Scribd Logo
Upload

Welcome to Scribd!

  • 42 views

    Probability Crash Course: Permutations and Combinations: School of Mathematics and Statistics

    Uploaded by

    dvdmega
    math

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd

    Probability Crash Course: Permutations and Combinations: School of Mathematics and Statistics

    Uploaded by

    dvdmega
    42 views19 pages
    math

    Original Title

    math

    Copyright

    © Attribution Non-Commercial (BY-NC)

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    math

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    42 views19 pages

    Probability Crash Course: Permutations and Combinations: School of Mathematics and Statistics

    Uploaded by

    dvdmega
    math

    Copyright:

    Attribution Non-Commercial (BY-NC)
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    You are on page 1of 19

    School of Mathematics and Statistics

    Probability Crash Course: Permutations and Combinations


    Paul Hewson Overview: This weble is designed as a revision aid to some introductory concepts in probability. It is intended to supplement a formal encouter with a text book or a set of lectures. These notes are meant to be slightly interactive, mysterious green dots, squares and boxes appear which you can click on to answer questions and check solutions.
    Copyright c 2008 Last Revision Date: 7th September 2009 paul.hewson@plymouth.ac.uk Version 0.1

    1. Overview Weve spent a lot of time looking at dice and coins. If we have 3 dice, each can show one of 6 faces. The number of possible outcomes are given by n1 n2 n3 = 63 . We would say each outcome can occur with 1 probability 63 But more generally, what about an experiment with equally likely outcomes, the problem of nding the probability of an event reduces to counting the number of outcomes in the event and in the sample space. But with realistic numbers this can get quite tedious. Consider the manager of a small plant wishes to determine the number of ways which he can assign sta to the rst shift. There are 15 sta who can serve as operators of the production equipment, 8 who can serve as maintenance personnel and 4 who can be supervisors. If a shift requires one operator, one maintenance person and one supervisor, how many dierent ways can the shift the staed? The answer is 480, which is rather tedious if one had to draw the tree diagram.

    Back

    Doc

    Doc

    Section 1: Overview

    1.1. N-tuple Denition 1 An N-tuple is a nite ordered set with an unspecied number of members For example, a 5-tuple is an ordered set with 5 members. Suppose that each outcome (i.e., element) of A takes the form of an n-dimensional vector (or n-tuple) such as (a1 , a2 , . . . an ). If there are N1 objects that can be used for a1 , and N2 for a2 then: n(A) = N1 N2 . . . Nn (1)

    assuming that sets A1 , A2 , . . . Ak have respectively n1 , n2 , . . . , nk dierent ways in which one can rst take an element of A1 , then an element of A2 and so on. So for our factory problem, we have N1 = 15, N2 = 8 and N3 = 4, with 15 8 4 = 480. This is rather an intuitive result, but it seems too convenient to be true that we needed exactly one maintenance person, one production person and one supervisor.

    Back

    Doc

    Doc

    Section 1: Overview

    We consider two scenarios: Permutations (the order of the elements is important) Combinations (the order of the elements is not important)

    Back

    Doc

    Doc

    Section 1: Overview

    1.2. Permutation Given a population of n distinct elements, how many ordered samples of size r can be drawn: (i) with replacement: nr (ii) without replacement n(n 1)(n 2) . . . (n r 1), calculation n! reduces to: (nr)!

    Back

    Doc

    Doc

    Section 1: Overview

    For a 6 digit telephone number, 1. how many dierent telephone numbers can be made if each digit can be 0,1,2,3,4,5,6,7,8,9: 2. by how much is this reduced if the rst digit cannot be 0: Points: Click on the Ans box to get the correct answer; shift + click to get the solution.

    Back

    Doc

    Doc

    Section 1: Overview

    1.3. Combinations Two sets are regarded as disjoint if and only if the properties of the membership are mutually exclusive. A partition is dened as a set of disjoint and exhaustive subclasses of a given class that is divided in such a way that each member of the given class is a member of exactly one such subclass. The number of ways a set of n objects can be partitioned into 2 distinct subsets of size r and n r is given by: n r = n! r!(n r)! (2)

    Back

    Doc

    Doc

    Section 1: Overview

    1. The UK National Lottery requires that you correctly choose 6 numbers out of possible 49. How many ways re there is selecting these numbers; 2. In a communications channel, messages are transmitted in the form of a sequence of 8 binary digits. How many distinct messages are there containing 5 zeros and 3 ones? Points: Click on the Ans box to get the correct answer; shift + click to get the solution.

    Back

    Doc

    Doc

    Section 1: Overview

    1.4. Partitioning into k-subsets The number of ways a set of n objects can be partitioned into k distinct subsets, where set 1 has r1 elements, set 2 has r2 elements, . . . , set k has k elements is: n! r1 !r2 ! . . . rk ! (3)

    1. Consider sampling from 6 Midwestern states, =(Iowa, Illinois, Wisconsin, Michigan, Indiana and Ohio). For the purposes of a study, we need a sample of 4 states. How many samples of size 4 are possible. 2. How many samples of size 4 contain the state of Illinois. Points:

    Back

    Doc

    Doc

    10

    2. Pascals Triangle Its not worth losing sleep over, but its impossible at this stage not to mention Pascals1 triangle:
    1 1 1 1 1 1 1 1 7 6 21 5 15 35 4 10 20 35 3 6 10 15 21 2 3 4 5 6 7 1 1 1 1 1 1 1

    Note that each number in the triangle is the sum of the two numbers either side of it in the row above. Now consider you have a pool of 7 objects, and wish to know how many dierent ways there are of arranging 7! them into sets of 3. We used above n = 7 = 3!(73)! = 35. However, r 3 if we number the rows of Pascals triangle from 0 (at the top) to 7, take row 7 and count along to the 4th entry, we have 35. Isnt that cute?
    1 apparently discovered by Jia Xian in 1050, published by Zhu Shije in 1303, discussed by Cardano in 1570

    Back

    Doc

    Doc

    11

    3. A fun exercise This really is worth doing! Quiz Birthdays How many students must there be in a class in order for there to be a 50% chance that two will share the same birthday (a) 22 (b) 37 (c) 183 (d) 365 Maybe you should check this out. Can you get the dates of birth for all your fellow students (or is this a silly question)? If so, how many coincidences are there?

    Back

    Doc

    Doc

    Section 3: A fun exercise

    12

    Calculating the paradox. Note that there are 365n possible birthdays in a set of n students (we ignore leap years and assume birthdays are independently distributed throughout the year). 365! We have (365n)! possible permutations. The probability of 2 coincident birthdays is therefore given by 365! (365 n)!365n This can be rather a dicult computation (365! is rather large and we need to use logarithms). There is an inbuilt R function, we can use this to plot the probabilities for a range of class sizes:

    Back

    Doc

    Doc

    Section 3: A fun exercise

    13

    ## create a storage vector for the results pb <- vector("numeric",60) ## loop and calculate probability of coincidence ## for different class sizes for (i in 1:60){ pb[i] <- pbirthday(i) } ## plot plot(c(1:60), pb, type = "l", xlab = "Class size", ylab = "Prob of coincidence", main = "Birthdays") What do you notice about the curve? Do you get a dierent curve if you calculate this for yourself (in either R or Python) using a function with: 1 exp{log(f actorial(365)) log(f actorial(365 i)) i log(365)} Checking the R helple (?pbirthday) may be informative. What happens if you look for the probability that three or more people share a birthday pbirthday(i,coincident=3)

    14

    Solutions to Quizzes Solution to Quiz: The answer is given by the rst denition (we are sampling with replacement, as all the digits can be used in any position) hence we want 106

    Click on that green button to return to the quiz

    Solutions to Quizzes

    15

    Solution to Quiz: For ve positions we have simply 105 , but there are only 9 possibilities for the rst digit hence the total number available is 105 9 = 900, 000. This is 100, 000 less than the previous answer.

    Click on that green button to return to the quiz

    Solutions to Quizzes

    16
    49! 6!(496)!

    Solution to Quiz: We need to calculate

    Click on that green button to return to the quiz

    Solutions to Quizzes

    17
    8! 5!(85)!

    Solution to Quiz: We need to calculate

    Click on that green button to return to the quiz

    Solutions to Quizzes

    18
    n r

    Solution to Quiz: We want

    6 2

    6! (64)!4!

    = 15

    Solutions to Quizzes

    19

    Solution to Quiz: Correct answer is 10.

    You might also like