returns a vector<struct Complex> that contains the product according to the
above rule.
6. Using structure lists
Consider the following structure denition:
struct Student {char name[10]; int section; float grade;};
Write a function compatible with the following function call:
adjustGrades(list);
where list is a vector<struct Student> with size > 0. Your function will do the
following:
(a) Raise all the grades in the class 5% to a maximum of 100.
(b) Give everyone in the class whose name starts with Al a grade of 100.
(c) Sort the list by grades, highest to lowest.
228
8.5
Projects
1. Battleship! Sink this one!

You will write C++ code that will allow you to play the game Battleship! on your
computer following the rules given below.
The game is played on a 10 by 10 array of chars. A battleship is represented by a series
of sizeShip contiguous (joined) points that lie in a horizontal, vertical or diagonal line.
This can be represented by some special character in a 10 by 10 character array, say S.
For example, if the array everywhere was initialized to the character . and you wrote,
board[2][3] = S; board[3][4] = S; board[4][5] = S; board[5][6] = S;
it would represent a battleship on a diagonal occupying the above indices.
For each ship you will ask the computer to generate sizeShip such contiguous points
on a line starting from a random position within the entire array. For example, if you:
#include<cstdlib>
in the preprocessor area, the code:
i = rand()%10; j = rand()%10; board[i][j] = S;
picks a random point in the array called board and assigns it the value S, indicating
that it is one of the points representing a ship. This is the rst point on the ship.
The hard part is to pick sizeShip - 1 more points on a line that are contiguous with
the rst point and that remain within the array. This you have to gure out on your
own. The orientation of the battleship should be random as well. You have to make
the part of the program that determines the ships position and orientation a separate
C++ function and the array name must be an argument (in the parameter list) of this
function. When you come to the part where you are putting more than 1 ship on the
board, it gets even harder because no 2 ships can occupy the same space.
Now comes the fun part. You (the user of the program) have 50 + shipSize*nShips/2
bombs to toss at the battleship, where nShips is the number of enemy ships. The
battleship position and orientation are random and are kept secret from you until the
code nishes, so you do not know where it is while you are playing. You have to drop
the bomb on the array, which is equivalent to selecting an element of the array, by
asking the user to provide two indices that lie within the board. If that array element
is an S, you have hit the ship and the program informs you that you have hit the ship.
(See the example below.) You have to hit all locations of a ship to sink it and you have
to sink every ship to win the game.
If you hit a ship once, the game strategy simplies. You look for adjacent points (8
possibilities). If you hit it again, you know the orientation. Now, all you have to do is
make sure you get all the points that lie on the line.
Win or lose, print out a representation of the board that shows clearly: the location
of the ship, the points of the ship that were struck, and the points where bombs were
dropped but missed.
8.5. PROJECTS
229
Here is an example of a typical play of the game. Your computer output does not have
to look exactly as follows. However, you have to follow the above rules explicitly.
Number of enemy ships? 1
Size of the enemy ships? (<= 10) 4
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
Bombs
away
away
away
away
away
away
away
away
away
away
away
away
away
away
#
#
#
#
#
#
#
#
#
#
#
#
#
#
1:
2:
3:
4:
5:
6:
7:
8:
9:
10:
11:
12:
13:
14:
i
i
i
i
i
i
i
i
i
i
i
i
i
i
j:
j:
j:
j:
j:
j:
j:
j:
j:
j:
j:
j:
j:
j:
3
7
4
3
9
6
2
2
2
3
3
3
4
5
4
2
1
6
2
0
6
7
5
6
7
5
4
3
Ouch, that hurts!
Ouch, that hurts!

Ouch, that hurts!
Ouch, that hurts!
Ugh, ya got me! Glub, glub, glub...

. . . . . . . . . .
. . . . . . . . . .
. . . . . o * o . .
. . . . o * o o . .
. o . . * . . . . .
. . . * . . . . . .
o . . . . . . . . .
. . o . . . . . . .
. . . . . . . . . .
. . o . . . . . . .
A "." is "open sea".
An "*" is a "hit on an enemy ship".
An "S" is a "surviving enemy ship".
An "o" is a "missed bomb".
230
2. Ladies and Gentlemen: Start your engines!

Write a computer program that simulates a race car negotiating an S-curve track and
either crashing or reaching the nish line. The game is played by either one player
racing against the clock, or two players competing against each other.
The game starts with a printout of a 2-D grid representing the track made up of a twodimensional array of characters (chars) with 52 rows and 72 columns. The picture on
the following page represents the starting condition for a one-person game. The grid can
be printed to the terminal screen by writing the entire char grid to cout. If you have
trouble seeing the whole grid in a single window, try using <Control>-<right mouse button>
in an XTerm window and resizing the window to the whole screen.
The outer boundary of the track and the two impenetrable barriers are represented by
Xs, the open track by spaces , the nish line by Fs (lower right hand corner)
and the car, pictured at its starting location, grid[1][1] (upper left hand corner), by
an O. In the following picture, the rows go down the page top to bottom and the
columns go left to right.
Heres how the game is played. The user is continually prompted (until the game
ends) for a component of horizontal (positive is left to right) and vertical acceleration
(positive is top to bottom). (You imagine that this prompt is given once per second,
that is, the user has a chance to change the acceleration once/second.) The user
responds with either -1, 0, or 1 for each of these, eectively steering, accelerating and
braking the car. If the user responds with a number outside of this range, the car
crashes and the game is over, because either the engine explodes or the brakes and
tires burn upreal concerns for real race cars.
The velocity in both the vertical and horizontal directions starts at 0 and increases or
decreases after each prompt, for example, xVelocity = xVelocity + xAcceleration,
and the same for the y direction. Then, both components of the position are changed,
for example, xPosition = xPosition + xVelocity, and the same for the y direction. A car crashes when its position takes it outside the racetrack (except when it
crosses the nish line) or into a wall or barrier, or an improper acceleration was input.
Crossing the nish line wins the race.
8.5. PROJECTS
231
The starting grid in a one-person game (note prompt for acceleration at bottom):
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
XO
XXXXXXXXXXXXXXXXXXXX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXXXXXXXXXXXXXXXXXXXXXXXX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXFFFFFFX
Horizontal and vertical acceleration (-1,0,1):
To start the programming, write a working main routine and a function called initRace
that initializes the grid to the given starting conguration, and a function called
printRace that prints the grid. The walls and barriers have to be exactly as shown
in the gure on the last page. The barrier on the left is 20 columns wide and 35 rows
high. The barrier on the right is 25 columns wide and 35 rows high.
232

Then, program a one-person and a two-person game. Keep track of the total time for
each player. If a player crashes he is out of the race. The rst across the nish line
wins! Hint: Program a one-person game rst, then adapt your program for two people.
Warning: This game is addictive!
Consider modifying the program to do one of the following. These are quite challenging
but well within your capabilities.
In a one-person game, keep track of the real time it takes to respond and adjust
the algorithm to work with real time. (Use the standard time library ctime. You
will have to look up the standard time functions in a dierent book.)
Program a one-person-against-the-computer game with multiple plays. Initially,
the computer makes random moves. Once the player has crossed the nish line,
the computer learns the players solution and tries to improve it.
Program some nice graphics for this game. (Good opportunity to read ahead in
Matlab.)
Program a two-person game but one that generates many barriers randomly in a
game with very many columns that scrolls down the screen.
Dream up some other nice variant of the game.
Sample plays of the race on the following pages
8.5. PROJECTS
User input: (0,1),(0,1)(0,1),(0,1),(0,1),(1,1),(1,1),(1,1),(1,1),(1,1)
XO
X
XO
X
X
X
XO
X
X
X
X
X
XO
X
X
X
X
X
X
X
XO
X
X
X
X
X
X
X
X
X
XO
X
X
X
X
X
X
X
X
X
X
X
X O
X
X
X
X
X
X
X
X
X
X
X
X
X
X
O
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
O
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
O
X
X
X
X
X
X
X
X
X
XXXXXXXXXXXXXXXXOXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXFFFFFFX
Crashed after 10 seconds
Note the car buried in the wall! (Ouch!)
233
234

Almost. Lost control at the very end.
XO
O
O
X
XO
O
O
X
X
X
XO
O
O
X
X
X
X
X
XO
O
O X
X
X
X
X
X
X
XO
O
OX
X
X
X
X
X
X
X
X
XO
O
O
X
X
X
X
X
X
X
X
X
X
X O
O
X
X
X
X
X
X
X
X
X
X
X
X
X
X
O
O
X
X
X
X
X
X
X
X
X
X
X
X
O XXXXXXXXXXXXXXXXXXXX
O
X
X
X
X
X
X
X
X
X
X
O
O
X
X
X
X
X
X
X
X
O
O XXXXXXXXXXXXXXXXXXXXXXXXX
X
X
X
X
X
X
O
O XXXXXXXXXXXXXXXXXXXXXXXXX
X
X
X
X
O
O
X
X
O O
X
XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXFFFFFFX
Crashed after 33 seconds
8.5. PROJECTS
Richard Petty, eat your heart out!
(Solution by Ali Mehmet Kutman, Fall 2000 ENG101/CoE student at UoM)
XO
X
XO
X
X
X
XO
X
X
X
X
X
XO
X
X
X
X
X
X
X
XO
X
X
O
O
X
X
O
O
X
X
X
X
O
O
X
XO
X
X
X
X
OXXXXXXXXXXXXXXXXXXXXXXXXXO
X
X
X
X
X
X
X
XO
O XXXXXXXXXXXXXXXXXXXXXXXXX O
X
X
X
X
X
X
X
X
X
X O
O
XXXXXXXXXXXXXXXXXXXXXXXXX O
X
X
X
X
X
X
X
X
O
O
X
X
X
X
X
X
O XXXXXXXXXXXXXXXXXXXX O
X
X
X
X
O
O
X
X
O
O
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
Finished after 31 seconds (Can anyone beat Alis 31 second solution?)
3. Mastermind: human vs. machine!

For this project you will write a variant of the game called Mastermind.
Here is a description of the game:
The computer picks 4 random numbers between 1 and 9 inclusive
Hint: hiddenNumber = rand()%9 + 1
235
236

for each of two players, those players being the human (person using the program)
and the machine (the computer). Some of these numbers can be the same, or they
can all be dierent, depending on what rand() returns.
On each play, the human picks 4 numbers and the machine picks 4 numbers. If
one of the numbers is correct and in the right position, the hidden number is
revealed.
The rst to reveal all 4 numbers wins the game. There are three possible outcomes
1) human wins, 2) computer wins, 3) its a tie.
The playing boards must be represented by integer arrays.
The algorithm that the machine employs to play the game must be designed so that
the machine is guaranteed to win in 9 moves, unless the human has already won.
Otherwise, you are free to choose whatever algorithm you like. Look at the example below. It gives a big hint as to how to design a simple algorithm that is
guaranteed to work well.
HUMAN
=====
? ? ? ?
------0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
MACHINE
=======
? ? ? ?
------0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
Enter your guess: 1 2 3 4

HUMAN
MACHINE
=====
=======
? ? ? ?
? ? ? ?
------------1 2 3 4
1 1 1 1
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
HUMAN
MACHINE
=====
=======
? ? ? 5
? 2 ? ?
------------1 2 3 4
1 1 1 1
2 3 4 5
2 2 2 2
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
8.5. PROJECTS
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0

HUMAN
MACHINE
=====
=======
? ? ? 5
? 2 ? ?
------------1 2 3 4
1 1 1 1
2 3 4 5
2 2 2 2
3 4 5 5
3 2 3 3
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
HUMAN
MACHINE
=====
=======
? 5 ? 5
4 2 ? ?
------------1 2 3 4
1 1 1 1
2 3 4 5
2 2 2 2
3 4 5 5
3 2 3 3
4 5 6 5
4 2 4 4
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
HUMAN
MACHINE
=====
=======
5 5 7 5
4 2 5 ?
------------1 2 3 4
1 1 1 1
2 3 4 5
2 2 2 2
3 4 5 5
3 2 3 3
4 5 6 5
4 2 4 4
5 5 7 5
4 2 5 5
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
Curses, human, you win!
237
238
Chapter 9
Miscellaneous Topics
9.1
Generating random numbers
The C++ standard library, <cstdlib>, provides a way of producing a random number, actually a pseudo random number using the rand() function. The rand() function produces
a pseudo random integer between 0 and RAND MAX, inclusive. The constant RAND MAX is also
dened in <cstdlib>.
Here is an example program, called rand0.cpp that calls rand(). Note the inclusion of
<cstdlib>.
//File: rand0.cpp
#include <iostream>
#include <iomanip>
#include <cstdlib> // Need this for the rand() function
int main(void)
{ cout << "RAND_MAX = " << RAND_MAX << "\n\n";
cout << "The first 10 random numbers are...\n";
for(int i = 1; i <= 10; i = i + 1)
cout << "Random number "
<< setw(2)
<< i << " = "
<< rand() // Random int between 0 and RAND_MAX inclusive
<< "\n";
return 0;
}
239
240
CHAPTER 9. MISCELLANEOUS TOPICS
Compiling and executing this results in:

RAND_MAX = 2147483647
The first 10 random numbers are...
Random number 1 = 1804289383
Note that RAND MAX and the implementation of rand() is system dependent. You should not
expect the same values on a dierent computer or even a dierent version of the operating
system on the same computer.
If you run this program again, you will note that the same random numbers are produced!
This sort of goes against the idea that rand() produces random numbers! Actually, this is
a benet. Since the behavior is predictable, it makes it easier to debug. Imagine if you were
trying to debug a program that always gave dierent results! You would go bonkers trying
to nd the cause of the error. So, predictable random numbers are good. But, this is why
they are called pseudo random numbers. They are made from a mathematical prescription
that is remarkably simple. Yet, they pass many dierent tests of randomness. They also
fail some tests as well but not ones that we are likely to try.
If you are not happy with the same random sequence every time, you can seed the random
number to produce another sequence. The <cstdlib> function srand() does this. All you
have to do is provide a positive integer to the argument list of srand(). Heres an example
called rand1.cpp:
//File: rand1.cpp
#include <iostream>
#include <iomanip>
#include <cstdlib> // Need this for the rand(), srand() functions
int main(void)
cout << "Input any int to seed the random number generator: ";
int seed;
9.1. GENERATING RANDOM NUMBERS
241
cin >> seed;

srand(seed);
for(int i = 1; i <= 10; i = i + 1)
<< setw(2)
<< i << " = "
<< "\n";
return 0;
}
Compiling and executing the above code results in:
RAND_MAX = 2147483647
Input any int to seed the random number generator: 123456789
Note the dierent sequence because the user provided the seed 123456789 to srand().
One way to make the start-up of a program that uses rand() appear to be dierent every
time, is to seed the random number generator with the int that is provided by the time()
function. The time() function is part of the standard library <ctime>. Heres an example
code called rand2.cpp that uses it:
//File: rand2.cpp
#include <iostream>
#include <iomanip>
#include <cstdlib> // Need this for the rand() and srand() functions
#include <ctime> // Need this for the time() function
242
int main(void)
srand(time(NULL)); // Seed the random number generator
for(int i = 1; i <= 10; i = i + 1)
<< setw(2)
<< i << " = "
<< "\n";
return 0;
}
Try running rand2.cpp and verify that you get dierent numbers every time. This unique
start-up feature is very useful for programming games, in order to produce a game that starts
dierently every time.
The interesting uses of rand() involve making decisions based on the output of rand() and
this simulates random behavior. In many applications you can use the modulus function to
do this. For example,
card = rand() % 13 + 1
has a one in 13 chance of being a one, a one in 13 chance of being a two, etc. Thus, you
could use the above statement to simulate the random dealing of cards and you could assign
a one to an ace, a 2 to a deuce, 11 to a jack, etc. If you were dealing with a nite
series of cards, however, you would have to take some care that you do not deal, say, 5 aces
and that would require some extra care with the coding. We wont go into further detail
here on this topic except to say that dealing 5 aces can get you into BIG trouble in certain
circumstances.
Another way of using random numbers, more common to scientic disciplines, is to convert
the random integer to a random double or oat, usually converted so that the number lies
between 0 and 1 (may be inclusive or exclusive of one or both of the endpoints).
Here is an example code, called rand3.cpp, that converts the random integer to a double
between 0 and 1 inclusive:
//File: rand3.cpp
#include <iostream>
#include <iomanip>
#include <cstdlib> // Need this for the rand() and srand() functions
#include <ctime> // Need this for the time() function
243
int main(void)
int nRandoms = 10;
cout << "The first " << nRandoms << " random numbers are...\n";
for(int i = 1; i <= nRandoms; i = i + 1)
{ double randomDouble = rand()/static_cast<double>(RAND_MAX);
<< setw(2)
<< i << " = "
<< randomDouble // Random double between 0 and 1 inclusive
<< "\n";
}
return 0;
}
Heres an example of its use:
RAND_MAX = 2147483647
Random number 1 = 0.174225
Random number 2 = 0.130327
.
.
.
Application: Calculating by throwing darts
If you are as bad a dart player as I am, throwing a dart at a circular object is truly a random
process. Imagine that you have a circle inscribed in a square. It is easy to show that the
ratio of the area of the circle to the area of the square is /4. This suggests a way to estimate
what is by throwing darts! However, it only works well if you are a bad dart player, like
me.
Set up a circular dartboard on a wall. Draw a square on the wall so that the circle dening
the dartboard touches each side of the square only once. Now start throwing the darts
(badly). Ignore all the darts that land outside of the square. Throw a lot of darts! Count
the ones that land on the dartboard and all the ones that land within the square. The ratio
should be close to /4. And, your estimate gets better the more darts you throw. Warning:
244
Dont do this in your dorms. Dont do this at home. Try it outside against a fence
or something and make sure that other people, pets, your mothers favorite vase, etc. do not
get damaged.
A much safer way is to do it with a computer code. The following code, called randomPi.cpp
calculates by random sampling:
//File: randomPi.cpp
#include <iostream>
#include <cstdlib> // Need for rand() and srand()
#include <cmath> // Need for pow() and atan()
#include <ctime> // Need for time()
/* Define a conversion function that converts the int returned from
rand() to a double between -1.0 and 1.0 inclusive
*/
#define MYRAND (2.*rand()/RAND_MAX - 1.)
int main(void)
{ int nDarts = 10000; // Number of darts to throw
int nInside = 0; // Counter for those that land inside
for(int i = 1; i <= nDarts; i = i + 1)
{ double xDart = MYRAND;
double yDart = MYRAND;
if (pow(xDart,2) + pow(yDart,2) <= 1.0) nInside = nInside + 1;
}
double pi = 4.0*nInside/nDarts; // Estimate for Pi
double Pi = 4.0*atan(1.0); // This is how you can get a very
// accurate estimate of the real Pi!
cout << "The estimated value of Pi is " << pi << "\n";;
cout << "The true value of Pi is " << Pi << "\n";;
return 0;
}
There are a few new features in this program that you have not seen before.
The statement
#define MYRAND (2.*rand()/RAND_MAX - 1.)
245
in the pre-processor area replaces every occurrence of MYRAND throughout the rest of
the le with (2.*rand()/RAND_MAX - 1.). This happens before compilation. It is
exactly as if you took an editor, looked for every occurrence of MYRAND and pasted in
its replacement (2.*rand()/RAND_MAX - 1.). You can be as creative as you wish with
#define s, but make sure that you understand this as a cut and pasting procedure.
Generally, you should avoid #define s because the C++ parser does not help you
much with syntax errors. But, #define s are a part of the language and using them
can make life a lot easier sometimes. Just be careful with them.
The expression (2.*rand()/RAND_MAX - 1.) produces a random double between -1
and 1 inclusive. Why? rand() produces an int between 0 and RAND_MAX, inclusive.
Following the rules of precedence and the rules of variable type promotion multiplying
it by 2. produces a random double between 0. and 2.*RAND_MAX. Dividing
by RAND_MAX produces a random double between 0. and 2.. Finally, subtracting
1. from it produces a random double between -1. and 1.. There is a lot going
on in this simple-looking expression.
When you throw a dart on a 2-dimensional board, you are really choosing randomly
an x-coordinate and a y-coordinate. So the statements:
double xDart = MYRAND; double yDart = MYRAND;
pick a random point in a square dened by the corners:
(x, y) = (1, 1), (1, 1), (1, 1), (1, 1).
The pow() function, part of the standard library <cmath>, takes the power of the rst
argument to degree of the second argument. For example, pow(x,3) computes x3 .
So, the statement:
if (pow(xDart,2) + pow(yDart,2) <= 1.0) nInside = nInside + 1;
computes x2 + y 2 and compares it to 1. If x2 + y 2 1, the dart has landed in the circle
and it is counted.
Finally, the standard library math function atan(), also part of <cmath> computes
the arc tangent of its argument, assumed to be in radians. Since tan1 (1) = /4,
4.0*atan(1.0) is a very accurate determination of , as good as the math library can
provide, at least for a double.
Heres what 20,000 points tossed at a circle look like, generated by the program above.
246
9. MISCELLANEOUS
TOPICS
20000 random CHAPTER
points within
a square
n(inside circle)/n(total) = /4
9.1.1
Example: Integrating functions by random sampling
The same idea may be used to integrate functions by random sampling. The following code,
called randomIntegrate.cpp integrates exp(x) for 0 x 1.
//File: randomIntegrate.cpp
#include <iostream>
#include <iomanip>
#include <cstdlib> // Need for rand() and srand()
#include <cmath> // Need for exp(), the exponential function
// and abs(), the absolute value function
/* Define a conversion function that converts the int returned from
rand() to a double between 0.0 and 1.0 inclusive
*/
#define MYRAND (static_cast<double>(rand())/RAND_MAX)
247
int main(void)
{ int nDarts = 1000000000, under = 0;
double Ans = 1 - exp(-1);
for(int i = 1; i <= nDarts; i = i + 1)
{ double xDart = MYRAND;
double yDart = MYRAND;
if (yDart <= exp(-xDart)) under = under + 1;
if (0 == i%1000000) // Every million times
{ double ans = static_cast<double>(under)/i;
cout << setiosflags(ios::fixed) << setprecision(8)
<< "#" << setw(10) <, exp(), for the exponential
ex and abs() for the absolute value, are used.
setiosflags(ios::fixed) allows oating point numbers to print with trailing zeros,
nice for formatting. This only has to be done once.
setw() sets the width of the output of the following item. It has to be set every time!
setw() is dened in <iomanip>.
setprecision() set the precision of the following number. setprecision() is also
dened in <iomanip>.
The output of the above program looks like:
#
#
#
#
#
.
.
.
1000000:
2000000:
3000000:
4000000:
5000000:
I
I
I
I
I
=
=
=
=
=
0.63213700,
0.63193800,
0.63197933,
0.63180525,
0.63187300,
Delta
Delta
Delta
Delta
Delta
=
=
=
=
=
0.00001644
0.00018256
0.00014123
0.00031531
0.00024756
248
The program operates very much like the program that found by random sampling. We
nd an x and a y randomly. However, the only dierence is that we count only those points
that fall under the function as contributing to the area under the curve. The area under the
curve is the fraction of points that fall under the curve, times the area of the rectangle. In
the above example, the area of the rectangle is 1. If the rectangle did not have unit area,
how would you know what to multiply by? Thats what you have to gure out in assignment
4!
9.2
Simple Sorting: The bubble or sinking sort
This is called the bubble or sinking sort technique because small items bubble to the top
and large items sink to the bottom. The technique involves looping through an array
looking at pairs of values in order. If the upper item is bigger than the lower item, they
are switched. Then the next pair is considered. The lower item of the previous pair is
the upper item of the current pair. The iteration continues until you can go through the
entire array and no pair is switched.
For example, imagine that you have the following array {3, 2, 1} that you want sorted in
ascending order. We will use the convention that parentheses surround the pair under
consideration.
Here is how it would work:
Loop
Loop
Loop
Loop
1:
2:
3:
4:
{3, 2, 1} : {(3, 2), 1} {2, (3, 1)} : {2, 3, 1}

{2, 3, 1} : {(2, 3), 1} {2, (3, 1)} : {2, 1, 3}
{2, 1, 3} : {(2, 1), 3} {1, (2, 3)} : {1, 2, 3}
{1, 2, 3} : {(1, 2), 3} {1, (2, 3)} : {1, 2, 3}
Change, continue
Change, continue
Change, continue
No change, stop
Here is some boiler plate pseudocode that gets the job done:
bool whileThingsAreChanging = true;
while(whileThingsAreChanging)
{ whileThingsAreChanging = false; //Assume that the list is sorted
for(/* appropriate for-loop parameters */) //Careful!
{ //Loop through the list in an order fashion in one direction
//Compare adjacent pairs of things
//If the pair is out of order, switch them and set
// whileThingsAreChanging to true
}
//When the for loop completes, either nothing has changed and
//whileThingsAreChanging has remained false, or something changed and
//whileThingsAreChanging was changed to true
9.3. RECURSIONA FUNCTION CALLING ITSELF
249
}
//When the execution arrives here, the list is sorted.
9.3
RecursionA function calling itself
Sometimes in mathematics, there are very compact ways of writing rules for generating
complicated things. For example, the factorial can be written out explicitly as:
N! = N (N 1) (N 2) 3 2 1
or, it can be written more compactly as
N! = N (N 1)! N >= 0 where 0! 1
where N is an integer.
Mathematicians prefer the second expression because it is better dened (no assumptions
about what means) and it also suggests a mechanism or process for calculating N! C++
mimics this by providing the programmer a mechanism called recursionthe ability for a
function to call itself. Here is how the factorial can be calculated using this technique.
//File: recursiveFactorial.cpp
#include <iostream>
int factorial(int n)
{ if (0 == n) return 1;
else return (n * factorial(n - 1));
}
int main(void)
{ cout << "N? ";
int N;
cin >> N;
cout << N << "! = " << factorial(N) << endl;
return 0;
}
Note how compact the coding is, and how similar it is to the second mathematical denition.
Note also how the 0! is coded in the if statement. This is critical to stop the recursion process.
Try writing out the stack frames for a simple case, say 5!, to see how this works.
There are a few simple, general guidelines for using recursion:
250
Successive calls to functions should represent progressively simpler cases. (For example,
(N 1)! is simpler than N!)
There is an end to the recursion, the base case, that does not involve a recursive
call. (For example 0! = 1)
If a function calls itself more than once, and the nesting level gets deep, the calculation can be very inecient. (See the following example.)
Recursion is a great way to crash a computer or a program! If you do not program the
stopping criterion correctly, you will cause your stack to collide with your heap, and
your program will be in a heap of trouble.
To illustrate the third point above, consider the calculation of the Fibonacci numbers dened
as follows.
f (N) = f (N 1) + f (N 2) N > 0 where f (1) 1 f (2) 1
where N is an integer. The code can be written using recursive calls as follows:
//File: recursiveFibonacci.cpp
#include <iostream>
int fibonacci(int n)
{ if (1 == n || 2 == n) return 1;
else return (fibonacci(n - 1) + fibonacci(n - 2));
}
int main(void)
{ cout << "N? ";
int N;
cin >> N;
cout << "Fibonacci(" << N << ") = " << fibonacci(N) << endl;
return 0;
}
The fibonacci function calls itself twice. This is a NO, NO! unless N is quite small, like
less than 30. Dont believe it? Try calculating f (1000)!
Although the above program is elegant and compact, we sometimes have to give up clarity for
eciency, yet another compromise we must sometimes make. Here is how one can program
the Fibonacci numbers using loops.
9.3. RECURSIONA FUNCTION CALLING ITSELF
251
//File: loopFibonacci.cpp
#include <iostream>
int fibonacci(int n)
{ if (1 == n || 2 == n) return 1;
else
{ int fMinus2 = 1;
int fMinus1 = 1;
int f;
for (int i = 3; i <= n; i = i + 1)
{ f = fMinus2 + fMinus1;
fMinus2 = fMinus1;
fMinus1 = f;
}
return f;
}
}
int main(void)
{ cout << "N? ";
int N;
cin >> N;
cout << "Fibonacci(" << N << ") = " << fibonacci(N) << endl;
return 0;
}
Its ugly but its fast. Try comparing the speed of execution of the above two methods. You
will be surprised at the dierence.
Recursion relations play a big role in mathematics and consequently there are many examples
that could be coded in C++ and used as examples. However, another important use of
recursion relations relates to graphics. The following example recursively splits a line into
two smaller segments chosen randomly but with the same total length. This is a very simple
model of how a tree grows. The output of this program is the total number of branches
according to the users input of the largest branch length and the smallest branch length.
//File: tree.cpp
#include <iostream>
#include <cstdlib>
int tree(float L, float LMin)
{ if (L <= LMin) return 0;
252

else
{ float split = static_cast<float>(rand())/RAND_MAX;
return 2 + (tree(split * L, LMin) + tree((1 - split) * L, LMin));
}
}
int main(void)
{ cout << "LTotal, LMin? ";
float LTotal, LMin;
cin >> LTotal >> LMin;
cout << "Number of branches = " << tree(LTotal, LMin) << endl;
return 0;
}
There are two companion programs that will be demonstrated in class that provide graphical
output from this program. These are called treeGraphics.cpp and anotherTree.cpp,
both posted on the web. In class, the close connection to fractals will be evident from the
graphical output. Fractals are interesting not only mathematically, but have great uses in
communications, encryption and virtual cinematography!
Before closing the section on recursion, it is worthwhile to note that this technique is also
used in numerical simulation, something the author of this lecture is passionate about. In
fact, the coding in the tree programs shares a lot in common with, to cite a few examples,
scattering of light in the atmosphere, the movement of neutrons in a nuclear reactor, the
scattering of electrons and photons in the human body during diagnostic X-ray procedures
or treatment of cancer with radiotherapy beams. Just a few of the details are dierent. In
order to nd out what these details are, you will have to take a later course in numerical
simulation.
9.4
Input and Output using les
9.4.1
File and stream handling in C++
Why les?
Except for a few cases where we generated random data using quasi-random numbers,
so far we have been getting data into our programs by typing input at the keyboard
and any output that our programs have generated has gone to the computer screen.
This is really only useful for limited amounts of data, either on the input side or the
output side.
9.4. INPUT AND OUTPUT USING FILES
253
A large quantity of data is handled more easily and more reliably with les.
Moreover, the same data les can be used by more than one program.
How is data organized?
Computers only know about bits - binary digits, 0s and 1s. These are the only
fundamental data types understood by computer hardware.
Humans, on the other hand, are more comfortable with characters - and these, as we
have come to know are usually represented by 8 bits, a byte, by computers.
Characters can be combined into eldscombinations of characters that have some
unique meaning, like a oating-point number, an integer, a string of characters representing a persons name, etc
Character elds can be gathered into recordsof related elds, for example, a persons
name, social security number, age, salary, etc as many elds as are needed for a given
application.
Records can be combined into lescollections of records all somehow related, that
can be accessed by a computer program for some application.
Files are collected in directories. A single directory usually contains les with related data. Directories can also contain sub-directories with their own les and subdirectories. It is useful to think of a tree structure of directories. Each sub-directory
is another branch of the tree. The root directory is the trunk of the tree.
The disk on a computer can contain several such trees or partitions. The disk
controller, a hardware component in a computer, works with the operating system to
organizing or manage the les on the disk.
A single computer can contain several disks. Some computers, le servers, can house
many.
These disks (or portions of them) may be shared across networks with other computers either nearby (Local Area Networks [LANs]) or over long distances (Wide Area
Networks [WANs]).
The ability to connect computers and their data together makes possible the establishment of the World Wide Web [WWW], which permits access to some of the data that
resides on a computer connected to the network. This data can be made available to
anyone who knows its Uniform Resource Locator [URL]. The URL not only gives the
internet address of the data, but also conveys some information about what form the
data is in.
254
What is C++s view of les?

C++ views a le as a sequential stream of bytes. The end of a le is determined
anywhere by an end-of-le (EOF) marker, or at a specic byte count that the operating
system has been told about in a program.
END OF FILE (EOF)

Files are stored permanently by the operating system, usually on a hard-disk or some
other medium (oppies, tape, CDs, DVDs). The operating system is responsible for
the maintenance of les.
When a le is opened by C++ what it really means is that a request for data is sent to
the operating system and the data is put into a data streamand C++ only concerns
itself with handling the stream, not the hardware storage device. Once data enters
the stream its is the responsibility of C++ to interpret the data using formatting
instructions.
This allows most of C++ to be operating-system independent.
The standard header les iostream (keyboard and screen-related Input/Output [I/O])
and fstream (le-related I/O) contains all the stream-le operating system dependencies and has to be rewritten for each new operating system.
When a C++ program starts with #include <iostream> it knows about three streams:
stdin (standard input) is (usually) the keyboard which can be read using the
cin >> function.
stdout (standard output) is (usually) the screen which can be written to using
the cout << function.
stderr (standard error) is (usually) the screen which is written to using the
cerr << function. The stderr stream is used by C++ to inform the user, via
warning and error messages, that data may contain ambiguities or errors. It is
often useful to make a distinction between stdout and stderr even though they
output, by default, to the same hardware device.
255
It is possible in C++ to redirect any or all of the stdin, stdout or stderr les.
We will demonstrate how to do this only as required for some specic purpose. It is
operating-system dependent.
It is possible in C++ to dene, create and read other les. Just keep in mind that
C++ creates these as streams and lets the operating system handle all the other details
of le management.
9.4.2
Creating sequential access les and writing to them
What is a sequential access le?

A sequential access le is one in which the records do not have a xed, specied length.
One usually uses such les for writing and reading data that is not changed later and
that is read all at once.
It is the most ecient way of storing les which have variable length records.
Changing the contents of records or the numbers of records can be done, but not very
eciently.
Accessing the contents of some specic records can be done, but not very eciently.
Nonetheless, it is a useful form for storage of data with variable record length that is
all read into memory, manipulated and then all written out.
Sequential les contain character data that is human readable and transportable (usually) from machine to machine and (usually) from architecture to architecture as long
as all the machines understand ASCII character representation. The only essential
dierence is the interpretation of the end of record. UNIX uses line feed, MacOS
uses carriage return and DOS/Win uses carriage return followed by line feed.
There are 6 things to remember when writing to les
1. #include <fstream> must go in the pre-processor area.
2. Declare an output le object and name it using an identier in C++. This is done
using the ofstream keyword. For example,
ofstream myOutputFile;
256

associates the identier myOutputFile with a user-declared output le stream. You
should think of this as a declaration, much as a usual variable declaration, or string
or vector object declaration. It can go anywhere in a program. It is good practice to
locate it close to where you use it for the rst time.
3. Before writing to the user-dene output stream, it must be opened using the .open()
output le object member function. For example,
myOutputFile.open("someData.dat");
opens the output le stream object called myOutputFile. It also connects the output le identier myOutputFile with a le called someData.dat. Data going to
myOutputFile will be written physically in the le someData.dat located on the computers hard disk (usually). In this example, the le someData.dat would reside in the
same subdirectory as the load module that was used to start the program assuming
that the user started the program with a statement like:
> a.out
You can also use relative or absolute path names for the parameter of the .open()
member function. For example, a typical unix system usage would be:
myOutputFile.open("~/Private/someData.dat");
and a Dos/Win usage:
myOutputFile.open("C:\\eng101\\hw\\hw6\\hw6.dat");
Can you guess why there are double backslashes above?
Warning! If the parameter of the .open() function refers to an existing le,
that le will be destroyed unless 1) it is write-protected, or 2), you take
special care to see if it exists before opening it. (More on this later.)
4. Always check to see that the le was opened successfully using the .fail() member
function. The .fail() member function returns a true when the le open failed. For
example,
if (myOutputFile.fail())
{ //File open fail
return 1; // Return to calling function with abnormal termination code.
}
257
5. Write to the output stream. Remember it is just a stream, just like cout, but with
your own identier. For example,
int i = 1, j = 2;
myOutputFile << i << ", " << j << endl;
6. When you are nished writing to the le, close the le using .close member function.
For example,
myOutputFile.close();
The .close() member function disconnects the output stream object, in this case,
myOutputFile, with the physical le, in this case, someData.dat. Although the termination of a program will properly close all les that it opened, it is a good idea to
close les as soon as you are nished with them. This frees up some system resources
for other programs and in the event of a system crash, closed les are left intact or
may be recovered more easily. Files left unclosed during a system crash may be left
incomplete. (This is somewhat system dependent and dependent upon how sudden
the crash is. Power failures are particularly nasty in this regard.)
Here is an example called fileOutput.cpp that illustrates all these concepts, plus a few
other concepts.
//File: fileOutput.cpp
#include <iostream>
#include <fstream>
int main(void)
{ ofstream myOutputFile; // Create an output file object
cout << "What filename do you want to write to: ";
char fileName[20]; // 1D array of chars
cin >> fileName;
myOutputFile.open(fileName); //Connect the stream to the file
{ // File could not be opened
cerr << "File called " << fileName << " could not be opened.\n";
258

return 1; // Return to O/S with abnormal return code
}
else
cout << "File called " << fileName << " was successfully opened.\n";
bool keepReading = true; // Keep reading if true
do{
cout << "Two floats? ";
float x, y; // Two input floats
cin >> x >> y;;
if (cin.eof()) keepReading = false; // End of file <CNTL>-D detected
else myOutputFile << x << " " << y << endl;
}while(keepReading);
cout << "End of input detected\n";
myOutputFile.close(); //Disconnect the stream from the file
return 0;
}
This code makes use of the .eof() member function which tests to see if and end-of-le
(EOF) marker has been received in the stream that it is referring to, in this case, cin.
.eof() returns true if the EOF has been received and false otherwise. Note how this is
exploited in the above example. This is how you can control le input with an EOF sentinel.
On UNIX systems, EOF is signaled with a <RETURN><CNTL>-d, on MS systems by <CNTL>-z.
Note as well how the lename was obtained by inputting a 1D character array from the
keyboard. It would seem sensible to use a string class object to accomplish this, but many
compilers do not support the use of a string object as an argument to the .open() member
function.
9.4.3
Reading from sequential access les
Reading from sequential access les follows the same 6 steps except for the use of the
declaration ifstream rather than ofstream and the direction of the data as indicated
by the symbols << for output and >> for input! Thats all there is to remember.
Here is an example called fileInput.cpp that reads the output created by the previous
example:
//File: fileInput.cpp
#include <iostream>
259
#include <fstream>
int main(void)
{ ifstream myInputFile; // Create an input file object
cout << "What filename do you want to read from: ";
cin >> fileName;
myInputFile.open(fileName); //Connect the stream to the file
if (myInputFile.fail())
}
else
do{
float x, y; // Two floats
myInputFile >> x >> y;;
if (myInputFile.eof()) keepReading = false; // End of file detected
else cout << x << " " << y << endl;
myInputFile.close(); //Disconnect the stream from the file
return 0;
}
Opening les safely when writing

In order to protect against destroying an existing le when opening a le for writing, you
should rst open it for reading. then test to see if the le open fails. If it fails, it means that
that le does not exist and it is safe to write to that lename. If the open does not fail, it
means that the le exists. In this case you should prompt the user to see if the le should
260
be overwritten. Heres an example:

//File: fileSafeOutput.cpp
#include <iostream>
#include <fstream>
int main(void)
{ ifstream testFile; // Create an output file object
ofstream myOutputFile; // Create an output file object
cout << "What filename do you want to write to: ";
cin >> fileName;
testFile.open(fileName);
if (!testFile.fail())
{ cout << "File called " << fileName << " already exists.\n";
cout << "Overwrite (Input 0 for no, any other int to overwrite): ";
int clobberIt;
cin >> clobberIt;
testFile.close();
if (!clobberIt) return 0;
}
myOutputFile.open(fileName); //Connect the stream to the file
}
else
do{
cout << "Two floats? ";
float x, y; // Two input floats
cin >> x >> y;;
if (cin.eof()) keepReading = false; // End of file <CNTL>-D detected
else myOutputFile << x << " " << y << endl;
9.5. COMMAND LINE ARGUMENTS
261
myOutputFile.close(); //Disconnect the stream from the file
return 0;
}
9.5
Command line arguments
Most unix commands are written in C or C++. For example, if you have ever listed the
contents of a directory, you have probably used the command ls. If you just type in ls
at the unix command prompt, you get a list of les that currently reside in your current
working directory. If you type in ls *.cpp you will get a listing of all the les that end in
.cpp. If you type ls -l, you get a long listing of les with information about ownership
and permissions for each le. The command is written in C and so you should wonder, How
did the programmer do that?.
The programmer did it by using command line arguments. Heres an example code that
exploits command line arguments:
//File: commandLine.cpp
#include <iostream>
int main(int argc, char* argv[])
{ cout << "argc = " << argc << endl;
cout << "This programs name is: " << argv[0] << endl;
for (int i = 1; i < argc; i = i + 1)
{ cout << "Argument " << i << " is: " << argv[i] << endl;
}
return 0;
}
Note the new way of declaring the main routine:
{ .
262

.
.
}
When main is called, argc contains the number of tokens (think of these as words) on
the line that ran the program. For example, if you invoked the program with
a.out word anotherWord
argc would be 3.
argv[] is an array of addresses. (The * qualier on a variable means that the identier
contains one or more addresses.) These addresses contain the addresses of the starts of the
tokens of the command line. Resident at these addresses is the start of the one-dimensional
character arrays that contain the tokens. In the above example, the rst token is the name
of the executable le (usually a.out in our examples). The second token would be word and
the third and last token in this example would be anotherWord.
heres a program that can open a le based on a token given to it on the command line:
//File: fileOpen.cpp
#include <iostream>
#include <fstream>
{ ofstream myOutputFile; // Create an output file object
char fileName[20];
if (argc == 1)
{ cout << "What filename do you want to write to: ";
cin >> fileName;
myOutputFile.open(fileName);
}
else if (argc == 2)
{ myOutputFile.open(argv[1]);
}
else
{ cerr << "Usage: a.out [filename]\n";
}
9.5. COMMAND LINE ARGUMENTS
263
cerr << "File could not be opened.\n";
}
myOutputFile.close(); // Disconnect the stream from the file
return 0;
}
If only one token is detected, the user is prompted to input a lename. If two tokens are
detected, the second is used as a lename. If there are more than two tokens, the program
stops and prints an error message.
264
9.6
Problems
1. Double factorials
Write a recursive function that returns the double factorial as a function of its int
argument N.
This is how the double factorial is calculated:
N!! = N (N 2) (N 4) (2 or 1)
The algorithm stops when a reduction by 2 leaves a number that is equal to 1 or
2.
You may assume that N > 2 and you do not have to concern yourself with the
fact that there is a nite (usually 32) number of bits to represent ints.
Your function must be compatible with the main routine below.
#include <iostream>
//Function goes here
int main(void)
{ cout << "N: ";
int N;
cin >> N;
cout << N << "!! = " << doubleFactorial(N) << endl;
return 0;
}
red% a.out
N: 5
5!! = 15
2. Sorting structs
This problem uses a struct dened as follows:
struct Student {int studentNumber; float finalGrade;};
9.6. PROBLEMS
265
where studentNumber is a 9-digit student number and finalGrade represents the

nal grade for that student. Write a void function that accepts one called-by-reference
vector<struct Student>, and sorts the vector in descending order according to the
students grade.
266
Chapter 10
A Potpourri of Applications
10.1
Finding a zero using the binary chop
Lets say you were looking for a place where a function is zero in some interval. And lets also
suppose that the only thing you knew about the function was that it is monotonic (either
always increasing or decreasing) and that it crossed the axis only once.
In fact, you have seen such a function (in disguise) in Assignment 4. You were asked to nd
the maximum of the function:
x sin(x)
f (x) =
.
1+x
The derivative of this function is:
f (x) =
sin(x)
x cos(x)
.
+
2
(1 + x)
1+x
f (x) has the properties mentioned above for certain ranges of its argument. So, nding the
zero of f (x) is exactly the same as nding the position of the extremum (in this case the
maximum) of f (x). The solution you discovered to nd the maximum of f (x) involved a
laborious search over the function. The algorithm about to be described, call the binary
chop or mid-point bisection method, works much more eciently, as we shall see with only
about 40 iterations. The pseudocode for the algorithm is:
1. Start with an initial guess for an x0 and an x1 such that the function is zero for some
x0 x x1 . From the properties of the function, we know that the sign of f (x0 ) is
dierent from the sign of f (x1 ).
2. Evaluate the function at the mid-point, xm = (x0 + x1 )/2.
3. If f (xm ) < ! , where ! is the convergence criterion, xm is the estimated zero of the
function and the algorithm can stop. Otherwise,
267
268
CHAPTER 10. A POTPOURRI OF APPLICATIONS
4. If the sign of f (xm ) is the same as the sign of f (x0 ), assign x1 = xm .

5. Else if the sign of f (xm ) is the same as the sign of f (x1 ), assign x0 = xm .
6. Go back to step 2.
Here is an example of a working code.
//file findZero.cpp
#include <iostream>
#include <cstdlib>
#include <cmath>
double myF(double x)
{ static double zero = 0;
static double f = zero/zero;
static double Pi = 4*atan(1.0);
if (zero <= x && Pi >= x) f = (sin(x)/(1+x) + x*cos(x))/(1 + x);
return f;
}
int sign(double x)
{ if (0 == x) return 0;
else if (0 > x) return -1;
else if (0 < x) return 1;
else
{ cerr << "Stopping since sign can not be determined\n";
exit(EXIT_FAILURE);
}
}
int chop(double x0, double x1, double precision, double& xNew)
{ int counter = 0;
if (0 == sign(myF(x0))) // Zero at the x0 boundary
{ xNew = x0;
return counter;
}
else if (0 == sign(myF(x1))) // Zero at the x1 boundary
{ xNew = x1;
return counter;
}
while(true){ // Search for new root
counter = counter + 1;
10.1. FINDING A ZERO USING THE BINARY CHOP
269
xNew = (x0 + x1)/2;

if (precision > fabs(myF(xNew))) return counter;
if (sign(myF(xNew)) == sign(myF(x0))) x0 = xNew;
else x1 = xNew;
}
}
bool badLimits(double x0, double x1)
{ if (sign(myF(x0)) == sign(myF(x1))) return true;
return false;
}
int main(void)
{ double x0, x1, x;
do
{ cout << "Two limits? ";
cin >> x0 >> x1;
}while(badLimits(x0,x1));
double precision;
cout << "Precision? ";
cin >> precision;
cout << "\n\nFinding root for function between "
<< x0 << " and " << x1 << " to precision = "
<< precision << endl;
cout << "Number of iterations = "
<< chop(x0,x1,precision,x) << endl;
cout << "Function is near 0 at x = " << x << endl;
cout << "Value of function there = " << myF(x) << endl;
return 0;
}
Apart from the idea of the algorithm, there are a few new features in this code. One is the
output stream cerr. cerr is a dierent output stream than cout and it is intended to be the
stream to which diagnostic or error messages are sent. For example, the "int sign() function
above writes to cerr when it can not determine the sign of the function.
The other important use of cerr is when the user of a program has redirected cout to a le
via the unix redirection symbols. For example, output to cout could be captured in a le
names myOutput as follows:
red% a.out > myOutput
270
CHAPTER 10. A POTPOURRI OF APPLICATIONS
However, cerr would still print to the screen so that the user would know something had
gone wrong. If the error message had been written to cout, the user would not know until
he or she opened to le.
Another example is:
red% a.out < myInput > myOutput
which reads cin from the le called myInput rather than the keyboard. Note that the le
names are arbitrary. The names myOutput and myInput are just examples.
Chapter 11
Programming in MATLAB
11.1
A taste of MATLAB
Chapter 11: Basic info, logical operators, logic ow control, functions (read now)
Chapter 12: 2D and 3D graphics
Chapter 13: Some applications
What is MATLAB?
An interpreted computer language
The name of the program that interprets MATLAB language
Central ideaall data is represented by 2D arrays of doubles
Data types and variables:
Names of variables? Same as ANSI C++
Types of variables and declarations?
C++: Many types, MUST declare
MATLAB: 1 type (2-D arrays of doubles) only and dont declare
A MATLAB demo:
271
272
CHAPTER 11. PROGRAMMING IN MATLAB

>> i = 10;
>> disp(i)
10
>>
Statements
ANSI C++: Terminated by a semicolon
MATLAB: Terminated by and end-of-line (carriage return)
unless you put a ... at the end of the line to continue
Can also terminate with a ; which suppresses the echo of the statement
Flow controlwhile loops
ANSI C++:
i = 0;
while(i<10)
{ i = i + 1;
}
MATLAB:
i = 0;
while(i<10)
i = i + 1;
end
Flow controlif/else constructs

ANSI C++:
if (i <= 10)
{ j = 0;
}
else
{ k = 0;
}
MATLAB:
11.2. ARRAYS IN MATLAB
273
if (i <= 10)
j = 0;
else
k = 0;
end
Example of summing the integers, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10
ANSI C++:
int main(void)
{ int i = 0, s = 0;
while(i < 10)
{ i = i + 1;
s = s + i;
}
cout << s << "\n");
return 0;
}
MATLAB:
i = 0;
s = 0;
while(i < 10)
i = i + 1;
s = s + i;
end
disp(s)
Scripts and functions
Scripts (or script M-les) are MATLAB statements in a le that the MATLAB interpreter will execute
Functions (or function M-les) are used to implement functions.
Function names are derived from le names.
11.2
Arrays in Matlab
The idea: 2-D arrays of doubles are the fundamental data type
274
Creating arrays in MATLAB

zeros Creates and initializes arrays in MATLAB and lls the array with 0.0s
ones Creates and initializes arrays in MATLAB and lls the array with 1.0s
Some examples:
>> x = ones(10,1)
x =
1
1
1
1
1
1
1
1
1
1
>> whos
Name
x
Size
Bytes
10x1
80
Class
double array
Grand total is 10 elements using 80 bytes

>>
z = zeros(5,3)
z =
0
0
0
0
0
>>
whos
Name
x
z
0
0
0
0
0
0
0
0
0
0
Size
10x1
5x3
Bytes
80
120
Class
double array
double array
275

>>
The colon operator (:) gives a list of integers
Example:
>> clear % This is how to clear stored variables
>> % This is how to make a comment
>> x = 1:10
x =
1
>> whos
Name
x
Size
Bytes
1x10
80
10
Class
double array

>>
So x = 1:N creates a row with values 1, 2, 3...N
You can use two colon operators (a::b) to make a list of the form:
a, a + , a + 2, a + 3 a + N where a + N b < a + (N + 1).
>> y = 1:2:10
y =
1
3
Note that the upper limit, 10, is not reached because the next number in the
series would overrun it.
linspace creates a 1D array with uniformly spaced elements.
x = linspace(a,b,N)makes N equally spaced points between a and b inclusive.
(The statement x = a:(b-a)/(N-1):b seems to accomplish the same thing BUT
round-o errors may cause the end-point b to be excluded! DO NOT USE the x
= a:(b-a)/(N-1):b form!)
An example:
276

>> clear
>> x = linspace(0,1,5)
x =
0
0.2500
>>
0.5000
0.7500
1.0000
Literal creation:
Create an array with one row and 3 columns:
>> x = [1.0,5.2,9.7]
x =
1.0000
5.2000
9.7000
>>
Create an array with 2 rows and 3 columns:
>> y = [1.0,5.2,9.7;3.4,9.3,10.7]
y =
1.0000
3.4000
5.2000
9.3000
9.7000
10.7000
>>
Array indexing
Simple indexing
x(i) refers to the ith element of array x
There is no o-by-one nonsense!
x = ones(1,10) = x = [1,1,1,1,1,1,1,1,1,1]
x(1) is the rst element of array x
x(10) is the last element of array x
Array indices go from 1 to the number of rows/columns
>> x = [3,5;7,9]
x =
3
7
5
9

x(1,1)
x(2,1)
x(1,2)
x(2,2)
=
=
=
=
3
7
5
9
=
=
=
=
row
row
row
row
277
1,
2,
1,
2,
column
column
column
column
1
1
2
2
Slice indexing
The : symbol as an array index means the whole row or the whole column.
Example,
>> x = [1,2;3,4;5,6]
x =
1
3
5
2
4
6
>> y = x(:,2)
y =
2
4
6
>> z = x(1,:)
z =
1
Partial slices can be extracted by specifying ranges. Example,

>> u = x(1:2,1)
u =
1
3
>> v = x(2:3,:)
v =
278
3
5
4
6
Saving an array to a le
sindat is a 2 row, 1000 column array created and saved this way:
%ch11e1.m
sindat = zeros(2,1000);
sindat(1,:) = linspace(0,2*pi,1000);
sindat(2,:) = sin(sindat(1,:)); % Note: array operation!
save sindat -ascii -double
Reading an array from a le
The following reads sindat in and plots it.
%ch11e2.m
clear
load sindat
x = sindat(1,:);
y = sindat(2,:);
plot(x,y) % Simple plotting function
pause % Pauses until any keystroke is given
close % Closes the figure
plot(x,y.^2) % Note: ".^2" means element by element squaring operation
pause
close
plot(y,x)
pause
close
Summary
zerosCreate and initialize an array and ll the array with 0.0s
onesCreate and initialize an array and ll the array with 1.0s
clearClear variables from memory
whosList current variables, long form
linspaceCreate a 1D array with uniformly spaced elements
11.3. LOOPS
279
Literal creation of arrays

Simple array indexing
Slice indexing, full and partial slices
Writing/reading arrays to/from les.
11.3
Loops
11.3.1
The for loop
The general syntax of a for loop is:

for i = array
% Some loop using i where i can not be re-assigned
end
On each successive pass through the loop, i is set to the next column of array. On the rst
pass through the loop, i is array(:,1), on the second pass i is array(:,2), etc
This can seem a little confusing unless we remember that the fundamental data type is an
array of doubles.
Here are some simple examples:
sum = 0;
for i = 1:10
sum = sum + i;
end
disp(sum);
Heres another way of doing the same thing:
i = 1:10;
sum = 0;
for j = i
sum = sum + j;
end
disp(sum);
Heres yet another way of doing the same thing:
280
i = 1:10;
sum = 0;
for j = 1:length(i) % length(array) is a MATLAB function that gives the length
sum = sum + j;
end
disp(sum);
Heres a weird way of doing the same thing:
i = [1,2,3,4,5;6,7,8,9,10];
sum = 0;
for j = i % j is set to the columns of i
sum = sum + j(1) + j(2);
end
disp(sum);
Why does this work?
The number of columns of i is 5 and so, the loop is executed 5 times. But on each iteration,
y is set to a column of x, in this case a column vector with two values which are summed
with the statement sum = sum + j(1) + j(2);. Weird! But it works and illustrates the
more general form of the looping index.
Some useful functions
size gives the number of rows and the number of columns as a two-element row vector.
For example (using the i array above):
>> size(i)
ans =
2
5
If you want to capture both the number of rows and columns, use the following:
>> [nRows,nColumns]=size(i)
nRows =
2
nColumns =
5
length gives the maximum of the number of rows and the number of columns. For example
(using the i array above):
11.3. LOOPS
281
>> length(i)
ans =
5
error error(string) displays the string and causes an error exit from an M-le to the
keyboard. If the string is empty, no action is taken.
Heres an example of an integration script using length and error.
% Area under f as a MATLAB script
% Assumes that f and x already exist
if (length(x) ~= length(f))
error( x and f are different lengths. Stopping.)
end
sum = 0;
for j = 1:(length(f)-1)
sum = sum + 0.5*(f(j) + f(j + 1))*(x(j+1) - x(j));
end
disp(sum);
Heres a 2D plotting example:
N = 100;
f = zeros(1,N);
x = linspace(-5*pi,5*pi,N);
for i = 1:N
if (x(i) == 0)
f(i) = 1;
else
f(i) = sin(x(i))/x(i);
end
end
plot(x,f);
Heres a 3D plotting example (using the above f):
g = zeros(N,N);
for i = 1:N
for j = 1:N
g(i,j) = f(i)*f(j);
282

end
end
surf(x,x,g);
11.3.2
The while loop
The general syntax of a while loop is:

while expression
% Some loop that is executed when expression is true
end
Heres an example
change = 1;
while change >= 0.001
change = change/2;
end
disp(change);
Thats about all you need to know about while loops.
11.4
The if/else construct
The general syntax of an if/else if/else construct is:

if expression1
% Executed when
elseif expression2
% Executed when
elseif expression3
% Executed when
.
.
.
elseif expressionN
% Executed when
else
expression1 is true
expression1 is false and expression 2 is true
expression1&2 are false and expression 3 is true
expression1...N-1 are false and expression N is true
11.5. SOME MATLAB TERMINOLOGY
283
% Executed when expression1...N are false

end
Valid syntax requires the if and the end, zero or more elseifs and zero or one else. The
design is very much like C and therefore does not really need to be discussed further. The
use of indenting make the logical ow a little more comprehensible.
Heres an example that makes a decision based on a random number:
x = rand;
if x < 1/3
disp(x
elseif x <
disp(x
else
disp(x
end
is less than a third);

2/3
is greater than or equal to a third but less than two thirds);
is greater than or equal to a two thirds);
New function:
rand provides an array of uniformly distributed random numbers between 0 and 1.
11.5
Some Matlab terminology
scalar A scalar in MATLAB is a 1 1 array. For example:

>> whos
Name
Size
Bytes Class
s
1x1
8 double array
row vector A row vector of length M in MATLAB is a 1 M array. For example:
>> rv = ones(1,8);
>> whos
Name
Size
Bytes
rv
1x8
64
s
1x1
8
Grand total is 9 elements using
Class
double array
double array
72 bytes
column vector A column vector of length N in MATLAB is an N 1 array. For example:
284
>> cv = zeros(10,1);
>> whos
Name
Size
Bytes Class
cv
10x1
80 double array
rv
1x8
64 double array
s
1x1
8 double array
matrix An N by M matrix in MATLAB is an N M array. For example:
>> a = rand(3,4);
>> whos
Name
Size
Bytes Class
a
3x4
96 double array
cv
10x1
80 double array
rv
1x8
64 double array
s
1x1
8 double array
>> a
a =
0.7382
0.9355
0.8936
0.8132
0.1763
0.9169
0.0579
0.0099
0.4057
0.4103
0.3529
0.1389
11.6
Operators in MATLAB
11.6.1
Math operators in Matlab
Operation
Inner expression
Exponentiation
Multiplication, x y
Division, x/y, y\x
Addition, x + y
Subtraction, x y
Assignment
Symbol
(,)
^
*
/,\
+
=
Example
(1+2)/3
2^3
x*y
y/x = x\y
x + y
x - y
y = x
Precedence
Highest
Medium high
Moderate
Moderate
Medium low
Medium low
Lowest
These are all 2-D matrix operations!

If the operands are scalars (a 1 1 array) the functionality is similar to that of C++.
Here are some examples:
11.7. POINTWISE OPERATORS IN MATLAB
285
>> a = 2*ones(2)
a =
2
2
2
2
>> b = a/4
b =
0.5000
0.5000
0.5000
0.5000
>> a*b
ans =
2
2
2
2
>> a+b
ans =
2.5000
2.5000
11.7
2.5000
2.5000
Pointwise operators in MATLAB
Operation
Pointwise exponentiation
Pointwise multiplication
Pointwise division
Here are some examples:
>> x = 2*ones(2)
x =
2
2
2
2
>> y = [0,1;2,3]
y =
0
1
2
3
>> x.*y
ans =
0
2
4
6
>> x.^y
Symbol
.^
.*
./,.\
Example
x.^y
x.*y
y./x = x.\y
Expansion
x(i,j)^y(i,j)
x(i,j)*y(i,j)
x(i,j)/y(i,j)
286
ans =
1
4
>> x.\y
ans =
2
8
0
1.0000
11.7.1
Symbol
< , <=
> , >=
==
~=
& , |
~
0.5000
1.5000
Logical Operators in Matlab

Relational operation
less than, less than or equal to
greater than, greater than or equal to
equivalent to
not equivalent to
AND , OR
NOT
Note that these operators work on arrays and can return arrays. The relational operators
operating on arrays is not discussed in this course; we will be using them as if they always
operate on scalars. Therefore, in this sense, they behave exactly as the analogous operators
in C++,
There are some dierences in notation with C++ (~= vs. !=, & vs. &&, | vs. ||, ~ vs.
!) However, the biggest dierence with respect to C++ is the dierence in the order of
precedence of the operators. Note that even some textbooks are confused on this issue. Here
is the current state of the order of precedence of the MATLAB operators. Note that for
MATLAB Version 5.3 and lower, the precedence table is dierent!
Operation
Symbol
Inner expression
( )
Transpose, Exponentiation
^ .^
Unary plus and negation
+ - ~
Multiplication, division
.* .\ ./ * \ /
Binary addition, subtraction
+ Colon operator
:
Relational logical operators < <= > >= == ~=
Logical AND
&
Logical OR
|
Assignment
=
Precedence
0 (Highest)
1
2
3
4
5
6
7
8
9 (Lowest)
For more complete information, issue help precedence. This is what you get:
PRECEDENCE
Operator Precedence in MATLAB
11.7. POINTWISE OPERATORS IN MATLAB
MATLAB has the following precedence for the built-in operators when
evaluating expressions (from highest to lowest):
1. transpose (.), power (.^), complex conjugate
transpose (), matrix power (^)
2. unary plus (+), unary minus (-), logical negation (~)
3. multiplication (.*), right division (./), left
division (.\), matrix multiplication (*), matrix right
division (/), matrix left division (\)
4. addition (+), subtraction (-)
5. colon operator (:)
6. less than (<), less than or equal to (<=), greater than
(>), greater than or equal to (>=), equal to (==), not
equal to (~=)
7. logical AND (&)
8. logical OR (|)
Starting in MATLAB 6.0, the precedence of the logical AND (&)
and logical OR (|) operators now obeys the standard relationship
(AND being higher precedence than OR) and the formal rules
of boolean algebra as implemented in most other programming
languages, as well as Simulink and Stateflow.
Previously MATLAB would incorrectly treat the expression:
y = a&b | c&d
as:
y = (((a&b) |c) &d);
It now correctly treats it as:
y = (a&b) | (c&d);
287
288
The only case where the new precedence will impact the result
obtained is when | appears before & within the same expression
(without parentheses). For example:
y = 1 | x & 0;
In MATLAB 5.3 and earlier, this statement would yield 0,
being evaluated as:
y = (1 | x) & 0;
In MATLAB 6.0 and beyond, this expression yields a 1 as the result,
being evaluated as:
y = 1 | (x & 0);
We strongly recommend that you add parentheses to all expressions of
this form to avoid any potential problems with the interpretation
of your code between different versions of MATLAB.
NOTE: A feature has been provided to allow the system to produce an
error for any expression that would change behavior from the old to
the new interpretation:
feature(OrAndError, 0)
feature(OrAndError, 1)
11.8
M-les
11.8.1
Script M-les
%
%
%
%
warning for any expression that

changed behavior (default)
error for any expression that
changed behavior
MATLAB commands may be stored in a le and executed. These les are called script
M-les. Any commands that you can enter in the MATLAB command window can be
entered in a le with a lename that is arbitrary except that it should contain the extension
.m, for example myMatlabScript.m. Script M-les may be created with your favorite editor
11.8. M-FILES
289
or with MATLABs own editor.

If you type the name, say myMatlabScript in the MATLAB command window, the following
order of execution is followed:
MATLAB checks to see if myMatlabScript is a variable in the current workspace, the
environment of user-dened and pre-dened variables; if not
MATLAB checks to see if myMatlabScript is built-in function; if not
MATLAB checks to see if myMatlabScript.m is a le in the current directory; if not
MATLAB checks to see if myMatlabScript.m exists anywhere along the MATLAB
search path;
Some useful functions for use with MATLAB script M-les:
disp(variable) displays the value of the variable, e.g..
>> n = 10; disp(n)
10
>>
echo toggles echoing.
echo off turns it o (default value)
echo on turns it on and every command executed by a script M-le is written to the screen
(useful for debugging purposes)
echo switches (toggles ) between echo off and echo on.
input prompt the user for input.
>> x = input(Input a value for x: )
Input a value for x: 10
x =
10
>>
keyboard gives control to the keyboard. Return to the script M-le by typing return at
the command window prompt, (>>).
pause Pause. Continue after user presses any keyboard key.
pause(x) Pause for x seconds, then continue.
290
>> for i = 10:-1:1

disp(i)
pause(1)
end
10
9
8
7
6
5
4
3
2
1
>>
waitforbuttonpress Pause. Continue after user presses any keyboard key or mouse key.
sprintf Write formatted data to string.
S = sprintf(FORMAT,A,...)
formats the data in the variable A (and in any additional arguments), under control of the
specied FORMAT string, and returns it in the MATLAB string variable S. There is a similar
function called fprintf that writes the MATLAB string variable to a le. (Do a help
fprintf for more information.)
FORMAT is a string containing C++ language conversion specications. Conversion specications involve the character %, optional ags, optional width and precision elds, optional
subtype specier, and conversion characters d, i, o, u, x, X, f, e, E, g, G, c, and s. See
the Language Reference Guide or a C++ manual for complete details. The %g conversion
specier produces the nicest output for oating point numbers. See the example below.
The special formats \n,\r,\t,\b,\f can be used to produce linefeed, carriage return, tab,
backspace, and formfeed characters respectively. Use \\ to produce a backslash character
and %% to produce the percent character.
Heres an example of a script M-le called russianRoulette.m:
%File: russianRoulette.m
clear
echo off
prob = 1/6;
disp(With each spin you could win 1$)
11.8. M-FILES
disp(If you lose, you lose everything!)
disp(Its best to quit while youre ahead.)
disp(Press any key to start.)
disp(Good luck)
disp( )
pause
money = 0;
repeat = input(Play? (0 = no, 1 = yes) :);
while(repeat)
disp(Spin....)
pause(1)
if (rand < prob)
disp(BANG! You LOSE!)
return
else
disp(Click! You WIN!)
money = money + 1;
disp(sprintf(Your gain is $%g,money))
end
repeat = input(Repeat? (0 = no, 1 = yes) :);
end
disp(sprintf(Your gain was $%g,money))
disp(End of Game)
disp( )
Heres a typical play of the game:
>> russianRoulette
With each spin you could win $1.
If you lose, you lose everything!
Its best to quit while youre ahead.
Press any key or mouse button to start.
Good luck!
Play? (0 = no, 1 = yes) :1
Spin....
291
292
Click! You WIN!

Your gain is $1
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $2
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $3
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $4
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $5
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $6
Repeat? (0 = no,
Spin....
Click! You WIN!
Your gain is $7
Repeat? (0 = no,
Spin....
BANG! You LOSE!
11.8.2
1 = yes) :1
1 = yes) :1
1 = yes) :1
1 = yes) :1
1 = yes) :1
1 = yes) :1
1 = yes) :1
Function M-les
Compartmentalizing your code into functional subunits is...yada, yada, yada...

Its the same sermon you got in C++! However, MATLAB makes it easy to design
functions for repeated use in many applications.
The general syntax of a function call is:
[out1,out2,...,outN] = functionName(in1,in2,inM)
and the general syntax of the function denition is to have a separate le called functionName.m
that has the following form:
11.8. M-FILES
293
function [out1,out2,...,outN] = functionName(in1,in2,...inM)

%FUNCTIONNAME A brief one line description
% A more general description of functionName with a description of the inputs and
% outputs and perhaps an example. All this text must fit into contiguous comment
% lines.
%
% See also related list of functions
% After this line starts MATLAB executable functions, important examples below
.
.
.
error(This is an error) % Can use this to return control to the command
% windows if an error is detected
.
.
.
warning(This is a warning) % Can use this to issue warnings
.
.
.
return; % Control reverts to calling M-file if a return is seen
.
.
.
% Last line in functionName, control reverts to calling M-file
Some general features of functions:
functionName is an identier, identical to that of variables
Variables can only be output in the output string [out1,out2,...,outN]. There
can be zero or one output (in which case the square brackets are not needed).
There can be zero, one or more inputs. If you change the value of an input variable it
is hidden from the calling M-le. If you do not change the value of an input, MATLAB
will not make copies for the functions workspace for the sake of eciency. If you do
change the value of an input, MATLAB will make copies for the functions workspace
and the changes will be hidden from the calling M-le.
The command window and the functions have their own workspaces, places where
variables are stored independently, much like C++s stack frame handling.
The rst line of the function M-le is called the function declaration line. The function
name must agree with the le name. If it does not, MATLAB will use the le name.
294

This line also serves a similar role as the function prototype in C++, checking the compatibility of the function denition with the function call. (MATLAB is considerably
looser, however, than ANSI C++).
If the second line of the le is a comment, it is used by MATLAB to look for functions
and functionality using the lookfor MATLAB function.
Comment lines following contiguously are copied to the screen when you type
help functionName
You can have more than one function in a function M-le. However, only the rst (the
primary function) is available to other M-les and the command window. The others
are called subfunctions and cannot be used by other functions, scripts or the command
window.
Here is an example called quadRoots.m:
function [root1,root2] = quadRoots(A,B,C)
%QUADROOTS quadRoots returns the two real roots of A*x^2 + B*x + C = 0
%
% The two roots are root1 = (-B + sqrt(B^2 -4*A*C))/(2*A)
%
root2 = (-B - sqrt(B^2 -4*A*C))/(2*A)
%
% An error is detected if 4*A*C > B^2 (imaginary roots)
% A warning is issued if 4*A*C = B^2 (identical roots)
%
% The blame? BLIF <bielajew@umich.edu>
% This version 1.1, last modified November 27, 2000
% Warranty? None. Take it or leave it.
% Copyright: Regents of the University of Michigan
if 4*A*C > B^2
error(Roots would be imaginary. Get real!)
elseif 4*A*C == B^2
warning(Roots will be the same, degenerate)
end
root1 = (-B + sqrt(B^2 -4*A*C))/(2*A);
root2 = (-B - sqrt(B^2 -4*A*C))/(2*A);
return % This is optional
11.9. PROBLEMS
11.9
295
Problems
1. Check the appropriate circle for the following MATLAB statements.

MATLAB Statement
Valid Invalid
y=3^2;

if(23==x)

23=x;

for i=-123:234

while( ~(~(~(x<3))) )

a(1:10)=1;

b=[1 2 3; 4 5;6 7];

2. Consider the following snippet of code
for p = 1:10
for g = 1:10
x(g,p) = g*p;
end
end
This piece of code:
Fills the entire rst row of the array, then moves on to the next row, and so on to
ll the array.
Fills the entire rst column of the array, then moves on the next column, and so
on to ll the array.
3. For each of the following tasks, check the appropriate box to state which construct is
the most natural way of implementing that task.
Initializing the elements of an array of

known dimensions
Responding to a users selection from a
menu
Carrying out an iterative process until
some measure of convergence is reached
Dealing a hand of cards from a deck
Determining a students letter grade for
this nal from his or her numerical score
Computing the average score on this nal
using the scores of a known number of students
Counter loop Sentinel loop if/else

296
4. Consider the following small script M-le and function M-le.

x = 4;
y = [2 3;6 7];
a(1,1) = dumbFunction(x,y(2,1));
function r = dumbFunction(a,b)
r = b/a;
return;
Now, complete the following sentence by lling in the blanks.
The numerical values
and
are passed into the function
dumbFunction. The function returns the numerical value
, which
is then stored in the variable
.
List and describe the three capabilities that are needed to construct an algorithm.
6. MATLAB Function M-le design
Describe the roles of:
(a) the rst line of a MATLAB function M-le,
(b) the second line of a MATLAB function M-le (assuming it starts with a %),
(c) the third and subsequent contiguous comment lines in a MATLAB function M-le
(assuming the second line starts with a %).
How do the MATLAB lookfor and help functions interact with these lines?
7. MATLAB Function M-les
Write a MATLAB Function M-le that provides the indicated response to the lookfor
and help commands. A human typed the stu to the right of a >> prompt, the
computer typed the rest.
>> lookfor helpMe
HELPME Help! I need somebody.
>> help helpMe
HELPME Help! I need somebody.
Help! Not just anybody!
>>
11.9. PROBLEMS
297
8. True or false?
(a) The 3 statements
clear all; x = 1.0; i = &x;
are ALL valid MATLAB statements.
(b) The 3 statements
clear all; x = 1; i = i & x;
are ALL valid MATLAB statements.
(c) The set of statements
clear all; a = 2*ones(2,2); b = a/4; c = a*b; disp(c)
prints
1 1
1 1
to the screen.
(d) The set of statements (note use of the transpose operator )

clear all; disp(linspace(0,1,5))
prints
0 0.2500 0.5000 0.7500 1.0000
to the screen.
(e) The statement

a = [0,1,2; 2,3; 4,5;];
is a valid MATLAB statement.
9. Multiple choice
Circle all the correct answer(s). There may more than one correct answer but there
will be at least one choice that is correct.
(a) The following MATLAB code:
clear all; N = 5; a = zeros(N);
for i = (2:1:N)
a(i,i-1) = -1;
if i ~= N a(i,i+1) = 1; end
end
disp(a)
is supposed to print to the screen:
298

0
-1
0
0
0
1
0
-1
0
0
0
1
0
-1
0
0
0
1
0
-1
0
0
0
1
0
but it does not. The error is
There is an syntax error in the line: if i ~= N a(i,i+1) = 1; end

After a = zeros(5);, nothing is changed in the rst row.
There is an syntax error in the line: for i = (2:1:N)
There is a conceptual error in the statement: a(i,i-1) = -1;
None of the above, there is a syntax error somewhere else.
(b) The following code is supposed to sum the squares of the matrix elements of a
random 10 by 10 array and then divide the sum by the number of elements. It
does not work as intended.
clear all;
N = 10;
s = rand(N);
s2 = s*s;
for i = 1:N for j = 1:N sum = sum + s2(i,j); end; end
disp(sum/N^2)
N is not dened properly.

There is a syntax error in the line that starts with for.
sum should be initialized rst, before the for loops.
There is a syntax error in the statement disp(sum/N^2).
There is a conceptual error in the statement s2 = s*s;.
There is no error. The code works as intended.
There is a conceptual error in the statement s = rand(N);.
10. Slice indexing

The following code is written with 4 for loops.
N = 100;
a = zeros(N);
for i = 1:N
for j = 1:N
a(i,j) = 1;
end
end
11.9. PROBLEMS
for i = N/4:3*N/4
for j = N/4:3*N/4
a(i,j) = 0;
end
end
(a) Write it without the use of for or while loops.
(b) If the variable N was set to 4 instead of 100, what would the array look like?
11. Think like a computer
Write the output that the following MATLAB code produces.
clear all;
z = ones(6);
z(3:4,:) = 0;
z(:,3:4) = 0;
disp(z);
12. Think like a computer again
Consider the following script M-le.
sign = 1;
for i = 1:5
x(i) = i - 3;
y(i) = sign*i;
sign = -sign;
end
plot(x,y);
Draw the plot that would result from executing this script.
Write the output that the following MATLAB code produces.
clear all;
z = -ones(7); % Dont miss the minus sign!
z(2:3,2:3) = 1; z(2:3,5:6) = 4;
z(5:6,2:3) = 2; z(5:6,5:6) = 3;
z(4,4) = 0;
disp(z);
299
300

Consider the following script M-le.
a = [1 2];
b = [1 2 3];
c = 1;
for ai = a
for bi = b
fprintf(a = %f, b = %f\n,ai,bi)
if (bi^2 - 4*ai.*c < 0)
fprintf(No real solution\n\n)
elseif (bi^2 - 4*ai.*c == 0)
fprintf(Single root = %f\n\n,-bi/(2*ai))
else
fprintf(Root1 = %f\n,(-bi - sqrt(bi^2 - 4*ai.*c))/(2*ai))
fprintf(Root2 = %f\n\n,(-bi + sqrt(bi^2 - 4*ai.*c))/(2*ai))
end
end
end
What is the output that would result from executing this script?
15. The purpose of the script M-le below is to:
prompt the user for a number between 0 and 100
generate a row-vector of 50 random numbers between 0 and 100
invoke a function called nearest
print out the results returned from the function
The purpose of the function M-le below is to :
take in a row-vector and a scalar value
nd the entry in the row-vector that is closest to the scalar value
return the entry that was found, and its index in the array
There are four bugs in the code, at lines 5, 7, 17 and 18. Find them and x them.
% Script M-file testNearest.m
%
x = input(pick a number between 0 and 100: );
11.9. PROBLEMS
b=rand(1,50)*100;
[ind,y] = nearest(b(50),x);
fprintf(The nearest value to %f is b(%i)=%f\n,x,ind,b);
301
2
3
4
5
6
7
% Function M-file nearest.m

function [ind,y] = nearest(a,x)
oldDiff = 1000;
[rows,columns] = size(a);
for i = 1:columns
newDiff = abs(a(i) - x);
if(newDiff < oldDiff)
y = a(i);
index = i;
newDiff = oldDiff;
end
end
return
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
16. Plot like a computer

The code below produces a 2D plot. Indicate what it would look like in the box below.
clear all;
x = [0,-0.59, 0.95,-0.95, 0.59, 0];
y = [1,-0.81, 0.31, 0.31, -0.81, 1];
plot(x,y)
17. More code debugging
In the following code I want to plot f (x) = sin(x)
for 0 x 5 by making two row
x
vectors, one for x and the other for f (x). The code does not generate any MATLAB
errors and produces a plot. However, the plot does not look anything like what I
expected. Indicate what needs to be done to x it by writing a correct version of the
code in the box below. The code below has conceptual problems on lines 3 and 5. Take
care that your code does not have a divide-by-zero problem. (Hint: f (x) = 1 when
x = 0.)
302

clear all;
N = 1000;
f = zeros(1000);
x = linspace(0,5*pi,1000);
f = sin(x)/x;
plot(x,f)
%
%
%
%
%
%
Array size
Allocate memory for f
Make a linspace for x
Calculate f(x)
Plot it
Line
Line
Line
Line
Line
Line
001
002
003
004
005
006
18. Temperature conversion

Write a code that outputs a table of degrees Fahrenheit and degrees Centigrade starting
with -40 degrees Fahrenheit, ending with 140 degrees Fahrenheit in increments of 1
degree Fahrenheit. The conversion from degrees Fahrenheit to degrees Centigrade is:
C = (5/9) (F 32)
and must be coded as a separate MATLAB function. The output should look like:
-40,-40.0
-39,-39.4
-38,-38.9
.
.
.
138,58.9
139,59.4
140,60.0
Hint: The format statement that outputs one line of the table is
fprintf(%3.0f,%4.1f\n,F(i),C(i)).
19. Array manipulation
Create a complete set of MATLAB instructions that makes a 1000 by 1000 matrix
called a that is zero everywhere except for ones in the rst column and last column
and ones in the rst row and last row. There are also ones on the main diagonal of the
matrix, (a(i,i) = 1) for all i. For full credit you can not use more than one for or
while loop. Otherwise, one point will be deducted for each use of a for or a while
loop. Do it the MATLAB way! There are ve extra credit points on this problem for
those who remember and use correctly (from reading the book, or from attending the
last lecture) the command for making an array with ones on the main diagonal and
zeros everywhere else.
20. Matrix rotation and ipping
Write 4 dierent functions that:
Rotates a matrix in a clockwise fashion
11.9. PROBLEMS
303
Rotates a matrix in a counterclockwise fashion

Flips a matrix about its horizontal mid-line
Flips a matrix about its vertical mid-line
Hints and instructions.
Recall that [Rows,Columns] = size(a) can be used to get the number of rows
and columns of the matrix a
Your functions have to work on arrays of any size
Your functions do not have to provide output to the help and lookfor functions.
Heres an example:
>> a = rand(2,3)
a =
0.9218
0.1763
0.9355
0.7382
0.4057
0.9169
>> disp(rotateClockWise(a))
0.7382
0.9218
0.4057
0.1763
0.9169
0.9355
>> disp(rotateCounterClockWise(a))
0.9355
0.9169
0.1763
0.4057
0.9218
0.7382
>> disp(flipVertical(a))
0.9355
0.1763
0.9218
0.9169
0.4057
0.7382
>> disp(flipHorizontal(a))
0.7382
0.4057
0.9169
0.9218
0.1763
0.9355
21. Stars and Stripes forever
At the end of this problem description, parts of a script M-le are given. This program
may be used to draw the US ag. The US ag has 13 horizontal rows, alternating with
red and white. Row 1 is red, row 2 is white, and so on, for 7 red rows and 6 white
ones. A solid blue rectangle occupies the upper left corner, from rows 7 to 13 and it is
half the width of the ag. There are 50 white stars arranged in the blue patch.
Heres how you complete the code to nish the ag program.
Put in the code that will make the stripes and the blue patch. (Red value 0.15,
blue has value 0.4, and white has value 0.0.) Plotted as indicated it should look
like:
304
Read through the rest of the MATLAB code and understand what it is trying to
do. The bottom part is for placing the stars.
Write the function star(x,y,r). x and y are the points that dene the center of
the star, and r is its radius. Notice that the 5 points on a star are all located on a
circle centered at x and y with radius r. If you drew lines from the center to the
points of the star, those lines would be separated by 72 degrees, (72 = 360/5).
To draw the star, imagine that the top point of the star is numbered 1, the one
to its right 2, and so on. You can make a star easily by drawing a line from point
1, to 4, to 2, to 5, to 3 and back to 1. Heres an example of star(1,3,2).
You might nd it useful to to use the sin() and cos() functions. The argument
to these functions should be expressed in radians. In MATLAB, the conversion
from degrees to radians is given by: rads = pi*a/180, where a is an angle in
degrees and rads is the same angle expressed in radians.
You should use the plot() function using white solid lines. Here are the details:
PLOT
Linear plot. PLOT(X,Y) plots vector Y versus vector X.
Various line types, plot symbols and colors may be obtained with
PLOT(X,Y,S) where S is a character string made from one element
from any or all the following 3 columns:
b
blue
point
solid
11.9. PROBLEMS
305
4.5
3.5
2.5
1.5
1
1
0.5
g
r
c
0.5
green
red
cyan
1.5
o
x
+
circle
x-mark
plus
2.5
:
-.
--
dotted
dashdot
dashed
.
.
.
For example, PLOT(X,Y,c+:) plots a cyan dotted line with a plus
Your function has to respond sensibly to the lookfor and help functions.
%FLAG flag Solution to Stars and Stripes forever
clear all
%Code to make the flag without stars below
306

%Show the flag without stars using pcolor
show = flag;
show(end+1,1)
= 0; %Hidden point, fixes the lower bound of the colormap
show(end,end+1) = 1; %Hidden point, fixes the upper bound of the colormap
pcolor(1:41,1:14,show)
colormap vga
shading flat
axis off
hold on
%This code prompts the user to place the stars...
putStar = 1;
while putStar
putStar = input(Place a star? (0 = no, ~0 = yes));
if putStar
response = input(Specify [x, y, size] );
x = response(1); y = response(2); r = response(3);
star(x,y,r)
end
end
22. Turn pseudo-code into real code

Write a MATLAB code to nd the zero of a function. Your code will be made up of
three M-les:
a script M-le to drive the process
a function M-le that nds zeros using a Newton procedure (explained below)
a function M-le that contains the function for which you are nding the zero.
Pseudo-code for the three pieces is given below. Construct the MATLAB code.
Script M-le
(a) Prompt the user for an initial guess for the x value at which the zero occurs.
(b) Call the zero-nding function, with the initial guess as the input argument,
and the nal answer and number of steps it took to nd the nal answer as
the output arguments.
(c) Print out the nal answer and the number of steps it took to nd the nal
answer in some self-explanatory format.
Function M-le for Finding a Zero
11.9. PROBLEMS
307
(a) This function has one input argument, the initial guess for x.
(b) This function has two output arguments, the nal approximation of the xvalue at which the zero is found, and the number of steps it took to nd
it.
(c) Set a convergence criterion of ! = 106 .
(d) Initialize the number of iterations to zero.
(e) Carry out the following steps until the absolute value of f (x) < ! where f is
the function you are zeroing, and x is your current approximation of the zero.
Calculate an approximation to the derivative of the function at this point,
from
f (x + !) f (x !)
f (x) =
2!
Calculate a new approximation to the zero, from
x = x f (x)/f (x)
Update the number of steps that you have taken through the loop.
(f) Return to main
Function M-le f
(a) This function has one input argument, x.
(b) This function has one output argument, y.
(c) The relation is
y = x2 sin(x)/x;
23. Temperature along an iron bar
Consider an iron bar 100 cm long. One end of it is xed at 0 C while the other end is
xed at 100C. You must plot the temperature as a function of the distance between
its end-points by dividing the bar into 1000 elements and using only the following
information:
The temperature of the elements at either end is xed. That is,
TBar(1) = 0; TBar(1000) = 100;
Except for the xed endpoints, the temperature in any element is determined as
the average of its neighbors. That is,
TBar(i) = 0.5*(TBar(i-1) + TBar(i+1))
The solution is obtained by iteration within a while loop until the average temperature in the bar, sum(TBar)/1000, changes by less than 0.01 C from the average
temperature in the bar from the previous iteration.
308

Write two versions of MATLAB code that solve this problem. Both versions will employ
a while loop which computes the new average temperature in the bar.
One version will have a for loop (internal to the while loop) over the elements of the
bar. The other version may not employ any for loops, employing instead MATLABs
array indexing method: the colon (:) syntax.
As the starting condition for the while loop, assume that the bar is 0 C everywhere
except at the warmer endpoint, TBar(1000).
24. Code Creation

The MATLAB script M-le below calls a MATLAB function M-le that returns the
minimum, maximum, and average value of a 2D array. The average value is the sum of
the matrix elements divided by the number of elements in the array. In the box below,
write the function. You are not required to provide any comment lines in the function.
clear all;
Nrows = 37; Ncolumns = 93;
a = rand(Nrows,Ncolumns);
[minValue,maxValue,avgValue] = minMaxAvg(a,Nrows,Ncolumns)
25. Code Debugging and Renement
The MATLAB function M-le below called quadRoot.m was discussed in earlier in this
chapter.
%File: quadRoot.m
function [root1,root2] = quadRoot(A,B,C)
%QUADROOT returns the two real roots of A*x^2 + B*x + C = 0
%
% The two roots are root1 = (-B + sqrt(B^2 -4*A*C))/(2*A)
%
root2 = (-B - sqrt(B^2 -4*A*C))/(2*A)
%
% An error is detected if 4*A*C > B^2 (imaginary roots)
% A warning is issued if 4*A*C = B^2 (identical roots)
if 4*A*C > B^2
error(Roots would be imaginary. Get real!)
elseif 4*A*C == B^2
warning(Roots will be the same, degenerate)
end
root1 = (-B + sqrt(B^2 -4*A*C))/(2*A); % This is line 15
root2 = (-B - sqrt(B^2 -4*A*C))/(2*A); % This is line 16
11.9. PROBLEMS
309
Unfortunately, the Professor wrote an incomplete program! When the function is

employed as follows:
>> [r1,r2] = quadRoot(0,1,1)
the following error is reported by MATLAB.
Warning: Divide by zero in quadRoot.m at line 15
Warning: Divide by zero in quadRoot.m at line 16
r1 = NaN
r2 = -Inf
>>
A problem occurs when A is zero. When A is zero there should be only one solution,
x = -C/B unless B is zero as well. If both A and B are zero, there is no solution at all!
Provide a x-up to these problems by modifying the if/elseif/else construct above.
26. John Conways game of life
Write a MATLAB script M-le that plays John Conways game of life on a 50 by 50
grid. Here is how to initialize the grid:
Make a 50 by 50 matrix called grid of random numbers between 0 and 1.
Consider each cell in grid. If grid(i,j) is less than 0.4, set it to one. Otherwise,
set it to zero.
Denitions:
A cell that contains 0 is dead.
A cell that contains 1 is alive.
Heres how the game is played:
(a) Make an exact copy of grid into the array called oldGrid.
(b) Loop through every cell in oldGrid.
(c) Find the number of neighbors a cell has by summing the values of the eight
neighboring cells (left, right, up, down, and the four diagonal ones).
(d) If the cell is alive and it has less than 2 neighbors, it dies due to loneliness. (Set
grid(i,j) to zero.)
(e) If the cell is alive and it has 4 or more neighbors, it dies due to overcrowding.
(Set grid(i,j) to zero.)
(f) If the cell is dead and it has exactly 3 neighbors, it comes to life. (Set grid(i,j)
to one.)
310

(g) Any other condition, the cell stays as it was.
(h) After you have considered every cell, go back to step 1.
You do not have to write any code that produces graphical output. You just have to
code the algorithm described above.
Note that this program never terminates. Life goes on...
11.9. PROBLEMS
311
27. Slice this one up

Write MATLAB code that generates the 64 64 array that produces the following
plot.
2
60
1.8
1.6
50
1.4
40
1.2
1
30
0.8
0.6
20
0.4
10
0.2
10
20
30
40
50
60
You may use the following commands to plot the array:

plotIt = b; % b is the array that contains the data to be plotted
plotIt(65,:) = 0; % Increase the number of rows by 1
plotIt(:,65) = 0; % Increase the number of columns by 1
pcolor(1:65,1:65,plotIt); % Plot it
colormap(gray)
colorbar
Try to make a solution that uses no more than 2 for-loops.
The array contains the data 0, 1, 2 in a pattern that produces the above gure. The
64 64 array can be made up of repeated use of an 8 8 array.
312
28. Any way you slice it

Write two MATLAB codes that generate the 13 41 array that produces the following
plot. The rst version uses for-loops only. The second version contains no for-loops
but uses slice indexing of the form iStart:Stride:iEnd.
14
0.9
12
0.8
0.7
10
0.6
8
0.5
0.4
0.3
4
0.2
0.1
10
15
20
25
30
35
40
You may use the following commands to plot the array:

plotIt = a; % a is the array to be plotted
plotIt(14,:) = 0; % Increase the number of rows by 1
plotIt(:,42) = 0; % Increase the number of columns by 1
colormap(gray)
pcolor(plotIt) % Plot it
shading flat
colorbar
Some helpful hints:
(a) The vertical bars are 1 unit wide and created with a Stride of 4.
(b) Note that the rectangle is in front of one of the vertical bars and in back of two
others.
(c) The color of the vertical bars and edges are established by giving them a numerical
value of 1.0.
11.9. PROBLEMS
313
(d) The color of the background is established by giving it a numerical value of 0.0.
(e) The color of the inner rectangle is established by giving it a numerical value of
0.7.
29. Array striping
Write the MATLAB code that generates the array that produces the following plot. It
should work for any response to user inputs (see example below).
20
15
10
10
15
20
25
30
35
40
The array contains the data 0 (black stripe) or 1 (white stripe) in a pattern that
produces the above gure. The following response to the prompts produced the output
shown:
Number of stripes: 8
Number of Rows: 24
>>
30. Basket weaving
Write the MATLAB code that generates the array that produces the following basketweave plot.
The array contains the data 0 (background), 0.6 (vertical dark grey stripes), or 1
(horizontal light grey stripes) in a pattern that produces the above gure. The following
response to the prompts produced the output shown above:
314
How many times does the pattern repeat on each side? 5

>>
Hint: The solution is made a lot easier when you realize that there is a small 4 4
array that is repeated to make the full pattern. Look at grid(1:4,1:4).
31. Minor diagonals and crosses
The built-in function, eye(M,N) creates an M N matrix with 1s on the main diagonal
and 0s elsewhere. Example:
>> eye(3,4)
ans =
1
0
0
1
0
0
0
0
1
0
0
0
(a) Write a function called littleEye() that does the same thing except the 1s are
on the minor diagonal. Example:
>> littleEye(3,4)
ans =
0
0
0
11.9. PROBLEMS
0
0
315
0
1
1
0
0
0
You should do this by having littleEye() call eye() and manipulating array
indices using slice indexing.
(b) Write another function called cross() that sums eye() and littleEye(), except
that the maximum element returned can not be greater than 1. Example:
>> cross(3,3)
ans =
1
0
0
1
1
0
1
0
1
You can avoid using for loops by using the min() function. For example, if A is
a scalar and B is a matrix, C = min(A,B) returns a matrix C that has no value
greater than A.
32. A smooth operation
Consider the following MATLAB script M-le that calls the function called smooth.
NH = input(Horizontal size of plot: );
NV = input(Vertical size of plot: );
a = rand(NV,NH);
avgChng = 1;
while(avgChng > 0.001)
[a,avgChng] = smooth(a);
plotIt = a;
plotIt(NV+1,:) = 0; % Increase the number of rows by 1
plotIt(:,NH+1) = 1; % Increase the number of columns by 1
pcolor(0:NH,0:NV,plotIt)
colormap(gray)
colorbar
pause(0.01)
end
When a user types lookfor smooth the following is received:
316

>> lookfor smooth
SMOOTH smooth Smooth a 2D array
>>
When a user types help smooth the following is received:
>> help smooth
SMOOTH smooth
Smooth a 2D array
The smooth function smoothes a 2D array, its only input parameter, by

performing adjacent horizontal and vertical point averaging following the
prescription:
Assuming:
R is the number of rows in the array
C is the number of columns in the array
When 2 <= i <= R and 2 <= j<= C
b(i,j) = (a(i-1,j) + a(i+1,j) + a(i,j-1) + a(i,j+1))/4
However, when the (i,j)th point is in the first or last row or column of
the array, the algorithm averages only over the available horizontal or
vertical array points.
The smoothed array is returned to b, the first returned variable.
The average change is returned to the second return variable using the
following algorithm:
change = sum(sum(abs(b - a)))/(R*C)
>>
Write the function smooth.m. Your function has to respond to the lookfor and help
functions as described above.
33. Roots of a cubic equation
The roots (values of x such that f (x) = 0) of the cubic equation:
f (x) = Ax3 + Bx2 + Cx + D
11.9. PROBLEMS
317
where A, B, C and D are scalars, are given by the procedure below (attributed to
Francois Viete, 1615). You will code this algorithm into a MATLAB function called
cubeRoots.m where A, B, C and D are arrays:
If A = 0 the above is not a cubic equation, it is a quadratic equation. Assume
that there exists a function called quadraticRoots.m somewhere on the MATLAB
path that operates on arrays in the same way that cubicRoots.m does and will
return real roots if they exist and NaNs otherwise. Else,
Dene a, b, c as follows: a = B/A, b = C/A, c = D/A.
Dene Q = (a2 3b)/3, R = (2a3 9ab + 27c)/54.
If R2 > Q3 there are no real roots and so the 3 roots are NaNs. Else,
Dene = cos1 (R/ Q3 ). (The MATLAB acos(u) function returns an array

the same dimensions as u but with its elements being the cos1 of the elements
of u.)
The three roots are:

a/3
3

+ 2
= 2 Q cos
a/3
3

2
= 2 Q cos
a/3
3
x1 = 2 Q cos
x2
x3
where the MATLAB cos(u) and sqrt functions return an array the same dimensions as u but with its elements being the cos and of the elements of
u.)
Some notes:
Your function cubeRoots.m must accept four input arrays, A, B, C and D, all
with identical dimensions. Consequently, you must be careful with your use of
MATLAB operators.
Your function returns three arrays containing either NaNs or roots of the cubic
equation.
Your function must respond sensibly to the MATLAB lookfor and help functions.
To make a variable into a NaN, assign it as follows: root = NaN;
34. Random, Rotated, Really
Write a MATLAB code that executes the following 5-step pseudocode. Each step is
illustrated by a working example.
318

(a) Using the input() function, interrogate the user for R and C, a row size and
column size, respectively. Example:
Number of rows? 4
R =
4
Number of columns? 3
C =
3
(b) Using the input() function, generate a R C random array. Example:
a =
0.0153
0.7468
0.4451
0.9318
0.4660
0.4186
0.8462
0.5252
0.2026
0.6721
0.8381
0.0196
(c) Rotate the array counter-clockwise. The rst column becomes the last row, the
second column becomes the second-to-last row and so on. Example:
a =
0.2026
0.4660
0.0153
0.6721
0.4186
0.7468
0.8381
0.8462
0.4451
0.0196
0.5252
0.9318
(d) Set to zero any array element that has a greater column index than row index.
Example:
a =
0.2026
0.4660
0.0153
0
0.4186
0.7468
0
0
0.4451
0
0
0
(e) Change the sign of any array element that has a value less than 1/2. Example:
a =
-0.2026
-0.4660
-0.0153
0
-0.4186
0.7468
0
0
-0.4451
0
0
0
35. Can you (co)relate?

Write a MATLAB function M-le called correlate.m that does the following:
Accepts a 2-dimensional array of doubles of any size. This array is the only
variable in the input list.
11.9. PROBLEMS
319
Returns the smallest absolute dierence that exists between any two array elements.
Returns the indices of the two closest elements.
Responds sensibly to the lookfor and help functions
Note: abs(x) returns the absolute value of all the elements of x.
Heres a working example:
>> lookfor correlate
CORRELATE correlate, finds the smallest difference between two array elements
>> help correlate
CORRELATE correlate, finds the smallest difference between two array elements
Returns the difference and array indices of the two closest elements
Programmer: Alex Bielajew 12/12/00
Problem on Final Exam for Eng101/200, Fall 00
>> A = rand(3,4)
A =
0.9501
0.4860
0.4565
0.4447
0.2311
0.8913
0.0185
0.6154
0.6068
0.7621
0.8214
0.7919
>> [d,i1,j1,i2,j2] = correlate(A)
d =
0.0086
i1 =
2
j1 =
4
i2 =
3
j2 =
1
36. Minesweeper initialization
Consider the MATLAB script M-le started below:
320

%MINESWEEPER minesweeper: Initialize the minesweeper game
clear all
echo off
clc
R = input(Number of rows in the grid: );
C = input(Number of columns in the grid: );
nMines = input(Number of mines: );
if (nMines < 1)
error(Too few mines)
elseif (nMines > R*C)
error(Too many mines)
end
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% Your part goes here.
% Write your code on the following page(s).
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
disp(grid)
Heres is what you must do.
(a) Put nMines randomly on the grid by giving that grid point the value -1. For
example, grid(i,j) = -1 means that the i,jth point on the grid contains a
mine. You will need to use both the rand and int32 functions. (int32()
converts its argument to a 32-bit integer that may be used as an array index.) Be
sure that you deposit exactly nMines mines on the grid, no more, no fewer.
(b) For each grid(i,j) point that is not a mine, count the number of mines surrounding it, those that are in the range grid(i-1:i+1,j-1:j+1). Be careful
with the boundaries of the grid. Note: sum(sum(a)) sums up all the values on
the points of the array called a.
(c) Here is an example use of the program:
>> minesweeper
Number of rows in the grid: 5
Number of columns in the grid: 3
Number of mines: 5
1
-1
1
11.9. PROBLEMS
1
0
1
-1
321
2
2
4
3
2
-1
-1
-1
>>
37. This problem is a mineeld
Consider the MATLAB script M-le started below:
%LANDMINE landmine: A game of landmine
clear all
echo off
clc
R = input(Number of rows in the grid: );
C = input(Number of columns in the grid: );
pMine = input(Probability that a tile contains a mine: );
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
%
% Your part goes here.
% Write your code on the following page(s).
%
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
showGrid = grid;
showGrid(R+1,:) = -1;
showGrid(:,C+1) = 1;
pcolor(0:C,0:R,showGrid)
header = sprintf(\\fontsize{20}There were %d survivors\n,nSurvived);
title(header)
axis off
This is a brief description of what you must do.

(a) Between 1:R and 2:C-1 use the random number function rand to decide whether
or not that gridpoint contains a landmine. In other words, if rand < pMine set
the grid point to -1 to indicate the presence of a mine. Otherwise, set it to 0, to
indicate that the gridpoint is safe.
322

(b) Make the rst column of the grid equal to 1. A 1 indicates a person on the
grid. These people are trying to make it to the last column on the grid without
touching a mine.
(c) Within a while loop whose logical condition is that there are still people within
grid(:,1:C-1), use a double for loop to look at every point within grid(:,1:C-1).
If that point has value 1, then do the following:
i. Attempt to advance the person one square to the right (horizontally) and
with equal chance: one square up, one square down, or no vertical change.
Restrict the vertical movements so that the person stays within the grid. Use
of the max() and min() functions, as described in class and in an earlier exam
problem, makes this restriction quite ecient.
ii. If the proposed new location contains another person, then nothing happens.
iii. If the proposed new location contains a landmine, then both the person and
the landmine are eliminated.
iv. If the proposed new location is within the last column, then that person has
safely reached home and should not move anymore.
(d) The game continues until there are no people remaining within grid(:,1:C-1)
(e) Be careful not to move a person more than once during each execution of the
double for loop.
38. War is heck

Write a MATLAB code for the game of War that is described as follows:
The game is played on an R C array, where R and C are inputs provided by the
user.
The array elements are lled with random numbers using the rand() function.
Go through each element of the array. If the element is greater than 1/2 reset the
array element to +1, else, set it to -1. Each one of these elements is a player in the
game. Those assigned +1 are the Positives. Those assigned -1 are the Negatives.
Both the Positives and Negatives think of each other not only as Opposites but
also as Oppositiondeadly opposition.
Print out the starting number of Positives and Negatives.
The Positives and Negatives are at war and here is how the battle is fought.
An outer while loop is executed under the condition that there are players of both
kinds still in the game.
During each execution of the while loop, the following happens:
(a) Each player tries to moves randomly but with equal chance in one of the 4
directions, North(N), South(S), East(E) or West(W) by one array index.
11.9. PROBLEMS
323
(b) If a player wants to move outside the bounds of the array, then the player
disappears from the game. The penalty for cowardice is severe. (Set that
array element to 0.)
(c) If the array element that the player wants to move to is empty (0), the player
moves there. (The +1 or -1 gets written into the new array element. The old
array element is set to 0.)
(d) If the array element that the player wants to move to is a player of the same
type, then nothing happens.
(e) If the array element that the player wants to move contains an Opposite, then
they ght. Three things can happen, all with equal chance:
i. Mutual destructionboth players perish. (Both array elements get set
to 0.)
ii. The Positive player wins and the Negative player is eliminated from the
game. (The new location gets set to +1 and the old location gets set to
0.)
iii. The Negative player wins and the Positive player is eliminated from the
game. (The new location gets set to -1 and the old location gets set to
0.)
When the while loop ends, there are 3 possible outcomes. Provide a message that
indicates how the game ended and the number of remaining players.
Heres an example of running the code.
>> war
Number of rows in the game: 10
R =
10
Number of columns n the game: 12
C =
12
The game starts with 64 +s and 56 -s
The -s have annihilated the +s. There are 17 -s remaining.
>>
39. Gaussian sum Write a MATLAB function M-le, called gauss(), that is called as
follows:
s = gauss(N) % where N in some integer
Your gauss.m responds to the lookfor and help functions as follows:
324

>> lookfor gauss
GAUSS gauss The Gaussian sum: gauss(N) = 1 + 2 + 3 + ... + N
>> help gauss
GAUSS gauss The Gaussian sum: gauss(N) = 1 + 2 + 3 + ... + N
>>
which gives the algorithm. Feel free to use the trick if you remember it.
40. Fibonacci sum

Write a MATLAB function M-le, called fibonacci, that is called as follows:
s = fibonacci(N) % where N is some integer
Your fibonacci.m responds to the lookfor and help functions as follows:
>> lookfor fibonacci
FIBONACCI fibonacci Returns the fibonacci number F(N)
>> help fibonacci
FIBONACCI fibonacci Returns the fibonacci number F(N)
The Fibonacci numbers are defined recursively thus:
F(1) = 1; F(2) = 1; F(N) = F(N - 1) + F(N - 2) where N >= 3
which gives the algorithm. The error function is used to return a message when the
parameter (N in the call statement above) is less than or equal to 0. For example...
>> fibonacci(0)
??? Error using ==> fibonacci
Fibonacci numbers are defined for N >= 1
You may not use recursive function calls.
41. Array striping
Complete the MATLAB code below that generates the array that produces the following plot, in response to the following request for inputs:
Number of horizontal (Gaussian) white stripes (>0): 10
Number of vertical (Fibonacci) white stripes (>2): 8
Your code should work for any positive (> 0) response to the prompts.
The array contains the data 0 (black rectangles) or 1 (white stripes) in a pattern
that produces the above gure. The ith vertical stripe is situated at the index
fibonacci(i) (See preceding problem.) while the ith vertical stripe is situated at
the index gauss(i) (See two problems back.) You may assume (if you wish) that the
gauss() and fibonacci() functions, as described in the 2 previous problems, are on
the Matlab search path.
11.9. PROBLEMS
325
Black has value 0, White has value 1

55
50
45
40
35
30
25
20
15
10
5
0
10
12
14
16
18
20
NH = input(Number of horizontal (Gaussian) white stripes (>0): );

NV = input(Number of vertical (Fibonacci) white stripes (>2): );
%Add your MATLAB code starting here...
%The code below plots the graph if you supply
%NRows, NCols and a(1:NRows,1:NCols)
plotIt = a;
plotIt(NRows+1,:) = 0; % Increase the number of rows by 1
plotIt(:,NCols+1) = 0; % Increase the number of columns by 1
pcolor(0:NCols,0:NRows,plotIt)
title(\fontsize{20}Black has value 0, White has value 1)
colormap(gray)
42. Waldo wanders again
Write a simplied version of the Waldo program you encountered in Chapter 5. Here
are the rules:
Waldos universe is a = zeros(20)
Waldo starts o at a(10,10) = 1
326

As indicated above, the indices of the array where Waldo is has a value 1, and 0
otherwise.
Waldo is trying to make it to a(16:20,16:20). If he makes it there, the game is
over and Waldo has found his way home.
On each step, Waldo attempts to walk randomly into one of the 8 adjacent squares,
up, down, left, right and the 4 diagonal ones.
If Waldo attempts to walk out of a(1:20,1:20), Waldo is lost and the game is
over.
You do not need to supply any graphical output.
11.10. PROJECTS
11.10
327
Projects
1. Is your dorm room cold?Register your complaint!

Consider that your dorm room is represented by a 3232 array, with 4 walls xed
at 20 C, a window 161 centered on the left wall which is xed at 0 C (its freezing
outside), and a heat register comprised of 8 contiguous pixels (array elements that all
touch another heat register array element on at least one side) that is xed at some
higher temperature. See Figure 1 for an example. Your job is to:
(a) Locate the register in the room somewhere (not occupying the same space as a
wall or a window!) so that the temperature in the room is as uniform as possible,
(b) Heat the room to between 21 and 22 C by raising the temperature of the register.
(c) Full marks for algorithm design will be awarded to those who nd a location for
the heat register that keeps the temperature in the room uniform to within 2
degrees. (How to calculate this is discussed later.)
Here is a step-by-step procedure for this problem:
(a) Open a le called a7.m with any editor. This will be your script M-le. Once
in MATLAB, you will run this program by simply typing a7 at the MATLAB
prompt.
(b) To start the problem, rst ll a 3232 array with zeros:
N = 32;
T = zeros(N);
(c) Then, make the rst and last pixels of each row and column equal to 20, setting
the temperature on the walls, then the 16 pixels of the window (xed at 0) and
nally the 8 pixels of the register (set at T = 65 C) as shown in Figure 1.
The picture in Figure 1 is displayed using the following MATLAB commands:
plotT = T;
% Copy the
plotT(N+1,:) = 0; % Increase
plotT(:,N+1) = 0; % Increase
pcolor(1:N+1,1:N+1,plotT); %
colorbar;
T array
the number of rows by 1
the number of columns by 1
Plot it
Increasing the row and columns by 1 is necessary to get MATLAB to show the
walls and the window, otherwise only the interior of the room would show.
(d) Inside a while loop, the i,jth cell gets it value from its neighbors, via an averaging
relation of the form:
T(i,j) = (oldT(i+1,j) + oldT(i-1,j) + oldT(i,j+1) + oldT(i,j-1))/4;
328

where oldT is the temperature array after the previous pass through the while
loop. You have to program this in the appropriate place in the while loop and you
have to make sure that you do not violate the boundaries of the array. You can
use for loops to do this although it is much more ecient to use slice indexing.
(e) In the above averaging procedure, you may have overwritten some of the array
that was meant to be xed by the initial conditions, like the register, for instance.
So, you had better make sure that you x these up in the appropriate place before
the while loop executes another pass.
(f) The while loop is executed until the average change temperature changes less
than 0.00001. The average change is calculated from:
difference = abs(T - oldT);
change = sum(sum(difference))/(N^2);
(Do a help on abs and sum to nd out what these functions do.)
(g) If you get everything correct, the nal results look like Figure 2 which is plotted
using the commands pcolor(1:N,1:N,T); shading interp; colorbar;.
(h) Calculate and print out the average temperature and its variation:
average = sum(sum(T))/(N^2);
variation = sum(sum(abs(T - average)))/(N^2);
avg
For the above example, I got the average to be Tnal
= 21.1514 while the variation
avg
was Tnal = 5.39824. You should verify that you get this result before continuing.
While this got the average temperature to within specications, the variation is
way out of spec. You have to do better. The easiest but most boring way to do it is
to move the register around by typing in the conguration in the script M-le and
running it again until you satisfy both the average and variation specications.
(BLIFs rule #197 of computer programming: Always do it the easy, boring way
rst!) In this case, however, you will need some intuition to get the answers to
within specications because there are many possibilities, both in the location of
the register and its shape. You should have a look in your dorm or other rooms
to see if you nd any common relationships between placement of windows and
heat registers. You should also hope that the architects and engineers thought
about this carefully because the impact on heating costs and human comfort are
quite large.
Alternatively, you can try to nd a way of automating the optimization procedure.
How you do this is up to you! Be creative!
Finally, Figure 3 shows a dierent starting position with the ending results shown in
Figure 4. This is a terrible solution! Not only is the variation worse, but the register
is so hot that you would burn yourself if you got close to it!
11.10. PROJECTS
329
60
30
50
25
40
20
30
15
20
10
10
5
10
15
20
25
30
Figure 11.1: Typical starting condition for the 3232 room with a register on the right wall
at T = 65 C.
330
30
60
25
50
20
40
15
30
20
10
10
10
15
20
25
30
Figure 11.2: Final result for a 3232 size room starting with the conguration of Figure 1.
avg
avg
In this case, Tnal
= 21.1514 and Tnal
= 5.39824.
11.10. PROJECTS
331
100
30
90
80
25
70
20
60
50
15
40
30
10
20
5
10
10
15
20
25
30
Figure 11.3: Another starting condition for the 3232 room with a register in one of the
corners at T = 105C.
332
100
30
90
25
80
70
20
60
50
15
40
10
30
20
5
10
10
15
20
25
30
Figure 11.4: Final result for a 3232 size room starting with the conguration of Figure 3.
avg
avg
In this case, Tnal
= 21.1643 and Tnal
= 6.1069.
11.10. PROJECTS
333
2. The plasma arc lampDo you get the point?

If you take two electrically conducting objects and connect then to opposite poles of a
battery, nothing happensusually. However, if you touch the objects together you can
get a spark. If the voltage dierence is great enough, you do not need to touch them
together to have something interesting happen. Two conductors, if connected across a
voltage that is high enough and brought close enough together, will cause an arc to
be produced. This is lightning on a small scale!
If causing an arc is desirable for some reason, it helps to have the conductors
pointed. This is why lightning rods are pointedto attract the lightning in an
ecient manner and then divert it into the ground safely away from things that can
be damaged by it.
You can also make a brilliant light using high voltage and pointed conductors. These
devices are called plasma arc lamps or plasma arc torches. The point (pun intended) of this problem is to make a plasma arc lamp, at least a virtual one, on the
computer. The pictures that are produced are a thing of beauty, well, at least to me.
The template for this problem is found at the end of this problem description.
Here is a step-by-step procedure for solving this problem:
(a) Type the template into a le.
(b) Start MATLAB and run the template program. The rst thing that pops up is a
graphical representation of the start to the problem. It should look like:
1000
30
800
600
25
400
200
20
0
15
200
400
10
600
5
800
10
15
20
25
30
1000
Figure 11.5: Starting conditions for a 3232 voltage grid.
334

Do not proceed unless you get this.
The initial conditions for the voltage in a square grid of points (the grid size must
be a multiple of 32) is determined as described below. The statement:
V = zeros(N);
creates an N x N array to represent the voltage and sets its values to 0. The
statement:
V(1:15*N/32,1:15*N/32) = 1000;
creates the high voltage (+1000 Volts) electrode in the lower left hand corner of
the gure, while the statement:
V(17*N/32+1:N,17*N/32+1:N) = -1000;
creates the low voltage (-1000 Volts) electrode in the upper right hand corner of
the gure. The statements:
V(1,15*N/32+1:N)
V(15*N/32+1:N,1)
V(N,1:17*N/32) =
V(1:17*N/32,N) =
= linspace(1000,0,N - 15*N/32);
= linspace(1000,0,N - 15*N/32);
linspace(0,-1000,N - 15*N/32);
linspace(0,-1000,N - 15*N/32);
establish a linear change in the voltage from the electrodes to the upper left and
lower right hand corners, so that these points at the upper left and lower right are
zero. Note the use of the transpose operator in two of the above statements.
(c) If you then type a <RETURN> in the MATLAB command window, the gure on
the previous page is replaced by a shading interp version and a second window
pops up which will be empty. Now you have to start some programming.
(d) Inside the while loop, the i,jth cell gets it value from its neighbors, via a relation
of the form:
V(i,j) = (oldV(i+1,j) + oldV(i-1,j) + oldV(i,j+1) + oldV(i,j-1))/4;
where oldV is the voltage array after the previous pass through the while loop.
You have to program this in the appropriate place in the while loop and you have
to make sure that you do not violate the boundaries of the array. You can use
for loops to do this although it is much more ecient to use slice indexing as
in the determination of the initial conditions above.
(e) In the above averaging procedure, you may have overwritten some of the array
that was meant to be xed by the initial conditions. So, you had better make
sure that you x these up in the appropriate place in the while loop.
(f) The while loop is executed until the voltage array does not change by a signicant
amount. Code to test for this is provided.
11.10. PROJECTS
335
(g) After the voltage array is calculated, you calculate the magnitude of the electric
eld from it. The magnitude of the x-component of the electric eld is obtained
from:
Ex(i,j) = V(i+1,j) - V(i,j);
while the magnitude of the y-component of the electric eld is obtained from:
Ey(i,j) = V(i,j+1) - V(i,j);
The magnitude of the electric eld is obtained from a relation of the form
E(i,j) = sqrt(Ex(i,j)*Ex(i,j) + Ey(i,j)*Ey(i,j));
Be very careful of the array bounds!
(h) Then the magnitude of the electric eld is plotted as provided by the template.
(i) If you get everything correct, the nal results look like:
120
100
80
60
40
20
20
40
60
80
100
Figure 11.6: Final result for a 128128 voltage grid.
120
336
120
100
80
60
40
20
20
40
60
80
100
120
Figure 11.7: Electric eld grid corresponding to the voltage in the previous gure.
11.10. PROJECTS
337
70
68
66
64
62
60
58
56
58
60
62
64
66
68
70
72
Figure 11.8: A zoom in on the previous picture to show the details of the arc. Even on a
ne 128128 grid you can see evidence of the digitization of the grid.
338
The code template...
N = 32;
% Array size, should be divisible by 32: 32, 64, 96, 128,...
V = zeros(N); % Creates Voltage array and sets its values to 0
% Sets initial values of the V array
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
V(1:15*N/32,1:15*N/32) = 1000;
% High voltage "point"
V(17*N/32+1:N,17*N/32+1:N) = -1000; % Low voltage "point"
% Linear ramps along edges from the high voltage point
V(1,15*N/32+1:N) = linspace(1000,0,N - 15*N/32);
V(15*N/32+1:N,1) = linspace(1000,0,N - 15*N/32);
% Linear ramps along edges from the low voltage point
V(N,1:17*N/32) = linspace(0,-1000,N - 15*N/32);
V(1:17*N/32,N) = linspace(0,-1000,N - 15*N/32);
% Plot the initialized array
%%%%%%%%%%%%%%%%%%%%%%%%%%%%
% The following lets all the cells of V be printed,
% useful for making sure the array got initialized OK.
% Without this, the Nth row and column of V would not
% be displayed.
% Should NOT do it this way if using shading interp
plotV = V;
% Copy the V array
plotV(N+1,:) = 0; % Increase the number of rows by 1
plotV(:,N+1) = 0; % Increase the number of columns by 1
pcolor(1:N+1,1:N+1,plotV); % Plot it
colorbar;
pause;
% Wait for a keyboard event before continuing
tic;
% Starts the clock
oldV = V; % Save the initial value of the array in "oldV"

% Enter a loop that ends when the change in voltage is
% equal to or less so some value
11.10. PROJECTS
339
change = 1; % Sets change to 1 to get the while loop started

while(0.001 <= change)
% Take the average voltage of the 4 adjacent cells
%...need student input here...
% Resets initial values that should stay constant (see above)
% Calculates the change in voltage
difference = abs(V - oldV);
% Prints out the change in voltage
fprintf(Average change = %g\n,change)
% Gives the oldV array the newly calculated voltage
oldV = V;
end
time = toc; % Stores how long the program took to run
fprintf(Elapsed time = %g\n,time) % Prints it out
% Plots a representation of the voltage
pcolor(1:N,1:N,V);
shading interp;
colorbar;
% Calculate the magnitude of the electric
Ex = zeros(N-1,N); % Allocate memory for
Ey = zeros(N,N-1); % Allocate memory for
E = zeros(N-1,N-1); % Allocate memory for
figure(2);
pcolor(1:N-1,1:N-1,E);
shading interp;
field
Ex
Ey
E
340
3. Battleship revisited
You will write MATLAB code that will allow you to play the game Battleship! on
your computer. There is only one ship, the size of the ship is 4 units and you only
have 25 bombs. Otherwise the game rules are exactly the same as for the Battleship
program discussed in Chapter 8.
However, you must use graphical output to show the status of the game board (with
the ship hidden) and use the mouse to generate the bomb drops.
Hints:
Initializing the game grid
N = 10;
showBoard = zeros(N); % Make the board, value of zero for "open sea"
For more explanation:
>> help zeros
Drawing the game grid If you have an array called showBoard, here is how to display it:
plotBoard = showBoard; % Makes a copy of showBoard into plotBoard
plotBoard(N + 1,:) = 0; % Boosts number of rows in plotBoard by 1, assigns 0s
plotBoard(:, N + 1) = 0; % Boosts number of columns in plotBoard by 1
figure(1); % Working with figure 1
pcolor(0:N,0:N,plotBoard); % Make the plot
axis off % Turn of axis labeling (may want to leave on during debugging)

>> help figure
>> help pcolor
>> help axis
Choosing the rst point of the ship
i = 1 + N*rand;
j = 1 + N*rand;
hiddenBoard(int32(i),int32(j)) = 1; %Ship point has value 1
Note that you will have to have two boards, one that you show during the game
and one that is kept hidden, where the ship points reside.
Note the use of the int32() function. The indices of an array should always be
integral. The statements i = 1 + N*rand; j = 1 + N*rand; are intrinsically
doubles. You need the int32() function to convert from doubles to integers.
11.10. PROJECTS
341
>> help rand

>> help int32
Dropping bombs with the mouse
[x,y] = ginput(1);
jDrop = int32(x + 1);
iDrop = int32(y + 1);
The above generates i,j indices on the game board. You can test
hiddenBoard(iDrop,jDrop) to see if that array elements represents open sea
(value 0), a part of the ship (value 1), or something else. In my version of the
game, I used a value of 0.25 to represent a bomb dropped on open sea, and a 0.75
to represent a struck ship.
>> help ginput
342
4. Swarm ball
This project was inspired by my 7-year-old daughters soccer team, The Muggles. If you
have ever watched a group of 7-year-olds play soccer, it bears very little resemblance to
the game called soccer. It resembles more closely what would should be called swarm
ball.
In swarm ball, the object of the game is to surround the ball with as many of your
5 other teammates as possible, for the chances of moving the ball in the direction of
the opponents goal is proportional to the number of your team surrounding the ball.
Of course, the other team has a similar strategy. So, around the ball is a swarm of
players all kicking at the poor thing. The ball usually blurts out in one direction and
the swarm reassembles around the ball. Goaltenders, while still a part of swarm ball,
are almost completely irrelevant. The ball is always rolling on the ground. Would you
bend down to pick up a soccer ball surrounded by 24 ailing feet? I think not.
So, this projects task is to program a simulation of swarm ball with the following
requirements.
(a) A 2-dimensional array, called field, contains the game, the players, the ball and
the goalposts. The width and length of the eld are dened by:
fieldWidth = 31; % Width of the field
fieldLength = 41; % Length of the field
(b) The following values of the elements of field determine what that point occupies.
Use the following as it gives a sensible rendition when field is plotted using
pcolor:
grass = 0.5; % Grass color
ball = 0.62; % Soccer balls color
red = 0.8; % Reds team color
blue = 0.2; % Blues team color
redGoal = 0.7; % Reds goal color
blueGoal = 0.4; % Blues goal color
boundary = 0; % Field boundary
(c) Dene the eld, the boundaries and the goals as follows:
field = zeros(fieldWidth,fieldLength);
field(2:fieldWidth-1,2:fieldLength-1) = grass;
field(10:22,1) = redGoal;
field(10:22,end) = blueGoal;
(d) Dene the starting position of the game using the following indices:
ballX = 21;
ballY = 16;
11.10. PROJECTS
343
field(ballY,ballX) = ball; % The balls starting position

% Red team starting positions
redX = [17,17,17,11,11,11];
redY = [13,16,19,10,16,22];
% Blue team starting positions
blueX = [25,25,25,31,31,31];
blueY = [13,16,19,10,16,22];
and then place these points on the eld.
(e) The game proceeds in a series of steps, not to exceed 120, when time runs out.
During each step, the following happens:
i. Advance each player towards the ball in either the vertical, horizontal or
diagonal directions. A player can only move to an open point (no other
players, boundaries, goals or ball). Be sure that you do this in a fair fashion.
If you do not and move, say, all the red players rst, they would block the
blue players from the ball and the red team would have an unfair advantage.
ii. Count the number of players (redKick, blueKick) of each team surrounding
the ball. You must check all 8 points, vertical, diagonal and horizontal, in
the vicinity of the ball. The team that kicks the ball will be determined by
the following prescription:
% Pick which team gets the kick
if (redKick == 0 & blueKick == 0)
kick = 0; % Neither team kicks
elseif (rand < redKick/(redKick + blueKick))
kick = 1; % Red team kicks
else
kick = -1; % Blue team kicks
end
iii. Pick the direction that the ball will go according to the following prescription:
% Pick a direction to kick the ball
if (kick)
% Add some randomness in the y direction, move towards the goal
randKick = rand;
if (randKick <= 0.3333)
kicky = -1;
elseif (randKick <= 0.6667)
kicky = 0;
else
kicky = 1;
end
344
% Make sure kick is usually in the correct direction

kickx = 0;
if (rand < 0.75)
if (kick > 0)
kickx = 1;
elseif (kick < 0)
kickx = -1;
end
end
if (kickx == 0 & kicky == 0)
kick = false; % No kick, a total miss
end
end
iv. When the kick direction is determined, place the ball in the rst available
point along the direction. If ends up in a goal, one of the teams wins and the
game is over. If the ball hits the boundary, there is no kick. (This is simpler
to program than a throw-in from the sidelines, or a goal kick.)
(f) Use a pcolor display after each repositioning and each kick.
11.10. PROJECTS
345
5. Going, going, going...Gone!

Baseball, like most sports, is a game of inches. (Except for soccer, of course, which is
a game of feet!) However, some things in baseball happen on a grand scale, like the
distance a ball travels when hit for a home run.
There are two parts to this problem, both to be solved with MATLAB. You can solve
them using any numerical/computational technique you like so long as it is done with
MATLAB. You must use a function called homeRunDistance which is described in a
little detail below.
(a) A baseball is struck at a height of 3 feet about the ground. It can have any velocity
between 0 (Bielajew territory) and 250 feet/second (McGuire/Sosa territory). It
can have any angle with respect to the ground from 0 (parallel to the ground) to
90 degrees (straight up in the air).
Using the equations of motion described below, write a MATLAB function that
starts something like this. (It can be dierent.)
function d = homeRunDistance(y0,theta,v,G,D)
%HOMERUNDISTANCE returns distance a ball travels (hits the ground)
% homeRunDistance(y0,theta,v,G,D) calculates distance a baseball travels
%
before it hits the ground assuming that:
% y0
is the initial height of the ball (ft)
% theta is the angle in degrees of the start of the trajectory with
%
respect to the ground (90 is straight up)
% v
is the initial speed (ft/s)
% G
is the magnitude of the acceleration due to gravity (ft/s/s)
% D
is the magnitude of the drag factor (1/ft)
%
% The components of acceleration are:
%
% ax = -D*vx*v
(x-direction in units ft/s/s)
% ay = -D*vy*v - G (y-direction in units ft/s/s)
%
% where
%
% vx = v*cos(pi*theta/180) (x-direction velocity in units ft/s)
% vy = v*sin(pi*theta/180) (y-direction velocity in units ft/s)
Using a script M-le to call the function homeRunDistance, make a 3D plot of
distance versus theta and v between the limits described above. (A 20 by 20
array should be sucient.)
(b) For an initial velocity of 250 ft/s, what angle should the ball be hit for maximum
distance? You can determine the maximum in any way you please as long as it is
346

part of the same script M-le used above and it gives you approximately the right
answer! When you obtain this angle make a 2D plot of the trajectory of the ball
both with and without air resistance (D = 0). Plotting the trajectory means
making a 2D plot where the y-axis is the height and the x axis is the distance.
Without air resistance it would have a parabolic shape.
11.10.1 Supplementary material: Trajectories without air resistance

If a ball starts o at height y0 , horizontal position x0 = 0, with components of velocity
vx,0 and vy,0 in the presence of constant gravity providing an acceleration towards the
ground, ay = G, the situation is one of constant acceleration.
The trajectory as a function of time may then be described by:
x = vx,0 t
1
y = y0 + vy,0 t Gt2
2
We can solve for the time of ight, tight , it takes for a ball to return to the ground
using the quadratic equation and obtain the time and distance as follows:
tight =
vy,0 +
2
vy,0
+ 2Gy0
d = vx,0tight
For example, if we use y0 = 3 (ft), vx,0 = vy,0 = 250/ 2 (ft/s), and G = 32.174
(ft/s/s) we get d = about 1950 feet for the distance! The inputs are realistic and so
we know that air resistance must play a big role. However, knowing that there is an
exact answer in the case of no wind resistance will serve as an important test of our
stepping algorithm described later.
11.10.2
tance
Supplementary material: Trajectories with air resis-
In this case, the equations of motion are dierent because the accelerations are not
constant. In this case they are:
ax = Dvx vx2 + vy2

ay = Dvy vx2 + vy2 G
11.10. PROJECTS
347
where D is a measure of the resistance of the air. The total air resistance is proportional
to v 2 and points in the direction opposite to the instantaneous direction of the ball.
The drag factor is related to the physical characteristics of the ball (or projectile) and
has the form:
1
D = CD,ball air Aball /mball
2
where air air is the density of the air, Aball is the cross sectional area of the ball, mball
is the mass of the ball and CD,ball is the drag coecient, a unitless quantity. For
smooth spheres, CD,ball is always very close to 0.5. However, for a baseball its closer
to 0.35 because of the seams. A pitcher can also change CD,ball by adding a little extra
to it (saliva, vaseline) or taking a little away (cutting the ball). Baseballs that are
newly manufactured have a shine that is rubbed o before the start of the game by the
umpires and occasionally by the pitchers. This IS legal. All they are doing is making
the CD,ball a little bit smaller to make the pitches faster.
The D for a major-league baseball can be worked out easily form the above relations.
Use D = 0.0019285 and G = 32.174 for this problem.
11.10.3
Supplementary material: Stepping algorithms
Lets go back to the equations of motion.

ax = Dvx vx2 + vy2

ay = Dvy vx2 + vy2 G

If we assume that the acceleration is constant over a small time interval, t, then make
a stepping algorithm as follows:
(a) Start with the initial conditions x0 = 0, y0 = 3, vx,0 = v0 cos and vy,0 = v0 sin ,
t0 = 0, and a step counter i initialized to 0.
(b) Increment the step counter i = i + 1.
(c) Calculate the acceleration at the start of the step:
2
2
ax,i = Dvx,i1 vx,i1
+ vy,i1
2
2
ay,i = Dvy,i1 vx,i1
+ vy,i1
G
348

(d) Update the positions and velocities assuming constant acceleration.
1
xi = xi1 + vx,i1 t + ax,i (t)2
2
1
yi = yi1 + vy,i1 t + ay,i (t)2
2
vx,i = vx,i1 + ax,i t
vy,i = vx,i1 + ay,i t
(e) Stop if yi falls below 0, otherwise go back to step 2.
There are a few subtleties you will have to work out. One is to guarantee that the
y never becomes negative. The other is to make a reasonable choice for t. Justify
your choice. Recall that we are assuming that the acceleration is constant over the
step. It isnt. However, a proper choice for t make sure that the acceleration is
nearly constant. This weeks lectures will be riddled with hints on the solution of this
problem!
11.10. PROJECTS
349
6. This problem sucks!

Have you ever wondered how one can make a vacuum?
One can make a good vacuum, comparable to the molecular/atomic density in deep
space by employing a multi-stage process. In the rst stage would we would take, say,
a cylinder with a piston in it (much like a combustion chamber in a car) and a pipe
attached to a closed canister that you want to evacuate. With a valve in the canister
(to the cylinder) open and a valve in the cylinder (to the outside) closed, withdraw
the piston. The pressure in the canister and the pipe is reduced. Now, close the valve
to the canister and open the valve to the outside, allow the piston to descend and
repeat. One can only go so far with this method. Soon there are too few molecules in
the canister and enough molecules in the mechanism of the outside valve that one gets
very little or no gain on each stage of the pump cycle.
The second stage (the one that this problem is concerned about) is to introduce a
cold trap. A separate chamber (attached to the canister) has a cold surface in it to
which molecules adhere. This further reduces the number of molecules. A third stage
might involve causing the molecules to become ionized and then they may be swept
out with an electric eld. Making a vacuum is an interesting combination of a lot
of dierent technologiesmechanical, electrical, chemical... Lets focus on the second
stage. With MATLAB, rst construct two chambers of identical dimensions, connected
4
30
3.5
3
25
2.5
20
1.5
15
1
0.5
10
0
5
0.5
10
15
20
25
30
Figure 11.9: Two chambers and a pipe (evacuated, value = 0) surrounded by walls (solid,
value = -1).
together with a pipe. The ones I made were made up of 2D arrays and look like those
depicted in rst gure. In my example I used chambers that were 30 units high and 10
units wide. The pipe was 16 units high and 10 units wide. (A unit is the width of
350

a cell.) Thus the entire geometry of the pipe, the two chambers and enclosing walls
could be represented by a 32 32 array in MATLAB. The space within the walls is
given the value 0 for vacuum and the walls are given the value -1.
Now, ll one of the chambers with molecules, placed randomly. In the example in
second gure, I placed 100 molecules. Note that I allowed more than 1 molecule per
cell. Each cell contains a number that represents the number of molecules within the
cell. Since MATLABs random number generator seeds dierently every time, your
distributions will look dierent.
4
30
3.5
3
25
2.5
20
1.5
15
1
0.5
10
0
5
0.5
10
15
20
25
30
Figure 11.10: The left chamber lled with 100 molecules.

Now, a time-step is executed. During this time-step, each individual molecule can move
randomly in any direction by one cell, upwards, downwards, sideways, or diagonally.
If the new direction would cause the molecule to hit one of the walls, the proposed
step is invalid and a new direction should be picked randomly. If a molecule enters
the rightmost column of cells, it sticks to the cold wall and can no longer be moved.
The third gure shows the distribution of molecules after 2000 time-steps.
Finally, the simulation ends when all the molecules are trapped in the cold trap. The
nal conguration of my example is shown in the fourth gure.
Here is what you must do to complete this problem.
11.10. PROJECTS
351
6
30
5
25
4
20
15
10
10
15
20
25
30
Figure 11.11: After 2000 time steps, 73 molecules have stuck to the cold wall in the cold
trap and 27 are still free.
352
8
30
6
25
5
20
4
15
2
10
1
10
15
20
25
30
Figure 11.12: After about 8000 time-steps all the molecules have stuck to the cold wall of
the cold trap.
11.10. PROJECTS
353
You must write a MATLAB script M-le that does the following:
(a) Prompt the user for:
i. Chamber height and width
ii. Connecting pipe height and width.
The pipe height must be less than or equal to the chamber height and greater
than 0.
iii. The number of molecules to simulate.
(b) Display the initial geometry as in rst gure.
(c) Place the molecules in the left chamber and display the initial conguration as in
the second gure.
(d) Execute the time steps until all the molecules stick to the right most wall. Display
the nal conguration as in fourth gure as well as the number of time-steps it
took to do the simulation.
Some hints:
When testing your code, use geometries with just a few cells for height and width
and just a few molecules. This will allow the simulations to complete quickly and
will test the more subtle parts of the code, the wall collisions and the action of
sticking to the cold wall.
When testing your code, make lots of use of visual output. Put in some graphical output for some or all of the time-steps with a pause statement following
the graphics instruction. For example, the third gure shows some intermediate
graphical output. This is a very valuable way of debugging the operation of your
code.
You may need the following MATLAB function in your code (it is not mentioned
in the book!) to place the molecules randomly at the start. Using help uint32:
UINT32 Convert to unsigned 32-bit integer.
I = UINT32(X) converts the elements of the array X into unsigned
32-bit integers. X can be any numeric object (such as a DOUBLE). The
values of a UINT32 range from 0 to 4294967295. Values outside this
range are mapped to 0 or 4294967295. If X is already an unsigned
32-bit integer array, UINT32 has no effect.
.
.
.
See also ...UINT8, UINT16, INT8, INT16, INT32.
354
7. Solve the universe, plane and simple

Programmers are an ambitious lot, and so this problem is to simulate the evolution of
the universe, assuming that you know the position and velocity of every object in it1 .
After solving this problem, anything else thrown at you in the rest of your career will
seem like a piece of cake. For simplicity we will assume that the universe is only
3-dimensional (two for space, one for time) instead of 4-dimensional and that each
object in the universe is point-like with mass but with no spatial dimensions, and is
travelling at a speed much less than the speed of light.
There are two parts to this problem, both to be solved with MATLAB. You can
solve them using any numerical/computational technique you like so long as you do
them with MATLAB. You must write and use a function M-le called timeStep that
evolves the universe over a small increment of time DT. This function and the algorithms that one can use in it are described in some detail below. You will also write a
script M-le that will interrogate its user for the following inputs:
Number of objects in your universe (greater or equal to 2),
Mass of each object in kilograms,
Starting x and y coordinates of each object in meters,
Starting x and y velocities of each object in meters/second,
Time alloted to each timestep, DT in seconds, and
The total number of timesteps to devote to the evolution, NSteps.
Note that the magnitude of the velocity, sqrt(vx^2 + vy^2) must be much less than
the speed of light, 3.0e8 m/s, for the equations described below to work sensibly. So,
make sure that the input speed is less than 3.0e6. The user should be prompted
repeatedly for inputs until sensible inputs are provided.
The script M-le will also execute a loop over all the timesteps in the evolution, and
provide nal graphical output. Examples of the program output will be provided below.
11.10.4 Supplementary material: Trajectories of objects in

the universe
The x and y components of acceleration (ai,x , ai,y ) felt by the ith object in the universe
at position (xi , yi ) because of gravitational attraction to the other N 1 objects is:
axi =
N

i=j
Gmj (xi xj )
[(xi xj )2 + (yi yj )2 ]3/2
1
Interestingly enough, at the dawn of the 20th century, theoretical physicists actually thought that we
could do this. Then, quantum mechanics came along and things got a lot more uncertain.
11.10. PROJECTS
355
ayi =
N

i=j
Gmj (yi yj )
[(xi xj )2 + (yi yj )2 ]3/2
where G is the gravitational constant (6.673 1011 nt-m2 /kg2 ). Note that the sum
N
i=j excludes the ith object, as an object (being point-like) does not exert any gravitational force on itself.
11.10.5
Supplementary material: Stepping algorithms
If we assume that any object feels a constant acceleration over a small time interval, t,
then we may construct a stepping algorithm as described in the following pseudocode:
(a) Start with time t = 0.
At t = 0 the starting conditions:
xi (0), yi (0), vix (0), viy (0) for i = 1, 2, 3 N
are known.
(b) Calculate the acceleration at the start of the step for each particle in the system:
axi (t) =
N

i=j
ayi (t)
N

i=j
Gmj (xi (t) xj (t))

[(xi (t) xj (t))2 + (yi (t) yj (t))2 ]3/2
Gmj (yi (t) yj (t))
[(xi (t) xj (t))2 + (yi (t) yj (t))2 ]3/2
(c) Estimate the positions and velocities at the end of the timestep assuming constant
acceleration.
1
xi (t + t) = xi (t) + vx,i (t)t + ax,i(t)(t)2
2
1
yi (t + t) = yi(t) + vy,i (t)t + ay,i (t)(t)2
2
vx,i (t + t) = vx,i(t) + ax,it
vy,i (t + t) = vx,i(t) + ay,i t
(d) Increment the time at the end of the timestep, t = t + t.
(e) If the total time to evolve the universe has been reached, stop. Otherwise go back
to step 2.
There are a few subtleties you will have to think about.
356

How do you make a reasonable choice for t?
Basically, unless you are prepared to do a lot of extra math and physics to work
this out, you have to guess. Leave this up to the user. Ill be happy with an
algorithm that works as described above.
What basic inputs should I start with?
You can dream up your own universe. However, here are some astronomical facts
that you can use to set the scale of things:
G: 6.673 1011 nt-m2 /kg2

The Sun: Mass 1.989e30 kg
The Earth: Mass 5.9736e24 kg, Earth-Sun orbit radius 149,600,000,000 m
The Moon: Mass 7.35e22 kg, Moon-Earth orbit radius 384,400,000 m
Velocities? Work these out for yourself assuming that the heavier partner is
stationary and the fact that the Earth goes around the Sun once per year and
the Moon goes around the Earth once per month. If you want the data on other
planets, do a little research. I found the above numbers by surng the web.
Is my evolution realistic?
Only in the limit that t 0. You will nd that the above equations, the
longer you iterate them will increasingly violate the physical laws of the conservation of energy and momentum. There appears to be no satisfactory solution to
this problem although there are better algorithms than the one presented above.
Discussion of this would take us far beyond the scope of this course.
How do I check that my algorithm is reasonable?
Here are some suggestions:
Set G = 0 and verify that your objects travel in straight lines.
The two-object case has an exact mathematical solution. You can use this
to check your results. Almost every introductory physics book discusses this
solution.
Here is an example of the Earth and Moon executing one orbit around the Sun. In
this case, I used a timestep of 3600 seconds and 365.25 24 timesteps.
11.10. PROJECTS
357
Wobble of the Suns center of mass
x 10
yaxis (m)
xaxis (m)
10
5
x 10
Figure 11.13: This is the wobble of the center of mass of the Sun.
358
Earths orbit
11
x 10
1.5
yaxis (m)
0.5
0.5
1.5
2
2
1.5
0.5
0.5
xaxis (m)
Figure 11.14: This is the Earths orbit.
1.5
2
11
x 10
11.10. PROJECTS
359
Earth and moon together
10
x 10
7
6.5
yaxis (m)
5.5
4.5
3.5
2.5
1.2
1.25
1.3
1.35
1.4
1.45
xaxis (m)
1.5
1.55
1.6
1.65
11
x 10
Figure 11.15: This is part of the combined orbit of the Earth and the moon.
360
8. Re-Mastermind
You will write MATLAB code, a simpler version of the Mastermind program of Chapter
8, one in which you, the user, try to guess the 4 hidden numbers.
You already know the rules of the game. Only, in the case, to make things simpler, the
uncovered numbers are represented by 0s rather than ?s. (Mixing characters and
numbers in MATLAB is a bit of a pain.)
And, the game continues until you either quit, or win. Pretty simple, huh?
Hints:
How the computer nds the hidden numbers
answer = floor(9*rand(1,4) + 1);
Confused? Try:
>> help rand
>> help floor
How can you input 4 numbers at once?
guess = input(Input your guess (1 <= [a,b,c,d] <= 9), [0,0,0,0] to quit: );
The above line prompts the user for input (See next page.) and the user can type
in the elements of a row vector.
Confused? Try:
>> help input
An example play of the game is on the following page.
11.10. PROJECTS
>> MasterMind
Input your guess
uncovered =
0
0
board =
1
1
Input your guess
uncovered =
0
0
board =
1
1
2
2
Input your guess
uncovered =
0
0
board =
1
1
2
2
3
3
Input your guess
uncovered =
0
0
board =
1
1
2
2
3
3
4
4
Input your guess
uncovered =
0
5
board =
1
1
2
2
3
3
4
4
5
5
Input your guess
uncovered =
0
5
board =
1
1
2
2
3
3
4
4
5
5
6
5
Input your guess
uncovered =
7
5
board =
1
1
2
2
3
3
4
4
5
5
6
5
7
5
ans =
You win!
361
(1 <= [a,b,c,d] <= 9), [0,0,0,0] to quit: [1,1,1,1]

0
1
1
(1 <= [a,b,c,d] <= 9), [0,0,0,0] to quit: [2,2,2,2]
2
1
1
2
2
(1 <= [a,b,c,d] <= 9), [0,0,0,0] to quit: [3,3,2,3]
2
1
2
2
(1 <=
1
2
3
[a,b,c,d] <= 9), [0,0,0,0] to quit: [4,4,2,4]
1
2
2
2
(1 <=
1
2
3
4
[a,b,c,d] <= 9), [0,0,0,0] to quit: [5,5,2,5]
1
2
2
2
2
(1 <=
1
2
3
4
5
[a,b,c,d] <= 9), [0,0,0,0] to quit: [6,5,2,6]
1
2
2
2
2
2
(1 <=
1
2
3
4
5
6
[a,b,c,d] <= 9), [0,0,0,0] to quit: [7,5,2,6]
1
2
2
2
2
2
2
1
2
3
4
5
6
6
362
9. Can computers cure cancer?

No, they cant. But radiation can kill cancer and computers help. The following
project outlines the essential problems of killing cancer with radiation. But rst...
Some background
There is about a 50% chance that you will get cancer. You almost certainly know
someone in your immediate family who died of cancer. My father died from it 3 years
ago. My mother and sister have had it, and, I will most certainly get it if I live long
enough. Here are some startling facts:
In the US alone, there are 1.4 million new cancer patients each year.
Half of these will get treated with radiation. Cancer treatment using radiation is
called radiotherapy.
About 20% of those who are treated will die unnecessarily, even though cancer
was well localized. The radiation treatments were not eective enough to cure
them. If you do the math, thats 140,000 people per year in the US, about 50%
more people than live in Ann Arbor!
So, if you can make radiotherapy just 1% better, you will save 1,400 people per
year.
Thats what my research is all aboutmaking radiotherapy better, using computers
and computer algorithms. I thought I would give you a avor of what the main
problems are. And, I want to give you a MATLAB project a little more meaningful
than Mastermind!
Hints and demonstrations with be given in class.
Heres what you have to do.
(a) Make a model of a cross section of the human body 3040 cm that is represented
by a 30 row by 40 column array called body. This array will represent the energy
deposited in the body by the radiation beams.
Rows = 30; Cols = 40; body = zeros(Rows, Cols);
(b) Make a working array that pads an extra row or column on the exterior of body.
This accounts for the outside. A(2:Rows + 1,2:Cols + 1) = body;
A(1,:) = 0; A(:,1) = 0; A(Rows + 2, :) = 0; A(:,Cols + 2) = 0;
(c) The tumor is a 44 cm square in the center of the body. Use pcolor and plot
to show the body and the outline of the tumor before the radiation is present.
Heres a hint:
show = A(2:Rows + 2,2:Cols + 2); %Handles off by one quirk in pcolor
%Some code to scale show so that its maximum value is 1.
%Hint: help max
11.10. PROJECTS
363
hold on
pcolor(0:Cols,0:Rows,show)
shading flat
plot(...you figure out the arguments...)
It should look like this2 ...
30
25
20
15
10
10
15
20
25
30
35
40
Figure 11.16: Before the radiation

(d) Give the user of the program the capability of putting up to 4 beams on the
patient from the outside, one from the top, one from the bottom, one from the
left and one from the right. For example...
>> TreatCancer
Place beam top to bottom?
Place beam bottom to top?
Place beam left to right?
Place beam right to left?
which should look like this...
2
This is much prettier in color!
(0
(0
(0
(0
=
=
=
=
no,
no,
no,
no,
1
1
1
1
=
=
=
=
yes)
yes)
yes)
yes)
1
1
1
1
364
30
25
20
15
10
10
15
20
25
30
Figure 11.17: After beams are placed.
35
40
11.10. PROJECTS
365
The beams decay (or attenuate) exponentially from the surface with a decay
constant of 15 cm. So, if a surface tile has energy energy value 1, at depth i,
where i represents the depth in centimeters, the value should be exp(-i/15).
Note that the beams should cover the entire extent of the tumor.
(e) If that was the whole story, cancer would be beaten! Unfortunately, radiation
spreads, the tumor does not get a full dose and the healthy tissue can get damaged.
A simple model of this spreading is to repeatedly average the energy value as
follows:
Ai,j = (Ai1,j + Ai+1,j + Ai,j1 + Ai,j+1 )/4
Perform this averaging 5 times (perform the operation on the array and then
repeat the process 4 more times) to produce a result that looks like the following
gure.
30
25
20
15
10
10
15
20
25
30
35
40
Figure 11.18: What happens after the beams spread.
Open ended questions. (Optional.) Solve these and collect numerous Nobel prizes for
medicine and physics!
Model a real 3-dimensional person rather than the simple geometry above.
366

Can you optimize the treatment by using more beams? (The answer is yes. The
diculty is in nding the optimum conguration.)
What is the best decay constant to use? (This is related to the energy of the
beam.)
Develop the radiation physics to make the above calculation resemble reality.
Design a better radiation machine that will deliver better beams for radiotherapy.
Get this calculation done in 5 minutes or less!
Find out the real reason why radiation kills cells.
Find out what some radiation is good for you (far in excess of natural radiation
exposure). This is called radiation hormesis.
What role does radiation damage play in the development of cancers. This is
called oncogenesis.
Chapter 12
Graphics
12.1
Two Dimensional Plots
12.1.1
A basic plot
15
% Note the use of pointwise operators

10
x = linspace(-pi,pi,1000);
y = (x.^2).*sin(20*x) + x;
plot(x,y)
pause % Use a "pause" statement to stop
close % Closes the figure
10
15
4
12.1.2
%
%
%
%
%
%
%
%
%
Using dierent line colors
Choose from:
b: blue
g: green
r: red
c: cyan
m: magenta
y: yellow
k: black
w: white
367
368
CHAPTER 12. GRAPHICS

15
plot(x,y,b) % blue
10
10
15
4
15
plot(x,y,g) % green
10
10
15
4
15
plot(x,y,r) % red
10
10
15
4
15
plot(x,y,c) % cyan
10
10
15
4
12.1. TWO DIMENSIONAL PLOTS
369
15
plot(x,y,m) % magenta
10
10
15
4
15
plot(x,y,y) % yellow
10
10
15
4
15
plot(x,y,k) % black
10
10
15
4
15
plot(x,y,w) % white
% White!!!!!
10
10
15
4
370
12.1.3
Making titles
Plotting white on white is dumb!
15
% Titling your graph

plot(x,y,w) % white
title(Plotting white on white is dumb!)
10
10
15
4
12.1.4
Axis labels
Same boring plot!
15
10
yaxis
% Axis labels
plot(x,y,k)
xlabel(This is the x-axis);
ylabel(\fontsize{16}y-axis)
title(...
\fontsize{20}\it Same boring plot!)
10
15
4
12.1.5
Dierent line styles
0
This is the xaxis
The bell shaped curve

1
0.9
% A different 2D example
0.8
0.7
u = linspace(-3, 3, 1000);
v = exp(-u.^2);
plot(u,v)
title(The bell shaped curve)
0.6
0.5
0.4
0.3
0.2
0.1
0
3

%
%
%
%
%
%
%
%
%
371
Plotting with different line styles

Choose from:
solid line (default)
:
dotted line
-. dash-dot line
-- dashed line
syntax plot(u,v,:)
Note use of different colors and line styles
1
plot(u,v,k:)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
3
plot(u,v,g-.)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
3
plot(u,v,r--)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
372

1
plot(u,v,b-)
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
3
The log(x) function

5
% Example - The LogGamma function

4
s = linspace(0.01, 6, 20);
t = log(gamma(s));
plot(s,t)
title(...
\fontsize{20}The log\Gamma(x) function)
xlabel(\fontsize{20}x)
ylabel(\fontsize{20}y)
12.1.6
%
%
%
%
%
%
%
%
%
%
%
Point plots, plotting with point symbols
Plotting with point symbols

Choose from:
.
point
o
circle
x
cross
+
plus sign
*
asterisk
S
square
d
diamond
V
triangle down
^
triangle up
<
triangle left
>
triangle right
p
pentagram
h
hexagram
syntax plot(s,t,:)
373
% Note use of different colors and line styles

5
plot(s,t,k:.)
4
plot(s,t,g-.o)
4
1
5
plot(s,t,r--x)
4
374

5
plot(s,t,b-+)
4
plot(s,t,k:*)
4
plot(s,t,g-.S)
4
375
5
plot(s,t,r--d)
4
plot(s,t,b-V)
4
plot(s,t,k:^)
4
376
plot(s,t,g>)
4
plot(s,t,r<)
4
plot(s,t,bp)
4
377
5
plot(s,t,kh)
4
12.1.7
Plotting more than one thing at once
% More arguments to the plot function

% Changing axes limits
clear all
x = linspace(-1, 1, 30);
f0 = ones(1,length(x));
f1 = x;
f2 = (1/2)*(3*x.^2 - 1);
f3 = (1/2)*(5*x.^3 - 3*x);
plot(x,f0,k-,x,f1,k:,x,f2,k-.*,x,f3,k--o)
title(\fontsize{20}Plotting more than one thing on the same plot)
xlabel(\fontsize{20}x)
Plotting more than one thing on the same plot
ylabel(\fontsize{20}y)
axis([-1,1,-1,1.1]) %[xmin,xmax,ymin,ymax]
grid on % Turn the grid on
legend(P0,P1,P2,P3)
P0
P1
P2
P3
0.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
378

P0
P1
P2
P3
1
0.8
% Another way
hold on % Dont erase when drawing new plot
plot(x,f0,k-)
plot(x,f1,k:)
plot(x,f2,k-.*)
plot(x,f3,k--o)
legend(P0,P1,P2,P3)
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
P0
P1
P2
P3
1
0.8
% Another way - matrix form

g(1,1:length(x)) = f0;
plot(x,g)
legend(P0,P1,P2,P3)
12.1.8
0.6
0.4
0.2
0
0.2
0.4
0.6
0.8
1
1
0.8
0.6
0.4
0.2
0.2
0.4
0.6
0.8
Subplots
% Yet another way, subplots

subplot(2,2,1), plot(x,f0,k-),
title(Plot
subplot(2,2,2), plot(x,f1,k:),
title(Plot
subplot(2,2,3), plot(x,f2,k-.*), title(Plot
subplot(2,2,4), plot(x,f3,k--o), title(Plot
zoom on
2
1)
2)
3)
4)
Plot 1
Plot 2
1
1.5
0.5
0.5
0.5
0
1
0.5
0.5
1
1
0.5
Plot 3
0.5
0.5
Plot 4
0.5
0.5
0
0
0.5
0.5
1
0.5
0.5
1
1
0.5
12.2. THREE DIMENSIONAL GRAPHICS
379
12.2
Three dimensional graphics
12.2.1
Plot3 plots
35
30
% Example using plot3

clear all
t = linspace(0, 10*pi, 1000);
plot3(sin(t),cos(t),t)
25
20
15
10
5
0
1
0.5
1
0.5
0
0
0.5
0.5
1
12.2.2
Mesh plots
1
% Mesh plots
0.5
[x,y,z] = sphere(30);
mesh(x,y,z)
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
grid off % Turn off the grid

0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
380

1
grid on
% Turn on the grid

0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
box on % Turn on the box

0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
box off % Turn off the box

0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
12.2.3
381
Axis labelling
% Provide a label for the x-axis

xlabel(\fontsize{20}x-axis)
0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
xaxis
% Provide a label for the y-axis

ylabel(\fontsize{20}y-axis)
0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
0.5
1
yaxis
xaxis
This is a mesh plot
% Provide a title
title(\fontsize{20}This is a mesh plot)
0.5
0.5
1
1
0.5
1
0.5
0
0
0.5
yaxis
0.5
1
xaxis
382
12.2.4
meshc and meshz plots
% Two other forms of mesh plots

% meshc and meshz
[x,y,z] = peaks(30);
subplot(1,2,1)
meshc(x,y,z),title(\fontsize{20}meshc)
subplot(1,2,2)
meshz(x,y,z),title(\fontsize{20}meshz)
meshc
meshz
10
10
8
6
4
2
0
0
2
5
4
6
10
4
2
4
2
0
2
0
2
2
4
2
4
Example using surf
10
8
6
12.2.5
Surface plots using surf
4
2
0
2
surf(x,y,z)
title(\fontsize{20}Example using surf)
4
6
8
3
2
2
0
1
0
2
3
Using shading flat
shading flat % Flat shading

title(\fontsize{20}Using shading flat)
10
8
6
4
2
0
2
4
6
8
3
2
2
0
1
0
2
3

shading interp % Interpolated shading
title(...
\fontsize{20}Using shading interp)
383
Using shading interp
10
8
6
4
2
0
2
4
6
8
3
2
2
0
1
0
2
3
12.2.6
Surface plots using surfc
surfc(x,y,z) % surf with contours

title(\fontsize{20}Example using surfc)
Example using surfc
10
10
3
2
2
0
1
0
2
3
12.2.7
Surface plots using sur
surfl(x,y,z) % surf with lighting

shading interp
colormap gray
title(\fontsize{20}Example using surfl)
Example using surfl
10
8
6
4
2
0
2
4
6
8
3
2
2
0
1
0
2
3
384
12.2.8
Contour plots using contour
Example using contour

3
contour(x,y,z,20)
title(\fontsize{20}Example using contour)
3
3
With shading interp

3
6
2
colorbar % Show the colorbar

title(\fontsize{20}With shading interp)
4
1
2
0
0
1
2
2
4
3
3
12.2.9
Contour plots using contour3
contour3(x,y,z,20)
title(\fontsize{20}Example using contour3)
grid off
Example using contour3
8
6
4
2
0
2
4
6
3
2
2
0
1
0
2
3
12.2.10
Contour plots using pcolor
Example using pcolor

3
6
2
% Example using pcolor

pcolor(x,y,z)
title(\fontsize{20}Example using pcolor)
4
1
2
0
0
1
2
6
3
3
with shading interp
385
6
2
shading interp % Interpolated shading

title(\fontsize{20}with shading interp)
4
1
2
0
0
1
2
6
3
3
12.2.11
Contour plots using contourf
Example using contourf

3
6
% Example using filled contour

contourf(x,y,z,12)
title(\fontsize{20}Example using contourf)
1
2
6
3
3
Example using contour w/labels
12.2.12
Contour plots with labels
3
1.35
4.69
5.8
2.46
6.91
% Example using contour with labels

C = contour(x,y,z,12);
clabel(C)
title(...
\fontsize{20}Example using contour with labels)
3.57
1
1.98
0.24
1.35
2.46
0.872
0.24
3.57
2.46
1.98
5.32
2
3
3
4.21
3.1
0.872
386
Chapter 13
Miscellaneous topics
13.1
Pitfall review
13.1.1
Review of for loops
This module gives some review on MATLABs implementation of for loops mostly by way
of example.
The general syntax of a for loop is
for k = vectorOrColumnList
% MATLAB statements
end
There are several ways of stepping through the elements of a row vector.
Example (Need vector value but not index):
The most common usage is C++-like syntax
%File: fex1.m
% For loop requiring vector value but not the index
for k = 1:3
fprintf(k = %g\n,k)
end
% Prints
k = 1
k = 2
k = 3
387
388
or
%File: fex1a.m
for k = 1:2:8
fprintf(k = %g\n,k)
end
% Prints
k = 1
k = 3
k = 5
k = 7
or
%File: fex1b.m
for x = 1.0:-0.1:0.0
fprintf(x = %g\n,x)
end
% Prints
x = 1
x = 0.9
x = 0.8
x = 0.7
x = 0.6
x = 0.5
x = 0.4
x = 0.3
x = 0.2
x = 0.1
x = 0
or
%File: fex1c.m
theta = linspace(0,90,3);
for thetai = theta
fprintf(theta = %g\n,thetai)
end
% Prints
13.1. PITFALL REVIEW

theta = 0
theta = 45
theta = 90
Example (Need vector value and index):
%File: fex2.m
% For loop requiring vector value and the index
v = linspace(0,250,3);
for i = 1:length(v)
fprintf(velocity(%g) = %g\n,i,v(i))
end
% Prints
velocity(1) = 0
velocity(2) = 125
velocity(3) = 250
Example (Nested for loops):

%File: fex3.m
% Nested for loops
theta = linspace(0,90,2);
v = linspace(0,250,2);
for thetai = theta
fprintf(theta = %g\n,thetai)
for i = 1:length(v)
fprintf(velocity(%g) = %g\n,i,v(i))
end
end
% Prints
theta = 0
velocity(1) = 0
velocity(2) = 250
theta = 90
velocity(1) = 0
velocity(2) = 250
Example (when the index is a column vector):
%File: fex4.m
% For loop where the index is a column vector
389
390
a = [1 2 3
4 5 6];
[rows,columns] = size(a);
fprintf(a is a %g x %g matrix\n,rows,columns)
for b = a
disp(b);
disp(-------)
end
% Prints
a is a 2 x 3 matrix
1
4
------2
5
------3
6
------Note in this example that b is a new column vector on each iteration of the loop.
13.1.2
Review of functions
Some detail is given in Chapter 13 of Austin and Chancogne. However, the denitive reference for this course is given in the Chapter 11 that was handed out and is on the web.
This lecture gives some review on MATLABs implementation of functions mostly by way
of example.
Recall:
The general syntax of the function denition is to have a separate le called functionName.m
that has the following form:
function [out1,out2,...,outN] = functionName(in1,in2,...inM)
%FUNCTIONNAME A brief one line description (optional)
% .
% .
% More description (optional)
% .
% .
% .
391
% First executable statement follows this line

.
.
.
% Valid executable MATLAB statements and comments
.
.
.
% Last line in functionName, control reverts to calling M-file or command window
The function call is made with the following statement:
[out1,out2,...,outN] = functionName(in1,in2,...inM)
Functions with no input or output arguments
Heres a function called noEvil.m that does nothing except printing something silly:
function noEvil
%NOEVIL noEvil is a function that does nothing
%
% ENG101 class example
%File: noEvil.m
disp(See no evil)
disp(Hear no evil)
disp(Do no evil)
disp(A function that does absolutely nothing)
Functions with one output argument
Heres a function called square.m, that computes the square of its argument:
function a2 = square(a)
%SQUARE square is a function that squares an argument
%
% It can accept scalars, vectors or matrices
%
%File: square.m
a2 = a.^2;
392
and some examples of its use on a scalar, a vector, a matrix and the element of a matrix:
>> square(2)
ans = 4
>> y = linspace(0,10,11)
y
= 0
1
2
3
>> square(y)
ans = 0
1
4
9
>> x = round(10*rand(4))
x =
5
10
3
2
7
4
6
4
9
3
7
7
>> square(x)
ans = 25
100
4
49
36
16
9
49
9
16
81
49
10
16
25
36
49
64
81
100
2
8
6
1
4
64
36
1
>> x(3,3) = square(x(3,3))

x =
5
10
3
2
2
7
4
8
6
4
81
6
3
7
7
1
>> z = square(x(2,2))
z
= 49
Functions with more than one output argument
Heres a function called squareCube.m, that computes the square and cube of its argument:
function [a2,a3] = squareCube(a)
%SQUARECUBE squareCube is a function that squares and cubes its argument
%
%
%
393
%File: squareCube.m
a2 = a.^2;
a3 = a.^3;
and some examples of its use on a scalar, a vector and a matrix:
>> squareCube(2)
ans = 4
>> [u,v] = squareCube(2)
u = 4
v = 8
>> y = linspace(0,10,11)
y
= 0
1
2
3
>> squareCube(y)
ans = 0
1
4
9
>> [u,v] = squareCube(y)
u =
0
1
4
9
v =
0
1
8
27
>> x = round(10*rand(4))
x =
10
9
8
2
8
4
6
5
6
5
0
8
>> squareCube(x)
ans =
100
81
64
4
64
16
36
25
36
25
0
64
>> [u,v] = squareCube(x)
u =
100
81
64
4
64
16
36
25
36
25
0
64
v =
1000
729
512
8
512
64
216
125
216
125
0
512
10
16
25
36
49
64
81
100
16
64
25
125
36
216
49
343
64
512
81
729
100
1000
9
7
2
4
81
49
4
16
81
49
4
16
729
343
8
64
>> [a,b] = squareCube(x(2,2))
394
a = 64
b = 512
Note that if you want to employ all the outputs, you have to assign them, else only the
rst output is displayed. These functions work on scalars, vectors, matrices and elements of
matrices.
Subordinate functions
Lets rewrite squareCube.m in the following way:
function [a2,a3] = squareCallsCube(a)
%SQUARECALLSCUBE squareCallsCube is a function that .^2s and .^3s its argument
%
% squareCallsCube calls a subordinate function called cube.m
%
%File: squareCallsCube.m
a2 = a.^2;
a3 = cube(a);
return
function c = cube(b)
c = b.^3;
squareCallsCube.m works the same way as if you had called squareCube.m. However, the
function cube.m is only known to the function squareCallsCube.m.
>> cube(a)
??? Undefined function or variable cube.
>> help cube
cube.m not found.
>>
13.2. THE MATLAB WAY
13.2
395
The MATLAB way
One of the most important things to remember is that MATLAB is organized around the
variable type of two-dimensional arrays of doubles. Hence, MATLABs intrinsic functions
are optimized to use this variable type. If you are making use of this aspect of MATLAB,
you are doing things the MATLAB way.
If you are not (and could be), then your application will almost certainly run slower.
For example, lets consider a temperature diusion problem. (Examples of a heat diusion calculation in 1 dimension is given in the example codes ironbarPointJacobi.m and
ironbarGaussSeidel.m. A 2-dimensional example may be found in the example codes
heat.m, heatSlow.m, and heatFast.m) The part of the algorithm that takes all the time is:
for i = 2:N-1
for j = 2:N-1
T(i,j) = 0.25*(oldT(i-1,j) + oldT(i+1,j) + oldT(i,j-1) + oldT(i,j+1));
end
end
Here is a faster solution obtained by using the intrinsic matrix operations of MATLAB. The
essential change is made by converting the for loops to
and changing them to
T(2:N-1,2:N-1) = 0.25* ...
( ...
oldT(1:N-2,2:N-1) + oldT(3:N,2:N-1) + oldT(2:N-1,1:N-2) + oldT(2:N-1,3:N) ...
);
This will be demonstrated in lecture on a related problem. The speedup is highly dependent
on computer architecture. On the laptop the typical speedup is a factor of only about 2 for
a 50 by 50 array of temperature. On a high-end workstation, Ive obtained as high as 30
times faster.
13.3
Selected Applications
13.3.1
Free fall with air resistance
Lets consider another trajectory problem that we will solve using the stepping method.
Consider taking an object at some height, y0 and dropping it with zero velocity, v0 = 0.
396
This is a one-dimensional problem and can be solved completely with equations. However,
we will execute a computer solution as a model for solving more complex problems, like
trajectories of baseballs. The air resistance on a baseball is very signicant. If there were no
air resistance, the longest home runs (about 600 feet or so) would travel half a mile!
The acceleration on the object is:
a = Dv 2 G
(13.1)
where D is the drag and G is the acceleration due to gravity.

We can solve this in a formal fashion as follows:

y(t) = y0 + v0 t +
t
dt
0
dt a[v(t )]
(13.2)
In fact, even if a were dependent on position y and time t we could write a formal solution
as:
t
t

dt
dt a[y(t ), v(t ), t ]
(13.3)
y(t) = y0 + v0 t +
0
But, this is a formal solution only and does not lead to any practical answers, only methods
of attack on this problem.
One such attack on the problem is to divide the integral above over a set of time interval t
and set up an iterative equation as follows.

yi = yi1 + vi1 t +
ti1 +t
t
dt
ti1
dt a[y(t ), v(t ), t ]
(13.4)
ti1
which is still exact. However, if we make the approximation that the acceleration is constant
over the step, that is
ai [y(t), v(t ), t ] ai [y(ti1 ), v(ti1 ), ti1 ]
(13.5)
we get
1
yi yi1 + vi1 t + ai [y(ti1), v(ti1 ), ti1 ]t2
2
and an algorithm that we can program:
1. Set initial conditions y0 , v0 at t = 0.
2. Initialize a counter i = 1.
3. Increment the counter i = i + 1
4. Calculate the acceleration ai [y(ti1 ), v(ti1 ), ti1 ]
5. Take a small step yi = yi1 + vi1 t + 12 ai [y(ti1 ), v(ti1 ), ti1 ]t2
(13.6)
13.3. SELECTED APPLICATIONS
397
6. Calculate the new velocity vi = ai [y(ti1 ), v(ti1 ), ti1 ]t

7. Update the time ti = ti1 + t
8. Go back to step 3 unless some stopping criterion is reached.
Note that two-dimensional problems are not really more complicated. There would just be
more variables to carry around.
Now the central question arises.
How small should I make t?????
The answer to THIS question is the art of computer programming and lies at the heart of the
general eld of Numerical Methods. The eld of Numerical Methods is also concerned
with developing better and faster techniques for solving problems than by the brute force
method outlined above. However, for this problem, we can more or less guarantee that our
results are good if the acceleration changes very little over a time step, namely
da
t a
dt
(13.7)
At this point, lets simplify the discussion and consider just our problem. If we apply the
above condition to the problem we get
t
1
2Dv
(13.8)
Does this make sense? It does! If there is no drag, D = 0, and we can make t anything we
like! This is because with no drag there is constant acceleration and we dont have to make
the steps small to get an accurate solution! (We have to be careful that we do not violate
our stopping criterion, however. More on this later.)
What if the v is small? In this case we can set t to be large because the drag is not aecting
the motion very much. If D and/or v are large enough such that Dv is large, we had better
use small time steps because drag forces are strongly aecting the trajectory.
How do we know how to set t for our problem? The greater the velocity, the greater care
we have to take by setting smaller time steps. The greatest velocity we can reach with our
problem is called the terminal velocity. This is when the drag force and the acceleration
force exactly balance each other and there is no net acceleration. For free fall with air
resistance, this velocity is

G
vT =
(13.9)
D
For our baseball the terminal velocity is something like 129 ft/s.
398
This is the upper limit of the velocity and so we are conservative by saying that
1
t
2 DG
(13.10)
or t 2.
How much smaller than 2 should you make t? Usually, as small as we can get away with!
If you want the error to be of the order of 1% or less, you should set it to something below
0.02 seconds. At least we have some indication from a simple analysis.
Now we implement the algorithm with MATLAB coding. Here is what the essential working
part of the algorithm would look like:
y = 1000; % Starts at 1000 feet
v = 0
; % with 0 velocity
while (y > 0)
a = D*v^2 - G;
% Acceleration at beginning of step
y = y + v*dT + 0.5*a*dT^2; % Position at end of the step
v = v + a*dT;
% Velocity at end of the step
end
Heres what the complete code would look like. It gives some example of organization and
plotting:
%File: drag.m
clear
echo off
clc
% Clears variables
% Turns off echo
% Clears the screen
G = 32.174;
D = 0.0019285;
dT = 10;
% Force due to gravity

% Drag on a baseball
% Initial time step
vT = sqrt(G/D); % Terminal velocity

fprintf(Terminal velocity is %g ft/s\n,vT)
while dT > 0
dT = input(What delta time to use (0 or less to stop)? >);
fprintf( Using dT = %g\n,dT)
if dT <= 0
return % dT <= 0 used as an input sentinel
399
end
v = 0
; % with 0 velocity
index = 1; % Index into the counting arrays
Y = y; V = v; T = 0; % Plotting arrays
while (y > 0)
a = D*v^2 - G;
% Acceleration at beginning of step
y = y + v*dT + 0.5*a*dT^2; % Position at end of the step
v = v + a*dT;
% Velocity at end of the step
% Plotting arrays
index = index + 1;
Y(index) = y;
V(index) = v;
T(index) = T(index - 1) + dT;
end
subplot(1,2,1)
zoom on
hold on
plot(T,Y,-)
title(y as a function of t)
xlabel(time (s))
ylabel(height (ft))
subplot(1,2,2)
zoom on
hold on
plot(T,V,-)
title(v as a function of t)
xlabel(time (s))
ylabel(height (ft/s))
fprintf(Final velocity = %g\n,V(end));
end
When we run this code, as demonstrated in class, we see that the trajectory always over-
400
shoots. We always end up with a negative y at the end. We have to make t very small to
get the nal position close to zero. We can x this problem relatively easily.
What we can do is check to see if the t we are about to use will cause an overshoot and if
it will, to truncate it accordingly.
How do we truncate it? Recall that we are assuming that the acceleration over the course
of the step is constant. If that is so, we can calculate the t required to bring it exactly to
zero. It must satisfy the quadratic equation:
1
0 = yi1 + vi1 t + ai [y(ti1 ), v(ti1 ), ti1 ]t2
2
(13.11)
You can solve this with the quadratic equation and convince yourselves that the solution is:
dT = -(v + sqrt(v^2 - 2*y*a))/a;
Here is the new working part of the algorithm incorporating this idea. (See the example
code drag0.m.)
v = 0
; % with 0 velocity
while (y > 0)
a = D*v^2 - G; % Acceleration at beginning of step
yTest = y + v*dT + 0.5*a*dT^2; % Position at end of the step
if yTest < 0
% Handle the overshoot
dT = -(v + sqrt(v^2 - 2*y*a))/a;
y = 0;
else
y = yTest;
end
v = v + a*dT; % Velocity at end of the step
end
So now we have a trajectory that always ends in the right place and an algorithm that is
guaranteed to get us close to the right answer. We are now a long way towards solving more
complicated problems, like two-dimensional trajectories of baseballs.
13.3.2
Diusion of charged ions in an electric eld
This is the problem of positive and negative ions created somehow (like a pulse of radiation)
in a gas drifting in opposite directions under the inuence of a constant electric eld. Run
401
the following code called ion.m that resides in the classcodes area. Note how the leading
edge is more populated with ions than the trailing edges since the longer an ion spends in
the vicinity of ions of opposite charge the greater the chance that it will recombine and
neutralize.
%File: ion.m
clear all
echo off
clc
NSteps = 100;
NIon = 2000;
for i = 1:NIon
x1(i) = rand; y1(i) = rand;
x2(i) = rand; y2(i) = rand;
end
plot(x1,y1,b*,x2,y2,r*)
axis([-1 2 0 1])
axis off
title(\fontsize{20}Initial configuration)
pause
for time = 1:NSteps + 3*sqrt(NSteps)
for i = 1:NIon
if (rand < 2/NSteps & x1(i) < 1)
y1(i) = -100;
end
theta = 2*pi*rand;
x1(i) = x1(i) + (1 + cos(theta))*0.5/NSteps;
y1(i) = y1(i) + sin(theta)*0.5/NSteps;
if (rand < 2/NSteps & x2(i) > 0)
y2(i) = -100;
end
theta = 2*pi*rand;
x2(i) = x2(i) - (1 - cos(theta))*0.5/NSteps;
y2(i) = y2(i) + sin(theta)*0.5/NSteps;
end
402

plot(x1,y1,b*,x2,y2,r*)
axis([-1 2 0 1])
axis off
title(\fontsize{20}\leftarrow +ions | -ions\rightarrow)
pause(0.01)
end
13.3.3
Calculation of electric elds
Constant electric elds as employed in the problem above are really ctional. In this simulation we mock-up a realistic electric eld situation. Run the following code called phi.m
that resides in the classcodes area. Note how strong the electric eld is near the insulator.
(It helps to have come to class to get the full description of this problem!)
%File: phi.m
clear all
echo off
clc
N = input(Number of cells = N x N. Input N: )
delta = input(Convergence criterion delta: )
V = zeros(N);
V(1,N/2:N) = 1000;
V(N,N/2:N) = 1000;
V(:,N) = 1000;
oldV = V;
change = 10;
tic
counter = 0;
while(change > delta)
V(2:N-1,2:N-1) = ...
0.25*( ...
V(1:N-2,2:N-1) + V(3:N,2:N-1) + V(2:N-1,1:N-2) + V(2:N-1,3:N) ...
);
difference = abs(V - oldV);

counter = counter + 1;
if mod(counter,100) == 0
fprintf(Interation %g, average change = %g\n,counter,change)
end
oldV = V;
end
toc
pcolor(1:N,1:N,V);
shading interp;
xlabel(\fontsize{20}X-grid);
ylabel(\fontsize{20}Y-grid);
title(\fontsize{20}Voltage in the chamber);
pause
figure
[Ex,Ey] = gradient(V);
E = sqrt(Ex.^2 + Ey.^2);
warning off
pcolor(1:N,1:N,log(E));
shading interp;
title(...
\fontsize{20}Mag. of electric field in the chamber);
pause
figure
surf(1:N,1:N,log(E));
shading interp;
title(...
\fontsize{20}Mag. of electric field in the chamber);
403
404
13.4
Summary of the Course
By now you are conversant in two languages, C++ and MATLAB. You have noted the
dierences between them. C++ is hard to make work properly. It is a powerful, generalpurpose language that runs numerical calculations very fast.
MATLAB is specialized for engineering-type problems. It has lots of engineering-type utility
built in and has very nice plotting capabilities, particularly in 3 dimensions. It is easy to get
an application up and running, but it is relatively slow for large-scale problems.
Not every computer language is suitable for every purpose. Heres how a research scientist
would use both of these computer languages to his or her best advantage.
1. Design a computer algorithm in pseudo-code
2. Convert it to MATLAB
3. Debug the concept of the algorithm using MATLAB
4. Develop graphics output to interpret results
5. Develop a set of test problems that can be completed by MATLAB in a reasonable
time.
6. When the MATLAB version is acceptable, re-code the algorithm in C++ starting with
the MATLAB code.
7. Run the same set of test problems using the C++ code.
8. Make sure that the C++ results and the MATLAB results of test problems are identical.
9. When the C++ code passes the tests, increase array size or introduce sucient complexity to solve the problem you really want to solve.
10. Have C++ write data arrays that MATLAB can read in an display, using MATLABs
excellent graphics capabilities.
11. Publish results!
The upshot is that both ways of doing it has its strengths and weaknesses. Why not use
both?
Chapter 14
Programming Style Guide for C++
This programming style guide is a set of mandatory requirements for C++ code layout. These
stylistic laws are not made for arbitrary reasons. Experience teaches us that conforming to
this style, more or less standard through the industry with few exceptions, makes code easy
to maintain and allows it to be read less painfully by others. This style guide is mandatory
for ENG101, and can only be relaxed under specic instructions from GSIs and Profs.
This style guide is summarized by example:
Indent 3 spaces relative to the preceding for every new nesting level. Curly brackets
must always align vertically. Example,
int main(void)
{
int i = 0, k = 1;
if (i < k)
{ int temp = i;
i = k;
k = temp;
if (0 == k)
{ i = i*i;
k = -i;
}
}
return 0;
}
Variable and function names are in lowercase, terse but descriptive, enough information
conveyed within the programming context, with capital letters to start new words
internal to the identier. Example,
405
406
CHAPTER 14. PROGRAMMING STYLE GUIDE FOR C++

numStuds = 211;
degToRad = 180/PI;
Dont over do it! Bad example,
numberOfStudentsInTheENG101Setion200Class = 211;
conversionFactorForDegreesToRadians = 180/PI;
The rst encounter with a variable merits a short comment unless the meaning is
completely obvious. Example,
numStuds = 211; //Number of students in the ENG101 Section 200 class
degToRad = 180/PI; //Conversion factor for degrees to radians.
Never use underscores in function and variable names. Bad example,

this_variable_name_is_difficult_to_type = 211;
cute_______________but_dumb = true;
Constant names should be in uppercase. Leading underscores are forbidden. Underscores in the middle are acceptable although discouraged. Example,
const float PI = 3.14159;
const float EULER_CONSTANT = 0.577216;
A variable should be initialized near where it is rst used and should only be dened
within the scope for which it is relevant. Examples,
float x[SIZE], y[SIZE];
for (int i = 0; i < SIZE; i = i + 1) // i only needed with scope of for loop
{ x[i] = 0;
y[i] = 0;
}
float z[SIZE] = {0};
int i; // i needed outside scope of for loop
for (i = 0; i < rand()%SIZE; i = i + 1) z[i] = 1;
cout << "The first " <, >=, &&
and ||. Examples,
407
float result = x + y - z; //Good
float result=x+y-z; //Bad
There are no spaces before and after the binary operators *, /, and %. There is no
space following the unary operators - and !. Examples,
float result = x*y + z; //Good
bool decision = ! go; //Bad
No hardwired constants!
Bad example:
int a[10][10];
for (i = 0; i == 9; i = i + 1)
for (j = 0; j == 9; j = j + 1)
a[i][j] = 1;
Good example:
int a[SIZE][SIZE];
for (i = 0; i < SIZE; i = i + 1)
for (j = 0; j < SIZE; j = j + 1)
a[i][j] = 1;
Functions should be small enough so that they t on the screen. Thats about 30 lines
of code.
The use of goto, break and continue is disallowed.
Use of non-constant global variables is discouraged.
C++ Code layout

Every C++ le, including main() is laid out as follows:
Header comment block
#include statements
Constants
408
Global variables
Functions, with own headers. Function prototypes are optional unless required by
ordering constraints.
Here is an example of code layout:
409
// HEADER COMMENTS GO AT THE TOP ************************************
/********************************************************************
Copyright (C): University of Michigan
Project:
ENG101/200 - Class example
File:
C++example.cpp
Purpose:
Coding example to show code layout
Compiler:
g++
Programmer:
Alex Bielajew
Start Date:
01/18/01
*********************************************************************/
// INCLUDE statements ***********************************************
#include <iostream>
// CONSTANTS ********************************************************
// No constants in this example
// GLOBAL VARIABLES *************************************************
// No global variables in this example
// FUNCTIONS ********************************************************
float pow(float x, int n)
/*********************************************************************
Purpose: Calculate the integral power of a float using the algorithm
of al-Kashi
Receives: "x" which is to be raised to the power "n"
Returns: a float, the result of x raised to the power n
*********************************************************************/
{

if (0 > n)
{ n = -n;
410

x = 1/x;
}
result = 1;
while(1 <= n)
{ if (0 == n % 2)
{ n = n/2;
x = x*x;
}
else
{ n = n - 1;
result = result*x;
}
}
return result;
411
int main(void)
/*********************************************************************
Purpose: To prompt the user for a float x and and int n and call a
function that computes x to the power of n
*********************************************************************/
{
cout << "A calculation of x (float) to the power of n (int) \n";

float x; // The float to be raised to a power
int
n; // Power to be raised to
cout << "Input x (float) and n (int): ";
cin >> x >> n;
result = pow(x,n);
cout << x << " to the power " << n << " = " << result << "\n";
return(0);
412
Chapter 15
Syntax reference for beginning C++
4 Bit Computer Arithmetic
BINARY
0000
0001
0010
0011
0100
0101
0110
0111
1000
1001
1010
1011
1100
1101
1110
1111
HEX
0
1
2
3
4
5
6
7
8
9
A
B
C
D
E
F
UNSIGNED
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
SIGNED
0
1
2
3
4
5
6
7
-8
-7
-6
-5
-4
-3
-2
-1
BINARY ARITHMETIC (BASE 2)

0000 + ? = ? + 0000 = ?
0001 + 0001 = 0010
0010 + 0001 = 0001 + 0010 = 0011
0011 + 0001 = 0100
0010 + 0010 = 0100
0001 * ? = ? * 0001 = ?
0010 * 0011 = 0110
0011 * 0011 = 1001
i (integer), ~i (bit-complement of the integer)

The integer 5 in 4 bit binary is 0101
The complement of 5 is ~5 = 1010
5 + ~5
= 0101 + 1010
= 1111
5 + ~5 + 1 = 1111 + 0001
= 0000
-i = ~i (bit-compliment of i) + 1
-5
= ~5 + 1
1011 = 1010 + 0001
32 Bit Computer Arithmetic

BINARY
00000000000000000000000000000000
00000000000000000000000000000001
00000000000000000000000000000010
...
00000000000000000000000000001111
00000000000000000000000000010000
00000000000000000000000000010001
...
01111111111111111111111111111111
10000000000000000000000000000000
...
11111111111111111111111111111110
11111111111111111111111111111111
HEX
00000000
00000001
00000002
...
0000000F
00000010
00000011
...
7FFFFFFF
80000000
...
FFFFFFFE
FFFFFFFF
UNSIGNED
0
1
2
...
15
16
17
...
2,147,483,647
2,147,483,648
...
4,294,967,294
4,294,967,295
413
SIGNED
i =
00000000000000000000000000010001
0
~i =
11111111111111111111111111101110
1
2
i
00000000000000000000000000010001
...
+ ~i = + 11111111111111111111111111101110
15
------------------------------------16
11111111111111111111111111111111
17
...
2,147,483,647
-2,147,483,648
...
-2
-1
414
CHAPTER 15. SYNTAX REFERENCE FOR BEGINNING C++
Algorithms and Pseudocodes

An Algorithm is a set of simple (i.e. well-dened) instructions to be carried out in sequence to accomplish some task. It
has a start and a stop and can execute branches and loops.
A Pseudocode is a way of expressing the actions and ordering to be taken place on a computer in an exact yet informal
way, without details or syntax of a particular computer language.
e.g. IF such-and-such THEN DO
this-and-that
OTHERWISE DO
something-else
Order of Precedence
OPERATOR
()
* / %
+ < <= > >=
== !=
&&
||
=
ASSOCIATIVITY
left to right
left to right
left to right
left to right
left to right
left to right
left to right
right to left
Logical Expressions
T && T = T, T && F = F, F && T = F, F && F = F

T || T = T, T || F = T, F || T = T, F || F = F
!T = F, !F = T
False is the same as 0, True is anything else
415
If/Else
if(logical expression 1)
else ifs are optional
else is optional independent of the existence of else ifs
STATEMENT BODY 1
If there is only one statement in the statement body, you
}
else if(logical expression 2)
{
may ignore the {}, but for clarity it is better to keep them
STATEMENT BODY 2
}
// more else ifs if needed
else
{
STATEMENT BODY N
Loops
While
If the logical expression is False the rst
while(logical expression)
{
time, the statement body is never executed
STATEMENT BODY
If the logical expression is always True, then
you have an innite loop

Do/While
The statement body is evaluated once before
do
{
the logical expression is evaluated at all
STATEMENT BODY
Do NOT forget the ; at the end
}while(logical expression);
For
for(expression 1; logical expression; expression2)
expression 1 is optional
logical expression is optional
STATEMENT BODY
expression 2 is optional
;; are not optional

for(;;){} is an innite loop
Example Programs
CODE
#include<iostream>
OUTPUT
CODE
#include<iostream>
OUTPUT
416
CHAPTER 15. SYNTAX REFERENCE FOR BEGINNING C++

int main(void)
{ int i;
for(i = 1; i <= 10; i = i + 1)
cout <
int main(void)
i is 2. j is -2.
{ int i = 2,j = -2;
i is 1. j is -1.
while(j <= i)
{ if(j < 0 )
{ cout << "i is " << i <<". "
<< "j is " << j << ".\n";
i = i - 1;
}
j = j + 1;
}
return 0;
}
2
3
4
5
6
7
8
9
10
11
11
Chapter 16
Syntax reference for more advanced
C++
Addresses
The address of a variable is its location in memorytypically some number from between 0 and 232 1.
The address of a variable is obtained with the unary & operator.
The piece of code {int i; cout << &i << "\n";} should be read as i is declared an int. Print the address of i in hexadecimal
format..
Functions, call by value, call by reference

There are two important things to remember for functions: 1) the function denition and 2) the function call:
The function denition is written at the top of the C++ le outside of main or any other function. A function denition
must be given before any code that uses the function via a function call. Generic syntax:
[void|int|float|...]
functionName(variable type & name list separated by ","s){C++ code}
Examples: int function1(int a, float b){int c;...;return c;}

void function2(int& c, float& d){...;return;}
Note: Variable types and names in parameters lists of function denitions MUST BE INCLUDED.
If the return value is NOT void, the function must return the correct variable type.
If the return value IS void, the return; is OPTIONAL.
Functions can return at most ONE VARIABLE.
The function call is made inside of main or another function. Generic syntax:
[variable =] functionName(variable name list separated by ","s);
Example: int a,c;float b;...;c = function1(a, b);
Note: Variable names in parameters lists of function calls MUST BE INCLUDED.
Variable types in parameters lists of function calls MUST NOT BE INCLUDED.
417
418
CHAPTER 16. SYNTAX REFERENCE FOR MORE ADVANCED C++
Function call by value means that the parameter list of the function denition contains values. Example:
float function1(int c, float d){return d/c;} // function definition
x = function1(a, b); // function call
Note: Copies of these values are made available to the function. These are lost when the function returns.
Function call by reference means that the parameter list of the function denition contains addresses. Example:
function2(int& c, float& d){d = d/c;} // function definition
function2(a, b); // function call
Note: Functions called by reference can change the contents of the addresses that are made available to the
function in the reference list.
Vectors
Using vectors
#include <vector> // Must be included in the pre-processor area
Note: Vectors are a class dened in standard C++
Declaring vector<int> vectorName(10); // An int vector called vectorName of size 10;
Note: Vectors are initialized to 0 when they are declared.
Accessing vector elements
for(i = 0; i < 10; i = i + 1){vectorName[i] = i + 1}; // for loop assigns all elements of vectorName
i = vectorName[0]; // i is assigned 1st element of vector vectorName
vectorName[9] = i; // The last (10th) element of vector vectorName is assigned the value of i
Note the o-by-one syntax in the above three examples
Passing vector to functions
double function1(vector<double> b){...} // Typical definition for call-by-value
vector<double> a(10);...;x = function1(a); // Typical call-by-value
double function2(vector<double>& d){...} // Typical definition for call-by-reference
vector<double> c(10);...;x = function2(c); // Typical call-by-reference
Vector class functions
vector<double> a(10);cout << a.size(); // Prints the size of the a vector
vector<float> b(10);b.pop_back(); // Reduces the size of the b vector by one to size 9
vector<int> c(10); int d = 42; c.push_back(d); // Increases the size of the c vector by one to size 11,
// with value of int d in the last vector element
Two dimensional (2D) arrays

Declaring int arrayName[2][3]; // Array called arrayName of size 2 rows by 3 columns
Declaring and initializing
int arrayName[2][3] = {{1,2,3},{4,5,6}}; // Initialize all the elements of arrayName;
int arrayName[2][3] = {{1,2,3}}; // Initialize the first row of arrayName, rest to zero;
419
int arrayName[2][3] = {0}; // Initialize all the elements of arrayName to zero;
Accessing array elements
// for loops below assign all elements of arrayName
for(i = 0; i < 2; i = i + 1) for(j = 0; j < 3; j = j + 1) arrayName[i][j] = 3*i + j;
i = arrayName[0][2]; // i is assigned 1st row/3rd column
element of array arrayName
arrayName[1][1] = i; // The last 2nd row/2nd column element of array arrayName is assigned the value of i
Note the o-by-one syntax in the above three examples
Note: arrayName is the address of arrayName[0][0], that is, arrayName is the same as &arrayName[0][0]
Passing 2D arrays to functions Remember that an array name is the address of the start of the array
// Note that column dimension must be specified below
void function(double b[][3], int rowSize, int columnSize){...} // Typical definition
double a[2][3];...;function(a); // Typical call
Note that that the column size (second index) must be known by the function!
Characters and strings

chars are one-byte integers, strings are arrays of chars that includes a special ending character.
Declaring a character char ch; // Declares a character;
Declaring and initializing a character
char ch = a; // Declares a character called ch and initializes it to a. Note single quotes.
Declaring a string string name; // Declares a string
Declaring and initializing a character string
string name = "Albert Einstein"; // Declares a string
Note the use of double quotes in assigning strings.
Printing strings
string name = "Barry Manilow"; cout << name;
Structures
are compound data types. Structures have to be: 1) dened, 2) declared, 3) initialized, 4) used.
Dening a structure
struct AgeAndHeight {int age; float height;}; // Structure tag AgeAndHeight with an int and a float;
Declaring a struct
struct AgeAndHeight Wilma; // Declares Wilma as an AgeAndHeight struct
Declaring and initializing a character
struct AgeAndHeight Wilma = {26,65.5}; // Declares Wilma as an AgeAndHeight struct and initializes it
Referencing struct elements
420
CHAPTER 16. SYNTAX REFERENCE FOR MORE ADVANCED C++
Wilma.age = 27; // Wilmas age is assigned the value 27

cout << "Wilmas height is: " << Wilma.height << endl; // Prints Wilmas height
Struct vectors
vector<struct AgeAndHeight> theFlintstones(4); // A vector of structs of the AgeAndHeight type
cout << theFlintstones[0].height; // Prints the height of 1st element
Struct arrays
struct AgeAndHeight townOfBedrock[100][5]; // A array of structs of the AgeAndHeight type
Struct vectors and arrays are passed to functions in the same way as the standard variable types.
Recursion
Functions can call themselves. Factorial example:
int factorial(int n)
{ if (0 == n) return 1;
else return (n * factorial(n - 1));
}
The guidelines are:

Successive calls to functions should represent progressively simpler cases. (For example, (N 1)! is simpler than N !)
There is an end to the recursion, the base case, that does not involve a recursive call. (For example 0! = 1) Not doing this
will lead to a program or computer crash.
If a function calls itself more than once, and the nesting level gets deep, the calculation can be very inecient.
Chapter 17
Syntax reference for MATLAB
MATLAB Workspace and Variables
The basic variable type in MATLAB is a two-dimensional array of doubles (64-bit representation).
A scalar is a 1 1 array.
A row vector of length N is a 1 N array.
A column vector of length M is an M 1 array.
A matrix of dimensions M rows and M columns is an M N array.
Variable name conventions

MATLAB is case sensitive
Starts with A-Z, a-z
Up to 31 letters, digits and underscores
Default MATLAB variables

ans
eps
Inf
NaN
pi
realmax
realmin
result of last unassigned calculation

smallest number that can be added to 1.0
and still be dierent
innity e.g. 1/0 = Inf
Not a Number e.g. 0/0 = NaN
Value of Pi (3.1415...)
Largest real number MATLAB can represent
Smallest real number MATLAB can represent
421
422
CHAPTER 17. SYNTAX REFERENCE FOR MATLAB
User dened variables

clear
length
size
who
whos
clear all or selected variables (or functions)

from the current workspace
length of a vector or maximum dimension of an array
display dimensions of a particular array
display current workspace variable names
display current workspace variable names,
types and associated sizes
Command/Figure Window control

clc
clf
figure
figure(N)
close
close(N)
clear the command window

clear the gure window
start a new gure window
Make gure with index N active. If N is an
integer and figure(N) does not exist, create it
close current gure window
close gure with index N
Useful Workspace functions

help
lookfor
more
load
save
obtain help generally or for a specic function

obtain one-line help if it exists
toggles pagination, useful for longs helps
read variables in from a disk le
save all or selected variables to a disk le
MATLAB path and environment

cd
dir
path
change to a specic directory

list les in the current directory
display or modify the function search path
423
MATLAB punctuation
. decimal point usage: 325/100, 3.25 and .325e1 are all the same
... three or more decimal points at the end of a line cause
the following line to be a continuation
, comma is used to separate matrix subscripts and arguments
to functions, also used to separate statements in
multi-statement lines
; used inside brackets to indicate the ends of the rows of a
matrix, also used after an expression or statement
to suppress printing
% begins comments
Quote. ANY TEXT is a vector whose components are the
ASCII codes for the characters. A quote within the text is
indicated by two quotes. For example: Dont forget.
Assignment
Individual elements with a row can be delimited by a comma or
a space.
Explicit assignment using ;s to end rows
a = [1,2,3;4,5,6;7,8,9]
Explicit assignment using newline to end rows
a = [1,2,3
4,5,6
7,8,9]
Explicit assignment using continuation lines
b = [1 2 3 4 5 6 ...
7 8 9 10]
Useful ways to initialize variables

linspace(a,b,N)
424
Linearly spaced intervals between a and b (inclusive)

comprised of N points
zeros(m,n) An m by n array of zeroes
zeros(n)
An n by n array of zeroes
ones(m,n) An m by n array of ones
ones(n)
An n by n array of ones
eye(m,n)
An m by n array with ones on the diagonal
eye(n)
An n by n identity matrix
ones(n)
An n by n array of ones
rand(m,n) An m by n array of random numbers
rand(n)
An n by n array of random numbers
Vector/Matrix operators
These are Vector and Matrix operators
+
*
/
\
^
addition
subtraction
multiplication
left division
right division
exponentiation
transpose
Point-by-point operators
These operate on matrix elements in point-wise fashion
.*
./
.\
.^
point-wise
point-wise
point-wise
point-wise
multiplication
left division
right division
exponentiation
Logical operators
<
<=
>
less than
less than or equal
greater than
425
>=
==
~=
&
|
~
greater than or equal

equal
not equal
logical AND
logical OR
logical NOT
List generation/variable indexing

i:k:l
v(1)
v(end)
v(1:2:9)
v(2:3:9)
a(2,3)
a(:,3)
a(1,:)
a(1:2:end,:)
a(1:2,2:4)
a(1,end)
List generation syntax = 1stValue:Stride:LatValue

1st element of vector v
Last element of vector v
1st, 3rd, 5th, 7th, 9th elements of vector v
2nd, 5th, 8th elements of vector v
2nd row, 3rd column of matrix a
all elements in column 3
all elements in row 1
all odd rows of matrix a
sub-matrix of rows 1 and 2, columns 2 through 4
last element in 1st row
Script M-les
Sequences of MATLAB commands can be stored in text les with the extension .m. The
commands can be executed by typing the name of the les (without the extension) or through
the le management tools provided by the Command Window menu.
Useful M-le functions

Function M-les
Function denition
Dene a separate le called functionName.m with the following form:
function [out1,...,outN] = functionName(in1,...inM)
%FUNCTIONNAME A brief one line description (optional)
426
%
%
%
%
%
%
%
.
.
.
%
.
.
.
%
.
.
More description (optional)
.
.
.
First executable statement follows this line
Valid executable MATLAB statements and comments
Last line in functionName
Function call
The function call is made with the following statement:
[out1,out2,...,outN] = functionName(in1,in2,...inM)
Useful M-le functions

disp
fprintf
echo
error
input
keyboard
pause
return
warning
display a string
write data to screen of le
toggle command echo
display message and abort
prompt for input
transfer control to keyboard
wait for time or user response
return to caller
display warning messages
Performance monitoring
tic,toc
flops
stopwatch timer functions

counts oating point operations
427
Formatting
format
format
format
format
short
long
compact
loose
Scaled xed point format with 5 digits

Scaled xed point format with 15 digits
Suppress extra line-feeds
Puts extra line-feeds in the output
Program Flow control

for loops
for k = vectorOrColumnList
% MATLAB statements
end
while loops
while logicalExpression
% MATLAB statements
end
if/elseif/else construct
if logicalExpression1 % Mandatory
% MATLAB statements
elseif logicalExpression2 % Optional
% MATLAB statements
elseif logicalExpression3 % Optional
.
.
.
elseif logicalExpressionN % Optional
% MATLAB statements
else % Optional
% MATLAB statements
end % Mandatory
428
Plotting
contour
contour3
mesh
meshc
meshz
pcolor
plot
plot3
surf
surfc
surfl
Contour plot on a plane

3D contour plot with displayed depth
3D mesh surface
Combination mesh/contour plot
3D mesh with curtain
Pseudocolor (checkerboard) plot
Basic 2D plots
Plot lines and points in 3D space
3D colored surface
Combination surf/contour plot
3D shaded surface with lighting
Plotting annotation
clabel
colorbar
legend
title
xlabel
ylabel
Contour plot elevation labels

Display color bar (color scale)
Graph legend
Graph title
X-axis label
Y-axis label
Additional plotting functions

box
colormap
grid
hold
shading
subplot
zoom
Toggle the box display

Color look-up table
Toggle the grid state
Control multiple plots on a single gure
Color shading mode e.g. flat, interp
Control multiple plots in one window
Enable mouse-based zooming
Index
algorithm
al-Kashis power algorithm, 38, 39
Breakfast, 28
denition, 1, 2328
design, 29, 38, 39
the three pillars of, 26
raising a number to an integral power,
38, 39
roots of the quadratic equation, 29
simple sequence example, 24

sequencing, see instruction, sequencing
statement, see instruction
computer
architecture, 23, 40
owchart, 24
simple branching example, 25
simple looping example, 26
simple sequence example, 24
Gauss, Karl Friedrich 17771855 AD, 35,
44
instruction
branch, 25
conditional, 25
decisions, 25
denition, 23
ow, 24
jump, 25, 26
looping, 25, 26
sequencing, 23
logic
ow, 24
pseudocode, 24
simple branching example, 25
simple looping example, 26
429

Introduction To Computers and Programming

Uploaded by

Copyright:

Available Formats

Introduction To Computers and Programming

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Introduction To Computers and Programming

Uploaded by

Copyright:

Available Formats

Introduction to Computers and Programming

What this course is about . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Bits, nibbles, bytes and words . . . . . . . . . . . . . . . . . . . . . . . . . .

4-bit Binary, Hexadecimal and Decimal Digits . . . . . . . . . . . . . . . . .

Converting from binary to decimal . . . . . . . . . . . . . . . . . . . . . . .

Binary integer arithmetic on computers . . . . . . . . . . . . . . . . . . . . .

Twos-complement integer arithmetic . . . . . . . . . . . . . . . . . .

32-bit Binary, Hexadecimal, Unsigned and Signed Integers . . . . . . . . . .

3 Algorithms and Pseudocodes

Decisions, conditionals, branch instructions

Looping or jump instructions . . . . . . . . . . . . . . . . . . . . . . . . . .

Sequencing, branching and looping . . . . . . . . . . . . . . . . . . . . . . .

A mini-summary before the examples . . . . . . . . . . . . . . . . . . . . . .

The Breakfast algorithm, or, Bielajews Sunday Morning Internationallyfamous pancakes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

Solve for x: Ax2 + Bx + C = 0 where A, B, C are arbitrary constants

Iteration: A summing loop . . . . . . . . . . . . . . . . . . . . . . . .

Iteration: A product loop . . . . . . . . . . . . . . . . . . . . . . . .

An aside on computer architecture . . . . . . . . . . . . . . . . . . . . . . .

What does S = S + 1 mean? . . . . . . . . . . . . . . . . . . . . . . .

4 Getting started in C++

Simple input/output (I/O): A rst program in C++ . . . . . . . . . . . . . .

Compiling, linking, loading and running . . . . . . . . . . . . . . . . . . . .

Declaring and initializing variables . . . . . . . . . . . . . . . . . . . . . . .

Integer math in C++ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Floating point math in C++ . . . . . . . . . . . . . . . . . . . . . . . . . . .

The if/else if/else construct . . . . . . . . . . . . . . . . . . . . . . . . .

Logical expressions with AND or OR . . . . . . . . . . . . . . . . . .

Mixed arithmetic and logical expressions . . . . . . . . . . . . . . . .

The while loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The do/while loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

The for loop . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Motivation for functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123

User-dened functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129

Example: A problem tackled with teamwork . . . . . . . . . . . . . . . . . . 134

Call by value, call by reference, reference parameters . . . . . . . . . . . . . 138

The address of a variable . . . . . . . . . . . . . . . . . . . . . . . . . 138

Call-by-value vs. call-by-reference . . . . . . . . . . . . . . . . . . . . 140

The rules of scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145

7 More Variable Types, Data Abstraction

Representation of oating-point numbers . . . . . . . . . . . . . . . . . . . . 167

When is one not one? Floating point anomalies. . . . . . . . . . . . . 174

7.2 char: the character variable . . . . . . . . . . . . . . . . . . . . . . . . . . . 175

Character strings; the string class . . . . . . . . . . . . . . . . . . . . 177

The vector class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179

Introduction to vectors . . . . . . . . . . . . . . . . . . . . . . . . . . 179

Declaring and using vectors . . . . . . . . . . . . . . . . . . . . . . . 180

Vector syntax and rules: . . . . . . . . . . . . . . . . . . . . . . . . . 182

Vectors and functions . . . . . . . . . . . . . . . . . . . . . . . . . . . 187

8 More Data Abstraction, Arrays, Structures

Declaring a 2-dimensional array . . . . . . . . . . . . . . . . . . . . . 200

Initializing a 2-dimensional array . . . . . . . . . . . . . . . . . . . . 201

Passing a two-dimensional array to a function . . . . . . . . . . . . . 204

Character arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209

One dimensional character arrays . . . . . . . . . . . . . . . . . . . . 209

Two dimensional character arrays . . . . . . . . . . . . . . . . . . . . 211

The structure denition . . . . . . . . . . . . . . . . . . . . . . . . . 212

Where can structures be dened? . . . . . . . . . . . . . . . . . . . . 213

The members of a structure . . . . . . . . . . . . . . . . . . . . . . . 213

Declaring structure variables . . . . . . . . . . . . . . . . . . . . . . . 213