Optimization

Algorithms
Assistant Prof. Mansi Duse 1
 Content
 OVERVIEW OF OPERATIONS RESEARCH:

 OR models – OR Techniques

 LINEAR PROGRAMMING :

 Introduction
 Graphical solution, Graphical sensitivity analysis
 The standard form of linear programming problems
 Basic feasible solutions –
 unrestricted variables
 simplex algorithm
 artificial variables
 Big M method.
 Degeneracy
 alternative optima ,unbounded solutions ,infeasible solutions.

Assistant Prof. Mansi Duse 2

 Operational Research
 Operational Research based on data analysis..
 Given a problem set, we find solution and optimize it i.e minimize it. This is operational Research.
 Operational Research is a method of mathematically based analysis for providing a quantitative basis for
management decisions
 It is a discipline that deals with the application of advanced analytical methods to help make better
decisions..
 In OR, problems are broken down into basic components and then solved in defined steps by
mathematical analyses. (given all constraints must be Satisfied)
 field of mathematics which deals with obtaining efficient methods or obtaining better solutions in real life
problems using mathematics.
 eg: minimizing your travel time from house to workplace.
 eg:- Trying to optimize your phone calls to minimize bills.
 eg:- Pricing of a product depending on the market demand and popularity.

Assistant Prof. Mansi Duse 3

 Applications of OR
Planning of assembly line in any small or large scale industry to
minimize time and energy.
Allocation of resources according to the preferences.
Traffic management i.e planning of new rails or roads and
scheduling of airplanes and buses/trains.
Sports - A big industry for application of probability, statistics
and decision analysis is there.
Urban and regional planning - Efficient - drainage system or
water line system.

Assistant Prof. Mansi Duse 4

Operational Research Models :-
 We can look at the situation as a decision making problem whose solution
requires answering 3 questions :-
1. What are the decision alternatives? (x, y)
2. Under what restrictions is the decision made?
3. What is an appropriate objective criteria for evaluating the alternatives?
General OR-Model :-

Assistant Prof. Mansi Duse 5

 OR Model

Linear Programming
Iterative method
Using graphical method

Assistant Prof. Mansi Duse 6

Linear Programming Problems
• The Linear Programming Problems (LPP) is a problem
that is concerned with finding the optimal value of the
given linear function. The optimal value can be either
maximum value or minimum value.
• A model consisting of linear relationships representing a firm’s
objective and resource constraints.
• LP is a mathematical modeling technique used to determine a level of operational
activity in order to achieve an objective, subject to restrictions called constraints

Assistant Prof. Mansi Duse 7

 LP Model Formulation
• Decision variables
• mathematical symbols representing levels of activity of an operation
• Objective function
• a linear relationship reflecting the objective of an operation
• most frequent objective of business firms is to maximize profit
• most frequent objective of individual operational units (such as a production or
packaging department) is to minimize cost
• Constraint
• a linear relationship representing a restriction on decision making

Assistant Prof. Mansi Duse 8

Max/min z = c1x1 + c2x2 + ... + cnxn

subject to:
a11x1 + a12x2 + ... + a1nxn (≤, =, ≥) b1
a21x1 + a22x2 + ... + a2nxn (≤, =, ≥) b2
:
am1x1 + am2x2 + ... + amnxn (≤, =, ≥) bm

xj = decision variables
bi = constraint levels
cj = objective function coefficients
aij = constraint coefficients
Assistant Prof. Mansi Duse 9
 LP Model: Example
RESOURCE REQUIREMENTS
Labor Clay Revenue
PRODUCT (hr/unit) (lb/unit) ($/unit)
Bowl 1 4 40
Mug 2 3 50

There are 40 hours of labor and 120 pounds of clay available each day

Decision variables
x1 = number of bowls to produce
x2 = number of mugs to produce

Copyright 2006 John Wiley & Sons, Inc. Supplement 13-10 Assistant Prof. Mansi Duse
Maximize Z = $40 x1 + 50 x2
Subject to
x1 + 2x2  40 hr (labor constraint)
4x1 + 3x2  120 lb (clay constraint)
x1 , x2 0
Solution is x1 = 24 bowls x2 = 8 mugs
Revenue = $1,360

Assistant Prof. Mansi Duse 11

Assistant Prof. Mansi Duse 12
Assistant Prof. Mansi Duse 13
Assistant Prof. Mansi Duse 14
Assistant Prof. Mansi Duse 15
Assistant Prof. Mansi Duse 16
Assistant Prof. Mansi Duse 17
Assistant Prof. Mansi Duse 18
Assistant Prof. Mansi Duse 19
Assistant Prof. Mansi Duse 20
Assistant Prof. Mansi Duse 21
Infeasible Problem Linear Programming (LP)
 In some cases, there is no feasible solution area, i.e., there are no points that satisfy all constraints
of the problem.
 An infeasible LP problem with two decision variables can be identified through its graph. For example,
let us consider the following linear programming problem (LPP).
1) Solve the given LPP by using graphical method
𝑀𝑖𝑛𝑖𝑚𝑖𝑧𝑒 𝑧 = 200𝑥1 + 300𝑥2
subject to
2𝑥1 + 3𝑥2 ≥ 1200
𝑥1 + 𝑥2 ≤ 400
2𝑥1 + 1.5𝑥2 ≥ 900
𝑥1, 𝑥2 ≥ 0

22

Assistant Prof. Mansi Duse

 Solution Constraint point : solve for
x1 and x2 the one by one constraint
x1=600,x2=400
x1=400,x2=400
x1=450,x2=600
 The region located on the right of PQR includes
all solutions, which satisfy the first and the third
constraints.
 The region located on the left of ST includes all
solutions, which satisfy the second constraint.
 Thus, the problem is infeasible because there is
no set of points that satisfy all the three
constraints

Assistant Prof. Mansi Duse 23

Unbounded solution :-
A solution which increases or decreases the value of objective function of the
LP problem indefinitely is called unbounded solution.
Unbounded Solution: Graphical Method in LPP
 It is a solution whose objective function is infinite.
 If the feasible region is unbounded then one or more decision variables will increase
indefinitely without violating feasibility, and the value of the objective function can be
made arbitrarily large.
 Consider the following model:

1 2
subject to
1 2
1 2
1 2
1 2

Assistant Prof. Mansi Duse 24

The point (x1, x2) must be somewhere in the
solution space as shown in the figure by shaded
portion. The three extreme points (corner points) in
the finite plane are:
𝑃 = 90, 0 ;
𝑄 = 24, 22 𝑖𝑛𝑡𝑒𝑟𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝑝𝑜𝑖𝑛𝑡 𝑎𝑛𝑑
𝑅 = (0, 70)
The values of the objective function at these
extreme points are:
𝑍(𝑃) = 3600, 𝑍(𝑄) = 2280 𝑎𝑛𝑑 𝑍(𝑅) = 4200

Assistant Prof. Mansi Duse 25

1. Solve the following linear programming problem graphically:
Problems for practice:
𝑀𝑖𝑛𝑖𝑚𝑖𝑠𝑒 𝑍 = 200 𝑥 + 500 𝑦
subject to the constraints:
𝑥 + 2𝑦 ≥ 10
3𝑥 + 4𝑦 ≤ 24
𝑎𝑛𝑑 𝑥 , 𝑦 ≥ 0

Assistant Prof. Mansi Duse 26

1. Solve by using graphical method 4. Solve by using graphical method
𝑀𝑎𝑥 𝑍 = 4𝑥1 + 3𝑥2 𝑀𝑎𝑥 𝑍 = 4𝑥1 + 3𝑥2
Subject to Subject to
2𝑥 + 𝑥 ≤ 1000 4𝑥1 + 3𝑥2 ≤ 24
𝑥 + 𝑥 ≤ 800 𝑥1 ≤ 4.5
𝑥 ≤ 400 𝑥2 ≤ 6
𝑥 ≤ 700 𝑥1 , 𝑥2 ≥ 0
𝑥 ,𝑥 ≥ 0
2. Solve by using graphical method 5. Solve by graphical method
𝑀𝑎𝑥 𝑍 = 10𝑥1 + 6𝑥2 𝑀𝑎𝑥 𝑍 = 3𝑥1 + 5𝑥2
Subject to
Subject to 2𝑥1 + 𝑥2 ≥ 7
5𝑥 + 3𝑥 ≤ 30 𝑥1 + 𝑥2 ≥ 6
𝑥 + 2𝑥 ≤ 18 𝑥1 + 3𝑥2 ≥ 9 𝑎𝑛𝑑
𝑥 ,𝑥 ≥ 0 𝑥1 , 𝑥2 ≥ 0
6. Solve graphically
3. Solve by using graphical method
𝑀𝑎𝑥 𝑍 = 3𝑥1 + 2𝑥2
𝑀𝑎𝑥 𝑍 = 2𝑥1 + 4𝑥2
Subject to
Subject to 𝑥1 + 𝑥2 ≤ 1
𝑥 + 2𝑥 ≤ 5 𝑥1 + 𝑥2 ≥ 3 𝑎𝑛𝑑
𝑥 +𝑥 ≤4
𝑥 ,𝑥 ≥ 0 𝑥1 , 𝑥2 ≥ 0

Assistant Prof. Mansi Duse 27

7. Solve by using graphical method
𝑀𝑎𝑥 𝑍 = 6𝑥1 + 𝑥2
Subject to
2𝑥1 + 𝑥2 ≥ 3
𝑥 − 𝑥 ≥0
𝑥1 , 𝑥2 ≥ 0

8. Solve by graphical method

𝑀𝑖𝑛 𝑍 = 20𝑥1 + 40𝑥2
Subject to
36𝑥 + 6𝑥 ≥ 108
3𝑥 + 12𝑥 ≥ 6
20𝑥 + 10𝑥 ≥ 10 𝑎𝑛𝑑
𝑥 ,𝑥 ≥ 0

Assistant Prof. Mansi Duse 28

 Basic Definition:
1. Solution:- A set of values of variables 𝑋 (𝑗 = 1 2 … … 𝑛) is called of a LPP if it satisfies
constraints.
2. Feasible solution :-The set of values of decision variables 𝑋 (𝑗 = 1 2 … … 𝑛) which satisfy all the
constraints and non-negativity conditions of an LP problem simultaneously is said to be the
feasible solution to that linear programming problem.
3. Infeasible solution :- The set of values of decision variables 𝑋 (𝑗 = 1 2 … … 𝑛) which do not satisfy
all the constraints and non-negativity conditions of an LP problem simultaneously is said to
constitute the infeasible solution to that (LPP) linear programming problem.
4. Basic solution :- For a set of m simultaneous equations in n variables, a solution obtained by
setting (n-m) variables equal to zero and solving for remaining m equations in m variables is called a
basic solutions.
5. Non basic and basic variable :- The (n-m) variables whose values did not appear in this solution are
called non basic variable and the remaining m variables are called basic variables.
6. Optimum basic feasible solution:-A basic feasible solution which optimizes (Maximizes or
Minimizes) the objective function value of the given LP problem is called an optimum basic
feasible solution.

Assistant Prof. Mansi Duse 29

1. Basic feasible solution :-A feasible solution to an LP problem which is also the basic solution is called the basic
feasible solution. All basic variables assume non-negative values. Basic feasible solutions are of two types,
1) Degenerate: A basic feasible solution is called degenerate if value of at least one basic variable is zero.
2) Non-degenerate: A basic feasible solution is called non-degenerate if values all m basic variables are non-zero
and positive.
2. Unbounded solution :- A solution which increases or decreases the value of objective function of the LP
problem indefinitely is called unbounded solution.
3. Slack Variables:- If a constraints has sign (≤) in order to make it an equality we have to add something positive
to the left hand side. The positive variables which are added to left hand side to the constraints convert them into
equalities are called Slack Variables.
4. Surplus Variables: - If a constraints has a to sign (≥) in order to make it an equality we have to subtract
something positive to the left hand side. The positive variable which are subtracted to the left hand side to the
constraints to convert them into equalities are called Surplus variables.
5. Objective function:-In LPP a linear real valued function of decision variable which is to be optimized(maximize
or minimize) is called objective function.
The objective function is of the form Z = ax + by, where x, y are the decision variables. The function Z = ax +
by is to be maximized or minimized to find the optimal solution. Here the objective function is governed by the
constraints x > 0, y > 0.

Assistant Prof. Mansi Duse 30

 Slack Variable Example
The variables on the left, 𝑥 , 𝑥 𝑥 , are the basic
variables; the set of these variables is called
the basis. The rest of the variables are called non-
basic.

Solution:- introducing slack variable 𝑥 , 𝑥 , 𝑥

𝑀𝑎𝑥 𝑧 = 𝑥 + 2𝑥 − 𝑥 + 0𝑥 + 0𝑥 + 0𝑥

Assistant Prof. Mansi Duse 31

 Surplus Variable Example
Solution:-Introducing slack and artificial variable
𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 𝑧 = 𝑥 + 2𝑥 − 𝑥
𝑥 , 𝑥 , 𝑥 𝑎𝑟𝑒 𝑠𝑢𝑟𝑝𝑙𝑢𝑠 𝑣𝑎𝑟𝑖𝑏𝑙𝑒 𝑎𝑛𝑑 𝐴 , 𝐴 𝐴 𝑎𝑟𝑒 𝑎𝑟𝑡𝑖𝑓𝑖𝑐𝑎𝑙 𝑣𝑎𝑟𝑖𝑏𝑙𝑒
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜
𝑀𝑎𝑥𝑖𝑚𝑖𝑧𝑒 𝑧 = 𝑥 + 2𝑥 − 𝑥 + 0𝑥 + 0𝑥 + 0𝑥 − 𝑀𝐴 − 𝑀𝐴 − 𝑀𝐴
2𝑥 + 𝑥 + 𝑥 ≥ 14
𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜
4𝑥 + 2𝑥 + 3𝑥 ≥ 28
2𝑥 + 𝑥 + 𝑥 + 𝒙𝟒 + 𝑨𝟏 = 14
2𝑥 + 5𝑥 + 5𝑥 ≥ 30
4𝑥 + 2𝑥 + 3𝑥 + 𝒙𝟓 + 𝑨𝟐 = 28
𝑥 ,𝑥 ,𝑥 ≥ 0
2𝑥 + 5𝑥 + 5𝑥 + 𝒙𝟔 + 𝑨𝟑 = 30
𝑥 ,𝑥 ,𝑥 ,𝑥 ,𝑥 ,𝑥 ,𝐴 ,𝐴 𝐴 ≥ 0

Assistant Prof. Mansi Duse 32

 Simplex Algorithm
For the solution of any LPP by simplex algorithm the existence of
do initial basic feasible solution is always assume the following step for
the computation of on optimum solution are as follows:
• Step1:- Check whether the objective function of the given LPP is to be
maximize or minimize, if it minimize then we convert it into a
maximization problem. It by using the result min(z) = -max(-z)
• Step2:- Check whether all 𝒊 are non-negative. If any one
of the bi is negative then multiply the corresponding equation of
constraint by -1.
• Step 3:- Convert all the inequality of constraint into equality by
adding slack (Surpls variable in constraint.) Put the cost equal to zero.
i.e identity matrix.

Assistant Prof. Mansi Duse 33

• Step 4:- Obtain an initial basic feasible solution to the problem in the form and put it in the
Column of Solution of the Simplex table.
• Step 5:- To test optimality compute .
1. If all 𝒋 then the initial basic feasible solution 𝑩 is an optimum basic feasible solution.
2. If at least one 𝒋 proceed on to the next step
• Step 6:- If there are more than one positive 𝒋 then choose most positive of them and entre the
corresponding variable, called as entering variable Here Corresponding column is called as key
column.
1. If all 𝒋 then there is an unbounded solution .
2. If at least one then the corresponding vector enter the basis.
• Step 7:-
𝒙𝑩
1. Compute the minimum ratio 𝐣 and choose the minimum
𝒙𝒋 (𝒌𝒆𝒚 𝒄𝒐𝒍𝒖𝒎𝒏 𝒆𝒍𝒆𝒎𝒆𝒏𝒕)
of them.
2. Let the minimum of this ratio then the corresponding will be new, these variable called as
leaving variable & corresponding key row.
3. The common element in (key row and key column is known as key element) of the table.

Assistant Prof. Mansi Duse 34

• Step 9:-
1. For corresponding key row new element 𝒌𝒆𝒚 𝒓𝒐𝒘 = 𝒌𝒆𝒚𝒌𝒆𝒚
𝒓𝒐𝒘 𝒆𝒍𝒆𝒎𝒆𝒏𝒕
𝒆𝒍𝒆𝒎𝒆𝒏𝒕
To find new element elements
𝑵𝒆𝒘 𝒆𝒍𝒆𝒎𝒆𝒏𝒕
𝒄𝒐𝒓𝒓𝒆𝒔𝒑𝒐𝒏𝒅𝒊𝒏𝒈 𝒆𝒍𝒆𝒎𝒆𝒏𝒕 𝒊𝒏 𝒌𝒆𝒚 𝒓𝒐𝒘 ∗ 𝒄𝒐𝒓𝒓 𝒆𝒍𝒆𝒎𝒆𝒏𝒕 𝒊𝒏 𝒌𝒆𝒚 𝒄𝒐𝒍𝒖𝒎𝒏
= 𝒐𝒍𝒅 𝒆𝒍𝒆𝒎𝒆𝒏𝒕 −
𝒌𝒆𝒚 𝒆𝒍𝒆𝒎𝒆𝒏𝒕
• Step 10:- Go to Step 5 and repeat all procedure until either at optimum solution is
obtain or they is an indication of an unbounded soulution.

Assistant Prof. Mansi Duse 35

 Artificial Variables
Let us consider the LPP in which constraints may also have “ “ and “=“ sign after
ensuring that all .
In such problem basis matrix is not obtained as an identity matrix in the starting
simplex tables. Therefore we introduce a new type of variable called the Artificial
Variables.
This variable are function and cannot have any physical meaning.
The artificial variable technique is Marely a device to get starting B.F.S.
So that simplex procedure may be adopted as usual until the optimal solution is
obtained.

Assistant Prof. Mansi Duse 36

 Big – M method
1. The artificial variable are imaginary & are introduce only for computational purpose
and cannot have physical meaning.
2. By assigned a very large per unit penalty to each of artificial variable can be removed
from the simplex table as soon as they become non-basic. Such a penalty is –M for
maximization problem and M for minimization problem. Where
3. The constraint with a greater than or equals to in equality sing requires both a
surplus and artificial variable.
4. If one of the constraint in the problem. is = (equality) then an artificial variable has to
be added to get initial Solution.
5. It may be noted that at equality constraint requires neither a slack variable not a
surplus variable.
6. No feasible solution in exist to the problem if all the artificial variable cannot be driven
(out of) out in the Method optimal become simplex method.

Assistant Prof. Mansi Duse 37

• Step 1: Introduce surplus and artificial variable in each row.
• Step 2: put the artificial variable into the objective function for

• Step 3: Solve the Linear programming by simplex method.

• Note:
1. If all the artificial variable in the optimal solution equal to zero then solution is
optimal.
2. If any artificial variable positive in the optimal solution problem is infeasible

Assistant Prof. Mansi Duse 38