Visit : https://hemanthrajhemu.github.

io

Join Telegram to get Instant Updates: https://bit.ly/VTU_TELEGRAM

Contact: MAIL: futurevisionbie@gmail.com

INSTAGRAM: www.instagram.com/hemanthraj_hemu/

INSTAGRAM: www.instagram.com/futurevisionbie/

WHATSAPP SHARE: https://bit.ly/FVBIESHARE

 .
ot ‘ js Penal cs

Np scienNets ENS

. af
: j i 5
ahi
hy AMee
” fe
€' i 3:
hy *
>

. me mot ,
1 P / j
nn
FES ne _ tie tog " =
ee —_— oi i ie '

hit 0 eR ee, ae
Ul Gat ‘ Garfaes ‘ &
04 p ; ‘
Wao ih: ae
‘ i ‘ ipie

hy
‘be al Meatty , ’ reeks dy aihe i
1 ones TITroe aT At
4 Fe om RAE PALPei AC LA Wala

Zz ; i ye fy SN
7 "|
ier cee, et |, 2 at
Beer
vette)
oy bp 4 . Oy my rT
re _* 4s =e wit aye
Ties. acs ow ) oteres aU rrkemehy ‘ Pere EtJs “AP fal\FE veil
; Te ft < Nee eeeS” ne“ ‘aR aupd.Ary? -
os piaui
icsHat Cac‘aurea
oh Not a
Aa Jace = Oewor E
Fy
agus; é Oe
"i 44 SerSo
BOs.SoH
e ey
=9 r 4
PalF 4 + oh
eae b>
40 Dead c tots
my =Ae
gi eS ae Pets AE SA eT EM Tel ge
Snk
t r ¥ "Yre
4 re Lf FY ra r oO6. rar Tt Se Papas we
keg i
wg. ; a Yeee eee
pi Pieiy | hal Bs J
baa ‘
f OSs
iad é
hh H) ba‘ eyo MT
TA
‘bh , 4 hee Me Ms
‘ rey ee Mae
y
i> 1 BA FL h Peerigy
maam AY.Penne
metodo
mT, ye aeS s ey !
# ADYEi .
ee *<tr: ~~ .
2. y “i
; Abt ykaooe
wesleyoe LS

https://hemanthrajhemu.github.io
xiv
Operati'ONS RegSearey
i

2 Simplex Method - 1
2.1 Introduction
The Essence of Simplex Method 5]
Ww
WW

5]
Basic Terms / Definitions
iw
Nm

S2
Standard Formofan LP Problem (Characteristics of LPP)
in
NON

33
The Setting up and Algebra of Simplex Method
55
2.5.1 Steps of Simplex Methodin Brief
a7
2.5.2 Steps in Performing row Operations 58
2.6 Special cases of simplex method 78
2.6.1 Unbounded Solution 78
2.6.2 Tie Breaking in Simplex Method (Degeneracy) 80

2.6.3 Multiple optimal solutions 82

2.6.4 Infeasible Solution 83

84
Artificial Variable Techniques
N
~I

85
2.7.1 Big-M Method (Penalty Method)
86
2.7.2. Two Phase Method
101
ReviewQuestions
102
Problems

3 Simplex method - 2, Duality Theory 109

3.1 Introduction 109
3.2 The Essence of Duality Theory (Concept of Duality) 110
3.3 Economic Interpretation of Duality 111
3.4 Unrestricted Variables 11)
3.5 Key relationships between Primal and Dual problems ji1
3.6 Primal Dual Relationship 112
3.7 Characteristics of the dual problem
113
3.8 Advantages of Duality
j17
a9 Dual Simplex Method
129
3.9.1 Procedure of Dual Simplex Method
j30
Review Questions
Problems

https://hemanthrajhemu.github.io
 Tuble of Contents

Transportation an
d Assignment Prob
lems
4.1 Introduction
4.2 Formulation of
a Transportation
Problem
133
4,3 Initial Basic Feasible
Solution (IBFS)
134
4.4 Applications of Tr
ansportation Proble 134
ms
4.5 Optimality Check 134
4.6 Steps in solving a Tr
ansportation Proble 140
m
4.7 Variations (special Case
s) in a Transportation Prob
142
lem
4.7.1 Unbalanced Tran 152
sportation Problem
4.7.2 Degenerate Transp 153
ortation Problem
4.7.3 Restricted route (Prohibited 157
transportation problem).
4.7.4 Maximization Problem 160
4.8 The Assignment Problem 164
4.9 Applications of Assignment Problems 175

4.10 176
Algorithm for an Assignment Problem (Hun
garian Method)
4.11 Types of Assignment Problem 176
4.11.1 Unbalanced Assignment Pro 188
blem
4.11.2 Maximization Problem 188

4.11.3 192
Prohibited route (restricted assignment prob
lem) 197
4.11.4
Alternative Optimal Solution
4.12 Differences Between a Transportati 200
on Problem and an Assignment Problem
205
4.13 The Traveling Salesman Problem (Routing
Problem) 206
4.13.1 Formulation ofthe Traveling sales man
problem 206
Review Questions
210
Problems
211
5 Game Theory and Metaheuristics
5.1 Introduction
oan Basic Terms used in Game Theory
5.3 Formulation of Two Persons-Zero Sum Game
5.4 Properties (characteristics) of a Game
5.5 Assumptions made in game theory
5.6 Applicatons of Game Theory
5.7 Max. MinPrinciple
5.8 Min. max Principle

https://hemanthrajhemu.github.io
 simplex method - 2, Duality Theory

/
ie ;
Learning objectives
=.
After Studying this chapter, you should be able to
© appreciate the significance of duality concept
| © understand the relationship between the primal and dual LP
| problems
© convert a primal to dual problem and vice-versa
© interpretation of the duality
( © use dual.simplex method to solve LP problems i

3.1. INTRODUCTION
In the preceding chapters we have seen how complex and
cumbersometheiterative procedure in the simplex method can be. Every time a variable departs and
new variable enters, complicated calculations requiring large space (computer memory) have to be
performedto get the new element values. Successive iterations are carried out by using number of
row operations using simplex method and its variants (Big M method and Two Phase Method.
If a LPP with large number of variables and constraints (whichis in real life problems) has
to be solved by the simplex method, it will need a lot of time and computer space asall the tables
and data will have to be stored and retrieved repeatedly. In this chapter we discuss about a new
approachand variant of simplex method duality theory and dual simplex method.

3.2 THe Essence of DuAtity THEORY (Concept oF DUALITY)

The term ‘dual’ in a general sense implies two or double}'In LPP duality implies that each problem
can be analyzed in two different ways but having equivalent solutions. Each LP problem stated in
{09

https://hemanthrajhemu.github.io
Dr
 110 Operations R esearch
et

the original form is associated with another LPP called dual LPP or in short dual}.The two problems
are replica of each other, The dual ofthe primal problem is unique. The simplex procedureis such
that if the primal is solved it is equivalent to solving the dual.) Thus 4f the optimal solution to dual
can be obtained, if we knowthe optimal solution to primal, |

3.3 Economic INTERPRETATION OF DUALITY

In order to make the concept of duality clear, let us consider the following problem on diet.
Let the following table give the amounts of two vitamins V, and V, per unit present in two
different foods f, and f, respectively.

Food
Vitamins a I, Daily requirement
Vy 6 8 100
V 7 12 120
Cost per unit 12 20

Thelast column of the table represents the numberof units of the minimum daily requirement forthe
two vitamins whereas the last rowrepresents the cost per unit for the two foods. The problemis to
determine the minimum quantities of the two foods f, and f, so that the minimum daily requirement
of the two vitamins is met and that at the same time, the cost of purchasing these quantities of f,
and f, is minimum.
To formulate the problem mathematically let x, be the numberof units of food f, (J = 1, 2) to
be purchased, then the above problem is to determine two real numbers X, and x,, so as to
Minimize Z = 12x, + 20 x, subject to the constrains
6x, + 8 x, 2 100, 7x, + 12 x, 2 120, x,, x, 20 :
Nowlet us consider a different problem, associated with the above problem. Supposethere is
a wholesale dealer who sells the two vitamins V, and V, along with some other commodities.
The local shopkeepers purchase the vitamins from him and from the two foodsf, and f, (the details
are same as in the given table). The dealer knows very well that the food f, and f, have their market
values only because of their vitamins contents, the problem ofthe dealer is to fix the maximum per
unit selling prices for the two vitamins V, and V,, in such a waythat the resulting prices for food
f, and f, do not exceed their existing marketprices.
The selling prices are generally known as dual prices or shadowprices.
To formulate the problem mathematically, let the dealer decideto fix up the twoprices at W,
and W,per unit respectively.
Then the dealer’s problem can be stated mathematically as to determine two real numbers
W, and W, soasto

https://hemanthrajhemu.github.io
—
 WH
Simplex method - 2, Duality Theory

Minimize Z* = 100W, + 120 W,, subject to the constrains

6W, + 7 W, < 12, 8 W, + 12W, < 20, W,, W, > 0.

3.4 UNRESTRICTED VARIABLES

Usually in an LP problem, it is assumed that all the variables x, (1 = 1, 2..........., n) should have
non — negative values. In many practical situations, one or more of the variables can have either
positive, negative or zero value. Variables which can assume positive, negative or zero value are
called unrestricted variables. Simplex method requires that all the decision variables must have non
— negative valuc at each iteration, therefore, in order to convert an LP problem involving unrestricted
variables into an equivalent problem having only restricted variables, we have to express each of
unrestricted variable as two difference of the non — negative variables.
Let variable x_ u be unrestricted in sign.
g
We defined two newvariables say
y
x'u and x'!a such that
- = X,
X= yl _yll. yl x
xi; xi, yll 20

3.5 KEY RELATIONSHIPS BETWEEN PRIMAL AND DUAL PROBLEMS

Weak duality property: If 'x' is a feasible solution for the primal then and yts a feasible solu-
tion for the dual problems then, Cx < yb, this inequality must hold good for anypair of feasible
solutions of primal and dual.
Weak duality property describes the relationship between anypair of solutions for the primal
and dual problems, where solutions are feasible for their problems.
Strong duality property : If x* is an optimal solution for the primal and y* is an optimal solution
for the dual problem, then Cx* = y*b.
[The relationship imply that Cx < yb for feasible solutions if one or both of them are not
optimal for their respective problems, where as equality holds when both are optimal]
Complimentary solutions property: At each iteration, the simplex method simultaneously identi-
fies a CPF solution x for the primal problem and a complimentarysolution y for the dual problem.
In Cx = yb if x is not optimal for primal problem, then is not feasible for the dual problem.
Complimentary optimal solutions property: At the final iteration the simplex method simultane-
ously identifies an optimal solution x* for the primal problem and a complementaryoptimal solution
y* for the dual problem. The relationship can be expressed as Cx* = y*b

3.6 Prima DUAL RELATIONSHIP

Let the primal problem be
Max Z = Cc, x C, Koch. S n

https://hemanthrajhemu.github.io
 Operations Research,
12 Sa,

; ' : coat
Subject to a), X, +4), %, 4.7 aX, S b,
a,, X, + aX; +-.-ta,,%, Sb,

. ae .
Ba x, + 4.3 X; ose ra, <,,

Xp» x, +++ x >0

- n

Dual of the problem is defined as (let w, w, w, be the dual variables)

Min Z* = b,W, + by Wy + eeeeseeeeeees+bow,
Subject to 8, Wy tay Wy F evceececenes ta wi2c,
. ' +. : ae '

Big Wy + Bay Wy Heeesrccscceeeeseeneens +a w_2>C,

BW FB, Wy Teneceeeceeneceesseones tal wi2c,

Wye Woy cneneeeseeeecees w_>0

Where, w,, W,... W 2 0 are called dual variables

Note: Jn general Dual ofthe dual is primal.

3.7 CHARACTERISTICS OF THE DUAL PROBLEM

Duality in linear programming has the following major characteristics
i © Dual ofthe dual LP problem is primal,
ii. If either the primal or dual_of the problem has finite optimal solution, then the other
one also will have the same’
iii. (1f anyofthe two problemshas only an infeasible solution, then the value of the objective
function of the other is unbounded.
tv, ‘The value ofthe objective function for any feasible solution of the primalis less than
the value of the objective function for any feasible solution of the dual->)
v. air either the primal or the dual has an unbounded objective function value then the solu-
tion to the other problem is infeasible,
vi. df the primal has a feasible solution but the dual does not have, then the primal will not
have a finite optimal solution and vice-versa.)

https://hemanthrajhemu.github.io
 Simplex method- 2, Duality Theory 113

3.8 ADVANTAGES OF DUALITY

SS

Following are the advantages of duality.

i. It yields a number of useful theorems.
li. Solution of the dual checks the accuracyof the primal solution for computational errors.
iil. Indicates that fairly close relationship exists between linear programming and duality.
iv.. Economic interpretation of the dual helps the management in making future decisions.
v. Computational procedure can be considerably reduced.
vi. Duality can be used to solve an LP problem by the simplex method in whichthe initial
solution is infeasible.

Worked Examples
7) Write the dual of the following LPP
Max Z =x, + 2x, + x,
Subject to 2x, + %,-X,52
— 2x, + x,— 5x,2-6
dx + %,+ %, 5 6X, X,,%,20
Solution:
Max Z =x, + 2x,+ x,

Subject to 2x, +x,-x,<2

Note: Put the second constraint in the standard form (< form)
2x, —X,+ 5x, < 6, 4x, + x, + x,< 6 and x,, x,, x,20
Dual; Let w,, w,, w, be the dual variables

Then Min Z* = 2w, + 6 w,+ 6w,

Subject to 2w, + 2w, + 4w,> 1, w, —w,+ w,2 2,-w, +5 w y+ w,? and w,, W,,Ww, > 0

2. Write dual of the following LPP

Max Z = 3x,—x,+ X;
Subject to 4x, — x, < 8, 8x, + x, + 3x,2 12, dx, — Gx,< 13 and x,, x,, x,2 0

Solution:
Put the second constraint in the standard form (< form) i.e., — 8x, — x, — 3x, $12
Dual; Min Z* = 8w, — 12 w,+ I3w,

Subject to 4w, — 8 w, + 5w, 2 3, - w, — w, + Ow, > — 1 and Ow — 3w, + 6w,2 |

https://hemanthrajhemu.github.io
 il
Operatians Research
———Seareh

3. Write the dual ofthe following LPP

Max Z = 6x, + 10 x,
Subject to x, < 14, x, S 16, 3x, + 2x, < 18 and
Xx, X20

Solution:
Let w,, w, and w, be the dual variables
Min Z* = I4w + 16 w, + 18w,
Subject to w, + 0 w,+3w,>6
Ow, + w+ 2w.> 10 and Wi.
W., Ww, 20
4. Write the dual of the following LPP
Min Z = 0.4x, +0.5 x,
Subject to 0.3 X, + 0.2 x, 2.7, 0.5 x, +
0.5 x, =6
0.6 x, + 0.4 x, 26 and x,, x,2 0

Solution:
Max Z*= 2.7w, +6 w, + 6w,
Subject to 0.3 Ww, + 0.5 w,+ 0.6 w, 5 0.4
0.1 w+ 0.5 W,+ 0.4 w, < 0.6
W, S 0, w is unconstrained asit is equation,
w, > 0
5. Construct the dual problem for the following LPP
Maximize Z = 16x, +14x, + 36x,+6x,
Subject 14x, + 4x, + 14x, +8x, = 21: 13x, +17x, +80x,
+ 2x, < 48
Xj, X2 20; x3; X; to unrestricted

Solution:
Dual problem for the given primal problem
Minimize Z*=21 w,+48w,
Subject to
l4w, + I3w, 2 16,4w, + l7w, 2 14

l4w, + 80w,= 36, 8w, + 2w.= 6

W, is un-restricted,w, > 0
The above dual can also be solved by graphical method where
as primal can be solved using
only simplex method due to more than 2 variables.

https://hemanthrajhemu.github.io
—lex
Simp —method - 2, Duality Theory Hs

6. Obtain the dual of the following LPP

co 3x, + 4x,
x, x, 20
Subject to 2x, + 6x, s 16, 5x, + 2x, 2 20;
Solution:
Since the objective function of the given LP problem is of maximization, the direction of each
of ‘>’ has to be changed.
Multiplying the second constraint on both sides by —-1, we get
(-1) (5x, + 2x,) 2 (-1) (20) or
—5x, —2x, < -20
The standard primal LP problem so obtainedis:
Zw = 3%, + 4X,
Subject to the constraints
2x, + 3x, < 16, 5x, -2x, < -20 and x, 2 0, x, 2 0.
Let w,, W, are the duals then, the dual is
Zu = 16w, — 20w,
Subject to the constraints
2w, — 5w, 2 3, 3w, — 2w, 2 4 and w, 2 0, w, 2 0.

‘7. Write the dualforthe following LPP

Zax = 8X, + 5X, + Tx;
Subject to the constraints
%, + x, + 3x, < 10, 4x,-x, + 2x, 215
X,» X, = 0; x, is unrestricted variable.

Solution:
Zin = 1OW, — ISw,
Subject to the constraints
w,- 4w, 25, w, + w,2 5
3w — 2w, = 7, — 3w, + 2w, 2-7 or
3w, — 2w, < 7, w,, W, 2 0.
0 ||

Obtain the dual problem of the following primal LP problem.

Maximize Z = 40x, +120x, subject to the constraints, x,— 2x, < 8, 8x,+ 5x, = 90, 15x,
+ 44x, < 660,
x,2 0,x,20

https://hemanthrajhemu.github.io
 116 Operations Researey,
—

Solution:
Transform the ‘<* type constraint to a ‘>" type constraint by multiplying the constraints
by —1. Also write *=" type constraint equivalent to the constraint of the type “2’ and *<”’, then
the given primal problem can be written as,
Z__ min
= 40x, +120x,
«

Subject to the constraints

—x,+ 2x, 2-8, 3x,+ 5x, 2 90
3x,+ 5x, = 90 or —3x,— 5x, 2 -90
—15x, — 44x, 2 -660 and x,> 0, x2 0
Let w,, w,, W, and w, be the dual variables corresponding to the four constraints in given
order, then the dual of the given problem can be formulated as:
Z_., = 8w, + 90w, — 90w, — 660w,
Subject to the constraints
—w, + 3w, — 3w, -— 15w, < 40

2w, + Sw, — Sw, — 44w, < 120

w,2 0. w,2 0, w,2 0, w,2 0

Let w = w,- w,, the above dual problem reducesto the form.
Z_.. = 8w, + 90w — 660w,

Subject to the constraints

—w, + 3w- 15w, 40

2w, + Sw — 44w, < 120

w, > 0. w, > 0; wis unrestricted in sign.

Here, it may be noted that the second constraint in the primal is equality. Therefore the
corresponding second dual vanable w, should be unrestricted in sign.

3.9 Duat Simpcex MetHop

In simplex method, if one or more solution values (that is x,) are negative and optimality condition
is satisfied, then the solution may be optimum but not feasible [as it must satisfy the non negative
constraint]. In such cases, a variant of the simplex method called the dual — simplex method would
be used. In the dual simplex method we always attempt to retain optimality while bringing the
primal back to feasibility (that is x, > O for all i).

https://hemanthrajhemu.github.io
Simplex method - 2, Duality Theory 7
—_—

3.9.1 Procedure of Dual Simplex Method

Step 1
Rewrite the linear programming problem by expressing all the constraints in < form and transform
them into equations throughslack variables.

Step 2
Express the above problem in the form of a simplex tableau. If the optimality condition is satisfied
and one or more basic variables have negative values, the dual simplex method is applicable.

Step 3
Feasibility condition: The basic variable with the most negative value becomes the departing /
leaving variable (LV). Call the row in which this value appears as pivot row. If more than one
element for LV exists, choose one.

Step 4
Optimality condition: Form ratios by dividing all C, -Z values by the corresponding pivotal row,
the entering variable is the one having mminimumratio. If no element in the pivot row is negative,
the problem has nofeasible solution . c Y= O-k ~otey
-
mi FV
CE inthe cre jer & © ges
Step 5
Use elementary row operations to convert the pivot element to | and then reduce all the other
elements in the key/pivot column to zero.

Step 6
Repeat steps 3 through 5 until there are no negative values for the basic variables.

Note: The advantage ofdual simplex methodis, it can avoid theartificial variables introducedin the
constraints along with the surplus variablesas all ‘> constraints are convertedinto ‘<"type.

9. Use dual simplex method to solve the following problem.

Maximize Z = — 2x, - 3x,
Subject tox, +x, >2,2x,+x,< 10andx,+x,<8
with x, and x, non negative

Solution:

The given problem is of maximization type hence,it is in standard form. Converting ‘>’ as
‘<" of first constraint.
“XX, — X, 2-2

https://hemanthrajhemu.github.io
 a
Operations Research
—_—__ ee

Converting the constraints as equations,

—X,—X,
. .
+ 7 U, =--%14
=~ 2 where
}
u, u,, u, are slack variables
‘

2x, + x, + u,= 10

x, +x, tu, =8

Modified objective function is,

Zmax = ~ 2X, ~ 3X, + Ou, + Ou, + 0u,

First Simplex Table

EV 1
x x, u, u, u, B
Basis C,

[-1] PE ~] 1 0 o |-2<_PR
LV+u,| 0

“2 © 7 1} Oo 1 0 |10
1 l 0 0 g
u.| 0

C, -2 -3/0 oOo 0 |-

Z, 0 0}/0 oO 0 fo
-2 -3 0 0 0 |-
G-Z
T
PC
Since all the (C — Z) values are < 0, the above solution is optimal. However,it is infeasible
because it has a non positive value for the basic variable u, (= — 2). Since u, is the only non
positive variable, it becomesthe leaving variable (LV).

Table to Obtain the Ratio

x, x, uy u, uM,
C-Zrow -2 =. 0 0 0
Pivot row =| =i l 0 0
Ratio 2 3 an - -

As x, has the smallest ratio, it becomes the entering variable (EV)

Thus, the clement — 1, is the pivot element. Using the elementary row operations, we obtain
second simplex table.

https://hemanthrajhemu.github.io
Simplex method - 2, Duality Theory 119
—_

Second Simplex Table

Basis Cc, =, x, uy My, B

x, —2 l l — 1 0 0 |2

u, 0 [o]pe -1| 2 0 |6—PR

u, 0 0 0 1 0 1 6

5 —2 =3]) © 0 0 |-
Z, —2 —2 2 0 0 |-4
Cc -Z 0 -1]-2 0O o i}-

Since all the (C — Z) values are < 0, the above optimal! solution is feasible. The optimal and
feasible solution is x, = 2, x, = 0 with Z.. = - 4.

10. Use dual simplex method to solve the LPP

Zax = 2X, — x, subject to the constraints
bet x, — X25
Me me BKgyt 2x5 28

Solution:
Since the objective function of LP problem is of maximisation, therefore all the constraints should be
of < type. Thus convert the constraints of the < type by multiplying both sides by —1] and re-wmitting
the LP problem.

Zinax i —2x, + 0.x, _ x,

subject to the constraints:

—X, =X, tx, 5-5
—x, + 2x, — 4x, $-38
X,» Xp X, 2 0
Converting the constraints into equations by adding slack variables we get,
Lew = 72%, + Ox, — x, + Ou, + Ou,

subject to the constraints

-xX,—-X, +x, +u,=—-5

-x, + 2x, — 4x, + u,=-8

X. XX,» Uj,» u, 20

https://hemanthrajhemu.github.io
 Operations Research
——__.,

Simplex Table - 1

Basis Cy x, x, x. u, u, B
u, 0 -l -] ] I 0 -5
PR«L_: 0 -l 2 —-f 0 I -S L. pr
C -3 0 -l 0 0
Z 0 0 0 0 0 0
C-Z -2 0 -l 0 0

The solution is infeasible as u, = —5 and u, = —8, but it is optimal as all C, - Z, < 0. Thus we need to
apply the dual simplex method to get both feasible as well as on optimal solution.
Iteration - 1

u, = -8 is the most negative valueand hence u, leaves the basis.

Table to obtain the ratio.

x, x, x, u, u,
C-Z row —2 0 —] 0 0
Pivot row 1 2 4 0 1
Ratio 2 - WA = -
As x, has the smallest ratio, it becomesthe entering variable (EV). The intersection element of PR and
column corresponding to EV is PE = 2
Making the PE as unity and other element of PC as 1 bysuitable row operations we get
Simplex Table - 2

Basis G. x, x, a u, u, B
u, 0 5/4 -1/2 0 i 1/4 -7
Ij %, -1 1/4 -1/2 I 0 -1/4 2 te—PR
C -2 0 -] 0 0
Zz -1/4 1/2 —I 0 1/4
C-Z -7/4 -1/2 0 0 -1/4

Still the solution is infeasible as u, = -7, but optimal as all C, ~ Z, <0.

So we proceed to seconditeration.
Iteration - 2
Since, the variable u, = —7 is the only variable having negative value, we should select the
variable u, to leave the basis.

https://hemanthrajhemu.github.io
 oo
y Theory
Simplex method - 2, Dualit
——

Table to obtain the ratio

x, x, - u, Mu,
C ~ Z row -7/4 -I/2 0 0 -1/4
Pivot row -5/4 —-I/2 0 ! 1/4

Ratio 35/4 / — - =

The least ratio is | which corresponds to x, and hence x, enters into the basis. Making the
corresponding PEas | in the Simplex Table - 2 we get,
Simplex Table - 3
Basis C, Cs x, Xx, u, u, B
Ne 0 I/2 1 0 -2 -1/2 14
x, —I 3/2 0 I -l1 1/2 9
C —2 0 —I 0 0 _
Zz —3/2 0 -1 I 1/2
C-Z —1/2 0 0 -I ~1/2

From the abovetable, all C — Z < 0 and also all the basic variables are positive. Hence, the
solution is feasible and optimal.
Thus, x = 0, x5> 14,x,=9

Zam = 2 (0)-9=-9
‘11. Solve the following LPP using dual simplex method.
Zin = 8X, + X, Subjectto:
m.

x,+x,21, 2x, + 8x, 22, x, and x, 20

Solution:
| Converting the minimisation problem to maximization problem and the constraints of = type
to < type weget,

Zin = —3x,- %
subject to —x, —x, S—l, —2x, -3x, < -2
x, and x, 20
Adding the slack variables to the constraints,
Zmax = —3x, — X, + ou, + Ou,
Such that -x, — x, + u, = —l
—2x, — 3x, + u, = -2
XX U,, U, 2 0

https://hemanthrajhemu.github.io
 Operations Research
—t

Simple Table - 1

Basts % x,
1
x, u, u, B
; 0 -] —] 1 0 -I
LV< 0 -2 -3 0 ] —2 |«—PR
C -3 -] 0 0 -
Z 0 0 0 0 0
C-Z —3 —1 0 0 =

From the table, we can notice that the solution is infeasible as u, = —1, u, = —2 but it is op-
timal as all C — Z< 0. Thus, we need to apply dual simplex method to get both feasible as
well as optimal solution.
u, is the leaving variable as it has most negative value. To determine the entering variableit
is required to obtain theratio.

OS
C - Z row
Pivot row

a
Ratio 3/2 1/3
The least positive ratio is 1/3 which corresponds to column 2 and hence x, enters into the basis.
Replacing u, with x, weget,
Simple Table - 2
Basis C, x, Xe u, u, B
Ive ou, 0 -1/3 0 I -1/3 -1/3 |}*«—-PR
: -l 2/3 ] 0-1/3 2/3
C 3 -I 0 0
Z -2/3 -l 0 1/3
C-Z -7/3 0 0 -1/3

Solution is still infeasible as u, = —1/3 but optimal so proceedingto the next iteration.
Asu, is non - positive it becomes the LV,to determine the EV determinethe ratio as shown
below.

x, x, u, u,
C - Z row -7/3 0 0) -1/3
Pivol row -1/3 0 I -1/3
Ratio 8/3 - - 1

https://hemanthrajhemu.github.io
 123
ory
Simplex method- 2, Duality The
—_—_—

The min. ratio is 1 which corresponds to Column IV hence, u, becomes the entry variable.
Replacing u, with u, making PE as | in the Simplex Table - 2 we get,
Basis C, A ts u, i, B
u, 0 I 0 -3 1 I
x, -I ] -! 0 I
C -3 -1 0 0 -
Z, I -I 1 0
C-Zz —2 0 -1 0
As all C — Z < 0 andthe variables (of basis) are positive the solution is feasible and
optimal.
Hence, the optimal solution is
x, =0,; x,=1
7—=4
Note:
i. The least ratlo may corresponds to the slack variable also, in such cases the slack
variables becomesthe entering variable replacing the existing variable in the basis.
ii. In dual simplex method, the solution will be declared as final only when it is optimal
and feasible and i.e., only when all C, - Zz, < 0 and all the variables of the basis are
positive.

12, Solve the following LPP using dual simplex method.

ti = 2x,+ Xs

Subject to 3x, +x,23

4x, + 3x, 26, x, + 2x, 23 and x, x, 20.

Solution:
Converting the given objective function to maximization type
Zaan TO 2X — X
Converting all the constraints into < type.
—3x, -x, 3-3
— 4x, — 3x, s— 6.
—x,-2x,$-3 and x,,x,20.
Introducing the slack variables we get,
Zmax = — 2x, — x, + Ou, + Ou, + Ou,

https://hemanthrajhemu.github.io
 Operations Research
—.

Subject to = -3x, — x, tu, = -3

—4x, — 3x, + u, = -6
-x,— 2x,+u,=—-3 and x,, x, U,, u,, u, 20
Simplex Table - 1
EV
t
Basis ¢, x, x, u, u, uy B
u 0 -3 -I 1 0 0 —3
lv<u, 0 —4# pE| —3] 0 1 0 -6 |< PR
u, O —I -2 0 0 1 -3
C —2 -] 0 0 0
Z, 0 0 0 0 0 0
C-Z -2 -1 0 0 0 -

As all C — Z< 0, the solution is optimal. However, it is not feasible asit is having non-
positive basic variables.
The most negative basic variable is u, and henceit will be leaving variable.
Table to obtain the ratio.

x, = u, u, u,
C - Zrow -2 —] 0 0 0
Pivot row + —3 0 I 0
Ratio 1/2 1/3 - - -

As x, has the smallest ratio, it becomes the entering variable.

The clement — 3 is the pivot element, using the elementary row operations replacing u, with
X, in simplex table - 1, we get,
Simplex Table - 2
EV

Basis Ge x, x uy u, u, B
IVa ou, 0 pe|-5/3 0 I -1/3 0 -1 |<—PR
x, -1 4/3 I 0 -1/3 0 2
u, 0 5/3 0 0 —-2/3 I 1
C -2 -1 0 0 0
Z -~4/3 -1 0 1/3 0)
C-Z -2/3 0 0 -1/3 0

https://hemanthrajhemu.github.io
simplex method - 2, Duality Theory 125
50

Since all C, — Z, < 0 the solution is optimal but u, = —1 is the only negative value hence u,
becomes the leaving variable.

Table to obtain the ratio to decide the entering variable

x, x, u, u, u,
C,-Zrw -2/3 0 0 -I/3 O
Pivot row -5/3 0 1 -1/3) (OO
Ratio 5 = 0 1 -
x, is the entering variable asit's ratio is least/minimum.

Simple Table - 3
Replacing u, with x, and making PE as | by rowoperations
Basts Cy x, Xx, u, u, u, B
x, —2 1 0 —3/5 1/5 0 3/5
ms -l 0 I 4/5 -3/5 0 6/5
u, 0 0 0 I -1 1 0
C —2 -l 0 O 0
Z, —2 -1 2/5 1/5 0
C.— Z, 0 0 -2/5 —-1/5 0
Sinceall C, — Z, < 0 and the variables of the basis > 0 (non-negative) the solution ts optimal
and feasible.
Hence, x, = 3/5, x, = 6/5
Zax = 12/5 or f= 12/5 is the solution.

13. Use dual simplex methodto solve the following problem.

Minimize Z = 2x, + x, + 3x,
Subject to x, — 2x, + x, > 4, 2x, + x, + x,<8,x,-x, 20
and with all variables non negative.

Solution:

Asthe given problem is minimization, convert into maximization (standard form)

max = 2x, ~ Xx, —_ x,

Expressing the constraints in < form and adding the slack variables; the problem becomes
a Ou,
Zmax — 2x, = Xx, — 3x, a Ou, + Ou,

Subject to — x, + 2x, — x, tu, =-

https://hemanthrajhemu.github.io
 126 Operations Researeh
rc,

2x, +X, +%, +u,=8

-x,+x,+u,=0
With all variables non negative.

First Simplex Table

LEV
Basis C,, x, x, x uo ont B

iveu,| 0| FijpE 2 -1]}/1 0 0|-4e PR

u.| 0 2 1 1/0 1 O]8

u,| 0 -1 0 1/0 0 1/0
c -2 -I -3/0 0 0|-
Z, 0 0 o0|0 0 olfo
C=Z a2 =4 =-3)0 0 G |=
T PC
Sinceall the (C, - Z) values are < 0, the above solution is optimal. However, it is infeasible
because it has a non positive value for the basic variable u, (= — 4). Since u, is the only
negative vanable, it becomes the leaving variable (LV).

Table to obtain the ratio

x x x, u
C -Z.row| -2 -3 0
Pivot row -] -] l 0
Ratio 2 - 3 - = =
As x, has the smallest ratio, it becomes the entering variable (EV).
Thus, the element — 1 becomes the pivot element. Using the elementary row operations, we
gel,
Basis Cc. x, xy % u, ut, uy B

a -2|{1]peE -2 1] -1 0 O —PR4

MN, 0 0 5 -1] 2 1 0 0
u, 0 0 -2 2 /]-1 0 1 4
c -2 -1l -3 0 0 ~
Z, -2 4 -2] 2 0 0 -8
C-Z 0 -5 -] -2 0 0 =

https://hemanthrajhemu.github.io
 is
ory
Simplex method - 2, Duality The
—_
f

smnice all the variables have non negative values (x, = 4, u, = 0, u, = 4), the above solution
is optimal and feasible. The optimal and feasible solution is, x, = 4, x,= 0, x, = 0 with
Z min. = 8
[Substituting the values of x,, x, and x,, in the original objective function]

14. Use dual simplex method to solve the LPP

Max. Z = — 3x, - 2x,
Subject tox, +x,>1,x,+%x,<7,x, + 2x,>10,x,<3andx,x,>0

Solution:
Interchanging the = inequality of the constraints into <, the given LPP becomes.
Max. Z = — 3x, — 2x,
Subject to—x,-—x,<- 1
ot Xo 7

—K, ZX, 5 10

x, S3
Byintroducing the slack variables u,, u,, u,, u,, the standard form of the LPP becomes
Max Z = — 3x, — 2x, +0u, + Ou, +Ou, +0u,
Subject to — xX, —X, + u, =—-
X, +x, +u,=7
~4 $ 2x, +t: 10
Ox, +x, +u, =3

First Simplex Table

LEV
Basis C, *, x; uy Wy ou, ou, B
u,{| O |-1 -1 1 0 0 0 _]
u,}| O I l 0 1 0 0 7

Lv u,| 0 |-1 E2]PE] 9 O 1 © ~10~—pR

u,| 0 0 l 0 0 oO 1 3
Cc, -3 -2 0 0 0 0 _
Z, 0 0 0 Oo 0 90 0
Cc -Z, —-3 -2 0 0 0 0

https://hemanthrajhemu.github.io
- 128
ee
Operations Research
—r

Since all (C,— Z,) < 0 the solution is optimal but as x,, < 0 it is not feasible. u, = — 10 is the
most negative, the basic variable u, leaves the basis,
Table to obtain the ratio
[ x, x, u ny uy u,
—3 —2 0 0 0 0
C-Z.row|
Pivot row - | 2 0 0 0
Ratio 3 1 - - ~ ~

As x, has the smallest ratio, it becomes the entering variable (EV). The element — 2 is the
pivot element, using elementary rowoperations, we get,

Second Simplex Table

LEV

| Basis C. x, X, ua, u, u, B
u| O -1/2 O} 1 O -1/2 044
u,| 0 1/2 0 |0 1 1/2 0} 2
x,| -2 1/2 | 0 O -1/2 045

LV—u,| 0 Fajpe 0 |0 0 1/2 1 |-2—PR

c -3 -2}0 0 0 0 |-
Z -1 -2)/0 0 0 0 |-
C-Z -2 0|}0 O -1 0|-
APC
u, = — 2 is only the negative value. Hence, u, becomes leaving variable x, enters or is
entering variable.

Table to obtain the ratio

x x, u, u, u, u,
C - Zrow ae 0 0 0 md 0
Pivotrow -— 1/2 0 0 0 1/2 0
Ratio 4 0 = - = 7

https://hemanthrajhemu.github.io
 - 2, Duality Theory
Siimplex method
eT
129

ble
Third Simplex Ta

Basis Co “1 *) My Hy u, uy B
ut, 0 0 0 1 0 -1 -4 19

u, 0 0 01lo 4 \ t lo

x, -3| [iJrpE OJ]0 0 -1 ~-2]|4epr

C, —3 -21/0 0 oO 6 f-
Z, -3 -2/0 0 3 4 |-18
C-Z 0 0/0 0 -3 ~-4]-
As all C,— Z, < 0 and all x,, > 0, the solution is optimal andfeasible.
The optimal feasible solutionis,

Max Z = — 18 with x, = 4, x, =3

Review QUESTIONS
Give the characteristics of a dual problem.
S

Mention the advantages of duality.

NSN

Explain the economic interpretation of duality with an example.

kW

Explain the relation between the solution ofprimal and dual.

Briefly explain economic interpretation of duality.
Nw

Show that dual of dual is primal.

Hint Consider an LPP, write its dual and again write the dual of the obtain dual. This dual
will be same as that of the original LPP considered in the beginning.
Define duality in LPP and write important characteristics ofit.
Explain the following:
i. The essence of duality theory ti. Primal—Dualrelationship
Explain upper bound technique used to solve a LPP.
10. Write a note on
t) Weak duality property 11) Strong duality property
tii) Complementary solutions property iv) Complementary optimal solutions property

https://hemanthrajhemu.github.io
 130 Operations Researey
arch

PROBLEMS
i. Write the dual of the following LPP
7. =3x,-%, TX,
mana

Subject to the constraints 4 x, — x, 5 8, 8 x, + x, + aH> 12 and 5 x, -6x,< 13

Z*_,, = Sw, - 12w, + 13 w,

Subject to the constraints
4w,-8w,+ 5 w, 23, -w, -w, + Ow, 2-1, w, - 3 w, +6 w, 2 1 and w,, w,, w, > 0
Construct the dual of the LPP
Zoe = 4X, + OX, + 18 X,
Subject to the constraints x, + 3 x, 2>3 and x, +2x,25
Z* = Fw, +5 w,
Subject to the constraints
w, < 4,3 w,tw,s 6, 2 w,s 18 and WwW, W, >0

Z_, = 3x, + 17x, + Px,

Subject to the constraints x, — x, + x, 23 and—3x,+2x,<1
Write the dual for the above LPP
Z* = -3w, +,
Subject to the constraints w,— 3 w, 2 3, — w, 2 17, w, +2 w, 2 9 and w,, w, 2 0
Write dual of the following LPP
Zn = OX, + 10x,
Subject to 3x, + 2x, < 18, x, < 14, x, 5 16 and x,, x, 20
Z*min = Il4w, + 16w, + 18w,
Subject to w, + 3w, 2 6, w, + 2w, 2 10 and w,, w,, w, 2 0
Use dual simplex method to solve Max. Z = - 3x, - x, subject to x, +x, > 1, x, + 3x, 2 2, X)»
x, > 0.
Max Z=-1,x,=0,x,=1]

Using dual simplex method solve the following LPP

Min Z = 10x, + 6x, + 2x,
Subject to —x, + x, + x, > 1, 3x, + x, — x, 2 2, X,, X, x, 2 0
x= 1 /4,x,= 5/4,
%= 0,2. = 10

https://hemanthrajhemu.github.io
 hod - 2, Duality Theory
Simp lex met 137

7, Use dual simplex method and solve

Max. Z = — 2x, — X,
Subject to x, + x, — X, 2 5, x, — 2x, + 4x, > 8, x), x, x, 20
Ans: x,=0, x, = 14,x,=9andZ=-9

8. Using dual simplex method solve the LPP

Maximize Z =—3 x, -2x,

Subject to x, + x,21,x, +x,27,x,+2x,2 10,x,23 andx,,x,>0

Ans: x,= 4, x,= 3 and Z,=-18

9, Using dual simplex method solve,

Zain
m
= 2% + Xy
Subject to 3x, + x, > 3, 4x, +3 x, > 6, x, +2 x, > 3 where x,, x, > 0.

Ans: XxX} 2 X> == and Z min _ 12

5 5 5

10. Find the maximum of Z = 6x, + 8x,

Subject to 5x, + 2x, < 20, x, + 2x, < 10, x, and x, 22

by using dual simplex method
Ans: x, = 5/2, x, = 15/4 and Z= 45

https://hemanthrajhemu.github.io