Operations Research(MA30014)

Geetanjali Panda
Department of Mathematics
Indian Institute of Technology, Kharagpur, India
Syllabus
Role of Convexity in Linear Programming
Theory of simplex method
Two phase simplex Method
Dual Simplex method
Duality theory
Linear Integer Programming
Transportation Problem
Assignment Problem
Game Theory

1/22
Books:

1. Linear and Nonlinear Programming by David G Luenberger

2. Operations Research: An Introduction by Hamdy A Taha
3. Optimization Theory and Applications by S.S.Rao
4. Operations Research by S.D Sharma

2/22
 Some Optimization Models

Production Allocation Model

A firm produces three type of products X, Y and Z. Four chemicals A,B,C and
D are required to prepare these products. Each unit of these products uses
some units of these four products which are summarized in the following
table. The firm has a stock of some amount of these chemicals as provided in
the table. Firm gets profit Rs 20 in each unit of product X, 15 in each unit of
product Y and 16 in each unit of product Z. Formulate a mathematical model
to maximize the profit of the firm with available resource.
X Y Z Availability
A 3 0 5 30
B 2 1 3 40
C 3 0 2 30
D 1 1 2 50

3/22
Some Optimization Models

Suppose x1 , x2 and x3 are the number of units of the products X,Y and Z
respectively the firm has to produce so that the profit will be maximized.
Model:

max 20x1 + 15x2 + 16x3

subject to 3x1 + 5x3 ≤ 30
2x1 + x2 + 3x3 ≤ 40
3x1 + x3 ≤ 30
x1 + x2 + 2x3 ≤ 50
x1 , x2 , x3 ≥ 0

This model is a Linear Programming Problem(LPP). If the decision variables

x1 , x2 , x3 are integers then this becomes a Linear Integer Programming
problem.

4/22
Some Optimization Models

A general Linear Programming problem may be stated as

M aximize/M inimize c1 x1 + c2 x2 + ... + cn xn

Subject to a11 x1 + a12 x2 + .... + a1n xn ≤, =, ≥ b1
a21 x1 + a22 x2 + .... + a2n xn ≤, =, ≥ b2
.....
am1 x1 + am2 x2 + .... + amn xn ≤, =, ≥ bm
xj ≥ 0, j = 1, 2, ..., n

In matrix form LPP can be represented as

M aximize/M inimize cT x s.to Ax ≤, =, ≥ b, x ≥ 0,

x, c ∈ Rn , b ∈ Rm , A = (aij )m×n

5/22
 Some Optimization Models

Transportation Model

There are three sources (S1 , S2 , S3 )which store some items and send to four
dealers (D1 , D2 , D3 , D4 ). The sources have their individual storage capacity
and the dealers have their demand limits, which are provided in the table.
Transportation cost per unit item from source Si to dealer Dj ,
i = 1, 2, 3, j = 1, 2, 3, 4 are also provided in the table. Determine the optimum
solution for transporting the items at minimum transportation cost.

D1 D2 D3 D4 Storage Capacity
S1 6 4 7 4 50
S2 8 9 5 6 45
S3 7 3 2 8 40
Demand 40 30 30 35

6/22
Some Optimization Models

Model Formulation: Suppose xij =Units transported from Si to Dj . Find xij to

M inimize 6x11 + 4x12 + 7x13 + 4x14

+ 8x21 + 9x22 + 5x23 + 6x24
+7x31 + 3x32 + 2x33 + 8x34
subject T o x11 + x12 + x13 + x14 = 50
x21 + x22 + x23 + x24 = 45
x31 + x32 + x33 + x34 = 40
x11 + x21 + x31 = 40
x12 + x22 + x32 = 30
x13 + x23 + x33 = 30
x14 + x24 + x34 = 35
xij ≥ 0

This is a Linear programming problem problem.

7/22
 Some Optimization Models

General Transportation Problem

Notations:
xij : units transported from ith source to j th destination. m=number of
sources and n= number of destinations.
cij : transportation cost per unit item transported from ith source to j th
destination.
ai : capacity of ith source.
bj : Demand of j th destination so that m
P Pn
i=1 ai = j=1 bj
Model:
Xm X n
minimize cij xij
i=1 j=1
n
X
subject to xij = ai , i = 1, 2, ..., m
j=1
m
X
xij = bj , j = 1, 2, ..., n.
i=1
xij ≥ 0
Pm Pn
If i=1 ai = j=1 bj then this is a balanced transportation model.
8/22
 Some Optimization Models

Assignment Model

A manger has four assignments (A,B,C,D) to complete a project and he has

four officers (P,Q,R,S), who can be assigned these duties according to their
efficiency in terms of time to complete the work. Following table provides
number of hours to complete an assignment by an officer. Determine the
suitable policy to assign each work to an officer to complete the project in
minimum time.

A B C D
P 2 1 1 2
Q 1 2 1 3
R 3 1 2 1
S 2 3 3 2

9/22
Some Optimization Models

Model Formulation:
xij = 1 if ith job is assigned to j th person otherwise xij = 0, i, j = 1, 2, 3, 4.

M inimize 2x11 + x12 + x13 + 2x14

+x21 + 2x22 + x23 + 3x24
+3x31 + x32 + 2x33 + 1x34
+2x41 + 3x42 + 3x43 + 2x44
subject to x11 + x12 + x13 + x14 = 1
x21 + x22 + x23 + x24 = 1
x31 + x32 + x33 + x34 = 1
x21 + x22 + x23 + x24 = 1
x11 + x21 + x31 + x41 = 1
x12 + x22 + x32 + x42 = 1
x13 + x23 + x33 + x43 = 1
x14 + x24 + x34 + x44 = 1
xij = 1 or 0

This is linear binary programming problem.

10/22
 Some Optimization Models

General Assignment Programming

cij : cost associated with by assigning ith job to j th person.

n X
X n
min cij xij
i=1 j=1
n
X
subject to xij = 1, i = 1, 2, ..., n
j=1
n
X
xij = 1, j = 1, 2, ..., n.
i=1
xij = 0 or 1

11/22
 Some Optimization Models

Least Mean Square Problem

Consider a system of linear equations Ax = b, A = (aij )m×n is a matrix of

order m × n, Rank(A) = m, x ∈ Rn , b ∈ Rm . That is find (x1 , x2 , ..., xn ) so
that

a11 x1 + a12 x2 + ... + a1j xj + ... + a1n xn = b1

a21 x1 + a22 x2 + ... + a2j xj + ... + a2n xn = b2
....
ai1 x1 + ai2 x2 + ... + aij xj + ... + ain xn = bi
....
am1 x1 + am2 x2 + ... + amj xj + ... + amn xn = bm

12/22
Some Optimization Models

Solution of this problem can be found analytically by solving the optimization

problem:

min k Ax − b k22
x∈Rn

This is equivalent to
m X
X n
minn ( aij xj − bi )2
x∈R
i=1 j=1

This is an unconstrained quadratic Programming problem.

13/22
Some Optimization Models

Find the maximum volume of a rectangular parallelopiped whose surface

area is at most 10 and at least 6 units units. Solution of this problem can be
found by solving the optimization problem:

max xyz subject to 3 ≤ xy + yz + zx ≤ 5, x, y, z > 0

This is a constrained non linear programming problem.

14/22
 Some Optimization Models

Shortest path problem

Find the minimum distance from (1, 2) to the curve x2 + x − y = 1. Solution

of this problem is the solution of the optimization problem :

min (x − 1)2 + (y − 2)2 , subject to x2 + x − y = 1

This is a constrained non linear programming problem.

15/22
 Some Optimization Models

General Optimization Problem:

Optimization problems are of two categories: Constrained Optimization

Problem(CP), Unconstrained Optimization Problem(UP).

(CP ) : minx∈Rn f (x)

(U P ) : minx∈Rn f (x) or s.to gi (x) ≤ 0, i = 1, 2, ..., m
hj (x) = 0, j = 1, 2, ..., k

where f : Rn → R, gi : Rn → R, hj : Rn → R.
LPP: If all the functions f, gi , hj are linear then these optimization
models are known as Linear Programming Problems.
NLP: If at least on of the functions f, gi , hj is nonlinear then these
optimization models are known as Nonlinear Programming Problems.

16/22
Some Optimization Models

OBJECTIVE OF THIS COURSE IS TO STUDY LPP.

Existence of solution
Solution methodology
Duality theory
Different types of LPP
General LPP
Integer Programming
Transportation Problem
Assignment Problem
Game Theory

17/22
Solution of LPP

A general linear programming problem is stated as follows.

M aximize/M inimize c1 x1 + c2 x2 + ... + cn xn

Subject to a11 x1 + a12 x2 + .... + a1n xn ≤, =, ≥ b1
a21 x1 + a22 x2 + .... + a2n xn ≤, =, ≥ b2
.....
am1 x1 + am2 x2 + .... + amn xn ≤, =, ≥ bm
xj ≥ 0, j = 1, 2, ..., n

In matrix form LPP can be represented as

(P :)M aximize/M inimize cT x s.to Ax ≤, =, ≥ b, x ≥ 0,

x, c ∈ Rn , b ∈ Rm , A = (aij )m×n

Here x ∈ Rn is the decision vector. c ∈ Rn , b ∈ Rm are known vectors. A is a

given matrix. cT x is known as objective function, Ax ≤, =, ≥ b is known as
the set of constraints and xj ≥ 0 are known as variable restrictions.

18/22
Solution of LPP

To solve (P), it is necessary to understand the system Ax ≤, =, ≥ b. The

system of inequalities Ax ≤ b and Ax ≥ b can be converted
P to P x = b using
slack and surplus
P variables. Example: The P inequality j=1 aij xj ≤ bi can be
expressed
P as j=1 aij xj + ui = bi and j=1 aij xj ≥ bi can be expressed
as j=1 aij xj − vi = bi for some ui ≥ 0, vi ≥ 0. ui is known slack variable
and vi is known as surplus variable. Hence first we will study the system

a11 x1 + a12 x2 + ... + a1j xj + ... + a1n xn = b1

...................
ai1 x1 + ai2 x2 + ... + aij xj + ... + ain xn = bi (1)
...................
am1 x1 + am2 x2 + ... + amj xj + ... + amn xn = bm

where rank(A) = m, m≤n

19/22
 Solution of LPP

Gauss Jordan method

This system can be solved by Gauss Jordan method.(Refer class discussion

for details and any linear algebra book. This was a topic in your First year
Math course)
Gauss Jordan method in brief:
At kth stage suppose the system is

ak11 ak12 ... ak1j ... ak1n | bk1

...................
aki1 aki2 ... akij ... akin | bki
...................
akm1 akm2 ... akmj ... akmn | bkm

20/22
Solution of LPP

Row elimination process is used to convert j th column of kth stage to convert

a unit vector at (k + 1)th stage, whose ij place is 1 and rest places are 0(to
obtain basis matrix). That is
 k   
a1j 0
 .   . 
 k   
 aij  →  1  ,
   
 .   . 
akmj 0
akiq k akiq
using ak+1
pq = akpq − apj if i 6
= p otherwise a k+1
iq =
akij akij

21/22
Solution of LPP

Exercise:
Write C program to solve a linear system using Gauss Jordan iterative
scheme.
Define basic solution, Basic feasible solutions, Basic feasible degenerate
solutions and basic feasible non degenerate solutions of a linear system.
Write the set of solutions of the following linear system reducing the
system using Gauss Jordan method.(i)express x1 , x3 , x5 in terms of
x2 , x4 ,(ii)x2 , x3 , x4 in terms of x1 , x5 .
Find all basic solutions, Basic feasible solutions, Basic feasible
degenerate solutions and basic feasible non degenerate solutions.

3x1 + 2x2 − 2x3 + x4 − 3x5 = 6

2x1 + 2x2 − 3x3 + x4 − 2x5 = 2
x1 − 3x3 + 3x4 = 1

22/22