0% found this document useful (0 votes)
45 views8 pages

Two Phase Method Notes

Uploaded by

rookhum
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
45 views8 pages

Two Phase Method Notes

Uploaded by

rookhum
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

SIMPLEX ALGORITHM 221

S.1 4 JLLUSTRA TIVE EXAMPLES.


problem :
EX· 1. Use two phas_e r:zethod to solve the following
Maximize z = 3x1 - x2 , ( x 1, x2 ~ 0 )
subject to . 2x1 + x2 ~ 2,
X1 + 3x2 $ 2,
{North BengalHons., 2007 J
X1 $ 4.
blem reduces to
After introducin~ s~ack and surplus variables, the pro
Maxumze z = 3x1 - x2 + 0. x3 + 0. x4 + 0. x5
subject to · 2x1 + x2 - x3 = 2,
X1 + 3x2 + X4 = 2,
~ I + Xs= 4.
x to the first
Obviously, we are ·to introduce one artificial variable 6
ints become
constraint equation soJhat the new system of constra
X3 + 0. X4 + 0. Xs + x6 = 2,
2x1 + X2 -
X1 + 3x2 + 0. X3 + X4 + 0. Xs + 0. X6 = 2,

X1 + 0. X2 + 0. X3 + 0. X4 + Xs + 0. X6 = 4,

in which ~ ~ 0, j = 1., 2, 3, 4, 5, 6:
The au~iliary objective function to be maximized
is given by

z· =·0. X1 + 0. X2 + 0. X3 + 0. X4 + 0. X5 - x6
subject to the above system of equations.
es and x 6 is the
Here x1, x 2, x 3, x 4 and x 5 are the legitimate variabl
artificial variable.
and obtain
We then construct the simplex tableaux successively
PHASE I : TABLEAU 1
- I
- c·J 0 0 , 0 0
a.
0
as
-]
Cg
r--:: _ B
~
Xa b
2 [I]
81 82
I
83
- 1 0 0
°'I
x"
'O 84 2 I 3 0 1 0 0
0. X4
0
as Xs 4 l 0 0 0 1
I'- -
<.j - C; 2 l l 0 0 0
i J,
· LINEAR PROGRAMMING
r... •.
222
PHASE t : TABLEAU 2

Cj
0 0 0
31 32 83
Xs b
B -21 - .!.
1 1 2
Xi
0 0 -5 1
X4 1 2 2
0 84 1 1
3 0 2 2 0
as Xs
0 I

Zj- Cj 0 0 0

Here zi _ ci ~ o in this iteration and Max. z* = O. No artificial


. ble appears in the final tableau of phase I. We then pass on tothe
vana . . b . . .
phase II where the objective function to e max1m1zed 1s
z = 3x1 - X2 + Q. X3 + 0. X4 + _0. 'X5•
The last basis of phase I will·be the starting basis for phase nand
the last tableau of phase I will act as the initial tableau of phase nwith
a change only tn the objective row containing the prices. cs column
being ch~ged accordingly, there will be a new index row showing
( zi - ci ) in the initial tableau of phase II. As there is no artificial
vector in the basis of the last tableau of phase I, we forget about it and
delete this column . .
PHASE II: TABLEAU 1

c·J 3 -1 0 0 0
Cs B Xs., .b . 81 32 33 a. as
3 1 0 . 0
31 Xi 1 1 -1
2 2
0 1. 0
0
~

3s
X4

Xs
I
3
0
0
5
-
2
- .!
lTI I 0 \
2 2

Zj- Cj 0 -5
3 0
2 2

i
~
SIMPLEX ALGORITHM
I 223

is the entering vector and 8 4 is the departing vector.


e thata3 .
We se to cons truct the next tableau. •
ass on · .
111enweP PHASE II: TABLEAU 2 .

c·J 3 -1 0 0 0
___...,..--..--B XB b 81 82 83 84 85
___.
~ a, X1 2 1 3 0 1. 0
3 I

a3 X3 2 0 .5 1 ·2 0
0
, 2 0 !-3 0
0 as Xs - 1 1
i---
0 10
- 0
Zj- Cj 3 0
Here zi- ci ~ 0, for all J.. The optunality. ..
condition has been
satisfied. The optimal solution is
x1 = 2 , x2 = 0 and z~ = 3.2 - 0 = 6.
Ex. 2. Use two-phase simplex meth(!d to solve the problem
.
.
Maximize Z = 2x1 + Xi+ X3
subject to 4x 1 + 6.x2 + 3x3 ~ 8,
.
\
-
3x1 - 6.x2 - 4x3 ~ 1,
2x1 + 3x2 - 5x3 ~ 4,
x1 , x2 , ~ ~ 0. [ Nof'!h Bengal.Hons., 200 7]
To put the problem in the standard form, we introduce the slack
variables x4 ~ 0, x5 ~ 0 and a· surplus variable x ~ 0. The problem
6
is then stated.as .,,,,- .·
· Maximize 2x1 + X2 + X3 '. 0. _X4 + 0. X5 + 0. X6
Z=
subject to 4x1 + 6.x2 + 3x3 + X4 =· 8,
3x1 - 6x2 - 4x3 + x5 = 1,
2x1 + 3Xi - 5x3 :- x6 = 4.
c Obviously we are to introduce one artificial variable x., to the
st third
on raint equation so that the new set of constraints become
4x1 +. 6x2 + 3x3 ~ x4 = 8,
3x1 -6x2 -4x~ - +·x5 =1,
2xi + 3x2 - 5x3 - x6 + x., = 4. ·
LINEAR PROGRAMMING

224
.. biective function' to be maximized is .
The auxthafY O J given by
• Q + Q. x2 + 0. X3 + 0. X4 + 0. x5 + Q, X
z :::: . X1 6 - x1

. t those constraints.
subJect o
ct the simplex tableaux successively and
~~~s tJ.1.l ~

PHASE 1 : TABLEAUX 1 AND 2

0 0 0 0 ------r,-......
0 ... I
Cj 0

b a1 a2 33 a4 as -~
r--..
a.,

--
B
CB
-
~ 3 1 0 0 0
34 8 4
0 1
3 -6 -4 0 0 0
0 as · 1
3 -5 0 0 -1 1
a, 4 2 -
-1 0 1
·-2 -3 5 0 0
z;- Cj

4 2 1 -I -6I · 0 0 0
0 32 3 2.
3
0 -1 1 1 0 0
0 as 9 7
I 0 -1 1
-1 a, 0 0 0 --132 2

1 0
Zj- Cj 0 0
13 -2I 0
2
-

From the computation of the first tableau of phase I we observe tbat


. h . W nstr11cl
3 2 is t e entermg vector and a4 is the departing vector. e co .
the second tableau according!y and see that iill. ( Zf~ 'i ) : .
non~~egative ~d · hence an optimal basic feasible solution to
auxiliary L.P.P. has been obtained. We see further that

Max. z• = O
and the artific. 1 l vel. }lePce.
th . ia vector ,a, appears in the basis at the zero e ,naY
· note optimal
be
basic fi 'bl ·
. easi .e solution to the auxiliary L. · ·
p p maY oret tbi'
a basic feas'bl · · ·
1 e solution to the ongmal pro e · bl rn 'fog
'
. SIMPLEX ALGORITHM . 225 ,'

·we pass on to the


. feasible solution to the given problem,
se II problem is
...,al basic w objective function to the pha
oP11 .,. J. fhe ne 0. x6 - O. ~ .
p~s5e l z;; zx, + x2 + x3 + 0. x4 + 0. x5 +
t ·to the given
~aximize this objective function subjec
.a:. n, if any, will be·optimal
we 10 ~e optimal basic ~e~ible solutio The last l>asis of phase I
co11stra s1'ble solution of the ongmal problem.
t,s5iC fea he inifial basis in phaIse. II and
the last tableau is used as the
.
ak en as t u with a change. on Y m the ob'~ect1ve row containing the
is t 1ea h d. d' .
.~u·at tab um ns are c ~g e . acc or m~ ly. Th e md:x row is _
in c and col
J = 3, 4, 6, smce these.
poces d to show ( zj- . cj} fo~ ·all J excI ept
8
. . phase . Hence 1t . obv1o
. 1s .
aJculate ere strictly pos1t1 .. m
ve· us that
c
vecto~; ~ will not enter the basis. From _the first tableau of phaimal
se U
83
'. a, n that z - c1 < 0 and .
hence this ·does not give the ·opt
·t is see . ' next •tabJe ( zi- .ci)
1
• a enters the b~ is and a 5 leaves• , , it. In the• • •
solu tion. 1
baste feasible soJubon ts
. are all non-negative and hence an optimal
attained.
PH AS E II : TA BL EA UX l AN D 2

c·J 2 1 I ·o 0 0 0

a.c as a,
Cg B b 31
2
32 33
l I
0
"'0 0-
I 82
4 I 2 6
3 3
0 85 9 w 0 ~l 1 1 0 0
·J
.0 a, 0 0 0 _Q
2
--2I 0 -1

_ ·-4 ''
0 0 o.
Zj- Cj .
~ 3
_ _! 0
I 82
10
0 1 21
2
-21
9
0
I .o.
81
7
1 ' 7
0 a, 0 0 0
,o 1
t-- .
4 0
.........__ Zj- Cj 0 0 '
21·
~

The optimal basic feasible solution is 64


10 ·
. '- - 9
x, - 7 'X2 ::: 21 'X3 = 0-and _ Zmax = 21 . .
r · 226
J.,INEAR PROGRAM¥1NG

.
· S Ive the following problem by two.phas
. Ex, 3. 0 • e 111eth0
· Maximize z = -Sxi + 3x2 · d:
·su~ject to 2x, + X2 $ 1,
3x1 + 4x2 ~ 12
and X1 , X2 ~ 0.

To put the problem in t~e standard fonn, we n


vana . ble xJ in the _fir~t constramt, one surplus variable x4 10 one
h sl!l., ~
constraint and one artificial. . variable
• . . x5 in the second constrt .e . -~ ~~
xi , xz , x3 , x4 are-the legitimate vanables and x5 is the an·fi atQt 'h...
. • . . I Icial .•11ua
We now fonnulate the problem for phase I. · v~
~
Maximize z• = 0. + 0. x, + 0. x3 + o, X4- X , 5
subject to 2t1 + x2 + x3 + 0. x4 + o-Xs== l
3x1 + 4x2 + 0. x3 - x4 + Xs:: 12.
'
We construct the initial tableau as under.
PHASE I : TABLEAU I
...........
Cj . 0 0 -o 0 - .1
Cs B XB b 81 a2 83 ~
- 8s
-
0 33 X3 1 2 QJ 1 0 0
-I as Xs 12 3 4 0., 1 1
Zj- Cj .-3 4 0 1 0
I

A"
s is seen, this table do . . .
i J
the usual criteria w · hes not give the optimal solution. FollowiJ
· ' e see t at 3 2 enters the b~sis and a3 leaves it.
:· PHASE I: TABLEAU 2
Cj 0 0 0 O·
Cs B XfJ b 31 82 83 a.
0 32 Xi I 2 I 1 0
- I.
85
8 -5 0 -4 -1
ZJ- C·
J 5 4 . 1.
0
- - - -· 4 - - - ·- --- -
- ----- - --- -

SIMPLEX ALGORITHM . 227

. last tableau Zj ~ ci ~ 0, for all j. Thus the optimal solution


the · . h M • . .
Jfl ttained. But we notice t at ax. z· = - 8 < 0. Moreover, the
~3si,e_en ~ariable x5 corresponding to the artificial _ vector a5 appears in :
~ificta~ at the positive level. Hence the· original· L.P.P. does not
asis .
tbe b y feasible so1ut1on. · · .
5sess
an .
p0 4 Use two phase method_to show that the following L.P.P. has
£%· • .
ed solution :
bound
un Maximize z == 2x,.+ 3xi + X3' x, 'Xi' X3 ·~ 0
subject to . -3x1 + 2xi + 3x3 = 8,
-3x 1·+ 4xi + 2x3 = 7.
We are to introduce here two ~rt~ficial variables x4 and x5 to th_e
constraint equations. The auxiliary objective function which is. to
two I. . b . ·
be maximized in phase IS· g1 ven y
_ Maximize z• = 0. x 1 + 0. Xi + 0. x3 - x4 - x5
subject to -3x, + 2xi + 3x3 + X4 = 8,
-3x 1 + 4x2 + 2x3 • + x 5 = ?,

xi ~ 0, j = l , 2 , 3 , 4 ·, 5 .

PHASE I

Cj 0 0 0 - 1 -1

CB B XB b a1 82 a3 a. as
-1 34 X4 8 -3 2 3 1 0
-1 as Xs 7 -3 [I] 2 0 1

Zj- Cj 6· - 6 -5 0 0
1
--1
~ 1
9 3
34 X4
2
0 2
2
0 . 7 3 1. I
0 I
32 Xi 4 2 4
1---_ 4
3
Zj- Cj
3 0 -2 0 2
i----:.___ 2
0 I - l4
83 X3
9 - l 0 1 2. ,
4 4

0 5 . 3
1 0 _! 3
....___ 32 X2
·s 8 4 8

Zj- Cj 0 0 0 0 0
>

LINEAR PRQGRAMMfNG

228 .
- to sta. n. W.tth
ctors a4 and a5 gi. ve the• initial basis
. The Ve (p re vi ou s p~ ge ), a 2 enters th · As. is Seen
. the phase . I fir st tab lea u . h e ba
111
tru ct th e p as e I se co nd tab lea . _srs llll(I
aves it. Then .w•e co ns . S h . u. As I \
1e lea ve s 0 t at In the final Phase . S see. n,~
ba sis
nters the . . . an d 34
.
1t.
. I
e . e ba sts . So we sta rt the h. · tabJb..
there is no art.if1c1 al ~e cto
.
r tn th
b . . . P ase II '"""
.
tio n to e m ax 1m 1z ed w ill be , fabJea
where the ob Je tt1 ve fu nc
n
z = 2x1 + 3x2 + X3. •
be th e sta rting basis ·
p h :d th~ ~
The las t ba sis of ph aS e I w ill
tableau of phase ~
iti al ~a b lea u of
r ~ WrtJi a
.w! II ac t as t~ e in
change in the ob3ect1 ve ro ~ w1 th th e .pn ce s 2, 3, .1 fo
1

1s ch an ge d ac co rd in gl y. 'Xi ;-ti ·
respectively. The 8 c
_ co lu m n .
.

.The ne~ in de x ro w sh ow in g ( z1 - c 1 ) in the initial tab

. As th er e ar e no ar tif ic ia l Ve cto rs lef~ •w~eau Of


phase II are computed delere
the correspon di ng co lu m ns .
PHASE II
2 3
--
.. c·J 1

b. 31 82 83
-
B Xa
Ca -
X3
9 -
- -43 0 1
l 33 4

-45 - -83 1 0
3 32 X2
'
- -31 0 0
z·1 -. c·J 8

at cl = - ~l < 0, so that al enters the basis. But, since


We see th Z1 -

e nega tive, the prob lem w ill have unboundei


both .Yu _and Y21 ar .
s~utioo.

You might also like