Indian Institute of Technology Bombay

Department of Electrical Engineering

Academic Session: 2022-23, Semester: I

Course Project Report

Course: EE 491: BTP I

Student:

Sl Roll No. Name Email

1 190070035 Mohit Gusaiwal [email protected]

Project Mentors: Prof. Madhu Belur

Project Title: Linear Programming

Date of Submission: November 24, 2022

1
Contents

1 Abstract 2

2 Introduction 3

3 COMPONENT OF LINEAR PROGRAMMING PROBLEM 4

4 Algorithms for Linear Programming 5

5 Duality Theory in Linear Programming 9

6 Some Real Life Applications of Linear Programming 9

7 Conclusions 10

Acknowledgments 10

References 10

1 Abstract
Linear programming was developed during World War II when a system with which one
wanted to maximize the efficiency of resources was of utmost importance. The word
“programming” was a military term that referred to activities such as planning schedules
efficiently or deploying men optimally. Economics, have developed the theory behind
“linear programming” and explored its applications

An optimization problem with a linear objective function and linear constraints is

called a linear programming problem.
This project proposes to learn about some algorithms like the simplex method, steep-
est descent method, Interior point method, and ellipsoidal method. Also, I learn about
what is primal and duality of a linear programming problem are. In the end, we shall see
some real-life applications of linear programming.

2
2 Introduction
A division or subset of mathematical programming is known as linear programming (LP).
LP is a technique that uses quantitative analysis to achieve a purpose. In several fields,
it’s a crucial strategy for optimization. It is employed to find the ideal answer to the
issue within a set of limitations. There are some linear equality and inequality con-
straints, as well as an objective function. This method is used to maximize or minimize
the objective function of the given mathematical model comprising the set of linear in-
equalities, depending upon some constraints represented within the linear relationship.
It is concerned with optimizing linear functions under linear constraints. In particular,
LP assists in effectively allocating resources to maximize profits, minimize losses, or fully
utilize assembly capacity. LP always aids in the formulation of practical issues into a
rigid mathematical model. As a result, it is sometimes described as a mathematical
method for figuring out the best answer to a variety of real-world situations or for de-
termining the simplest value to use in a given situation. It is employed to determine
how to best use resources while incurring the fewest expenses. In other words, it deals
with the distribution of resources under certain constraints, such as price and availability.

Figure 1: Block Diagram characteristics of Linear programming problem

Source: https://images.app.goo.gl/yXTXJkZ7MeUVyQbj8

Mathematics and other areas of the social and physical sciences make substantial use
of LP. Decisions about business planning, industrial engineering, economics, telecommuni-
cation, energy, transportation, routing, manufacturing fields, various types of scheduling,
etc. are only a few examples of the areas where it is employed. It is the most straightfor-
ward and often applied approach to problem optimization. The multi-sector optimization
problem typically arises in real-world situations where regional development planning, the
development of water or electrical systems, urban planning, or environmental preserva-
tion are involved.

3
 This project report is divided into some parts: The first one is the introduction of
components of linear programming problems. The second one is some algorithms to solve
linear programming problems, and the third one is the comparison between some solvers
of LP problems, and at the end conclusions.

3 COMPONENT OF LINEAR PROGRAMMING

PROBLEM
The LP problem consists of different components. According to, the major components
of LP are stated as:

3.1 Decision variables

The physical quantities that the decision-maker uses to calculate and regulate are known
as decision variables, and they are denoted by mathematical symbols. These factors are
tangible things that the decision-maker can influence. Such variables are usually denoted
by x1 ,x2 ,.....xn , where n is a finite positive integer i.e. xi ≥ 0,i ∈ I + .In the LP model,
all decision variables are continuous, controllable, and nonnegative. These are unknown
quantities and are expected to be estimated as an output of the LP problem.

3.2 The objective function

The objective function of each LP problem is expressed in terms of decision variables to
optimize the criterion of optimality. The objective function Z = c1 x1 + . . . . . . + cn xn is
a linear function of the n decision variables. The constants c1 c2 . . . ..cn are real numbers
that are called the coefficient of the objective function. The objective function always
specifies a way of optimization, either to minimize or maximize.

3.3 The constraints

The degree to which an objective can be fulfilled is constrained by constraints, which
are limitations on the use of resources relating to physical, economic, legal, technologi-
cal, ethical, and other restraints. For instance, the outcome may be influenced by the
resources, such as raw materials, labor, machinery, money, personnel, space, etc. Such
constraints must be expressed as linear equalities or inequalities in regard to decision
variables.

4
 Figure 2: Block Diagram components of Linear programming problem
Source: https://images.app.goo.gl/gwNhvjkrov8ZbitF6

4 Algorithms for Linear Programming

4.1 Simplex/Tableau Method
A simplex method is an approach to solving linear programming models by manually us-
ing slack variables, tableaus, and pivot variables as a means to find the optimal solution
to an optimization problem.

Using the simplex method, required objective function and constraints in standard
form. Standard form means the problem must be a maximization problem, all linear
constraints must be in a less than equal to inequality and all variables are non-negative.

i.e. Max: z = C T X ;
AX ≤ B;
Xi ≥ 0

Example: Maximize: z = 8x1 + 10x2 + 7x3

x1 + 3x2 + 2x3 ≤ 10
x1 + 5x2 + x3 ≤ 8
x1 ,x2 ,x3 ≥ 0

Step 1: Now add slack variables in constraints

x1 + 3x2 + 2x3 + s1 = 10
x1 + 5x2 + x3 + s2 = 8
x1 ,x2 ,x3 ,s1 ,s2 ≥ 0

5
 Figure 3: Steps of Simplex Method
Source: https://images.app.goo.gl/CA5GbiamQ6MgVoK2A

Step 2: Now set Tableau method

x1 x2 x3 s1 s2 z b
1 3 2 1 0 0 10
1 5 1 0 1 0 8
-8 -10 -7 0 0 1 0

now check the smallest value in the last row and make it a pivot column.
R2 –> R52
R1 –> R1 - 3R2 , R3 –> R3 + 10R2

x1 x2 x3 s1 s2 z b
2 7 −3 26
5
0 5
1 5
0 5
1 1 1 8
5
1 5
0 5
0 5
-6 0 -5 0 2 1 16

now smallest value in the last row is -6 in column 1.

R2 –> 5R2
R1 –> R1 - 2R5 2 , R3 –> R3 + 6R2

So no negative value in the last row.

Basic Variable: A basic variable can be classified to have a single 1 value in its column
and the rest be zeros.

6
 x1 x2 x3 s1 s2 z b
0 -2 1 1 -1 0 2
1 5 1 0 1 0 8
0 30 1 0 8 1 64

So x1 , s1 ,z are basic variables and x2 , x3 ,s2 are non basic variables.

x1 = 8 , x2 0 , x3 = 0 , s1 = 2 , and s2 = 0.
Maximize z = 64

4.2 Interior-Point Algorithms

Interior point methods are a type of algorithm that is used in solving both linear and
nonlinear convex optimization problems that contain inequalities as constraints. Basi-
cally in the simplex method, we find an optimal solution on edges but in the interior
point method, the optimal solution exists in the feasible region. Interior point methods
came about from a desire for algorithms with better theoretical bases than the simplex
method. While the two strategies are similar in a few ways, the interior point methods
involve relatively expensive (in terms of computing) iterations that quickly close in on
a solution, while the simplex method involves usually requires many more inexpensive
iterations. From a geometric standpoint, interior point methods approach a solution from
the interior or exterior of the feasible region but are never on the boundary.

There are two important interior point methods: the barrier method and the primal-
dual interior point method.

4.3 Steepest Descent Method

The steepest descent method is also known as the gradient descent method. The steepest
descent method is a type of algorithm that is used to solve only minimization problems.
This method is used when it is not possible to find the minimum of a function analytically
and therefore must use an iterative method for obtaining an approximate solution. The
steepest descent approach, despite being straightforward, has contributed significantly
to the growth of optimization theory. Unfortunately, the method is not commonly em-
ployed because it is known to be quite slow in solving the majority of real-world situations.

Steepest method steps:

Step 1: Start with arbitrary initial point X0 .

Step 2: Find search direction di at X0 .

di = -∆fi = -∆f(X0 ).
∆ represents the gradient of function f.

Step 3: now find optimal step length α.

df
f(X0 + αdi ) and dα = 0 and find α.

7
 Figure 4: Steepest Descent Method approach
Source: https://images.app.goo.gl/MogAuHoLBo3AJ7s48

Step 4: Now set XK+1 = XK + αdi

Repeat step 2 till reach the optimal solution.

These steps we understand from figure 3 that we start the process from X0 then find
X1 , X2 ,X3 , X3 ... till reaching the optimal solution.

4.4 Ellipsoidal Method

The ellipsoidal method, proposed by Khachiyan, is the first polynomial time algorithm for
linear programming. This method is used on low-dimensional problems, such as planar
location problems, where it is numerically stable. This method generates a sequence of
ellipsoids whose volume uniformly decreases at every step, thus enclosing the minimizer
of a convex function. The ellipsoidal method is attractive because its complexity depends
on the number of columns and the digital size of the coefficients, but not on the number of
rows. Although the algorithm is theoretically better than the Simplex algorithm, which
has an exponential running time in the worst case, it is very slow practically and not
competitive with simplex.

8
5 Duality Theory in Linear Programming
An essential idea in the discipline of operations research is the theory of duality. The
original application of this idea was in relation to linear programming. According to the
idea of duality in linear programming, each linear programme has a dual that is a re-
lated linear programme. The primal refers to the original problem in respect to its dual.
The true core of duality theory is the relationship between the primal and its dual, both
mathematically and economically.

A general expression for the primal and dual linear program.

For Primal problem
Max Zx = CX
AX ≤ b
X≥0

For Dual problem

Min Zw = bT W
AT W ≤CT
W≥0

If primal problem is a maximization problem then the dual problem must be a mini-
mization problem and vice versa also true. The dual of dual problem is again the primal
problem.
We clearly see the number of variables in the primal problem is equal to the number of
constraints in the dual problem and the number of constraints in the primal problem is
equal to the number of variables in the dual problem.

6 Some Real Life Applications of Linear Program-

ming
As described earlier, linear programming(LP) is used for the utilization of resources for
the appropriate decision-making. So, it is used to select the best feasible approach from
a number of options.LP problems are mainly concerned with two important classes of
objects. The first is related to the limited resources such as land, labor, capital, materials,
machines, etc., and the second is activities related to production such as the production
of low carbon, production of more stainless steel, and production of high carbon steel,
etc.

6.1 Agriculture
First, from an economic standpoint, LP may be used in agricultural production planning,
farm and land management, and allocation of scarce resources like land, labor, money,
materials, water supply, etc. Thus, it should be utilized optimum allocation of crop pro-
duction, efficient production patterns, optimum use of resources, and a more efficient
product distribution system through the use of a linear programming model.

9
6.2 Industry
LP may be applied to several fields of industry. The industry uses LP, especially in the
manufacturing goods and service industry. Manufacturing industries use LP to maxi-
mize efficiency with minimum operation cost or maximum product at minimum cost.
The manufacturer can use the LP model to reconfigure their storage layout, adjust their
workforce and reduce the blockage so that they can maximize the production or minimize
the cost of the product.

6.3 Engineering
LP is the major tool for engineers in shape optimization that reduces cost and increases
efficiency. It helps them solve manufacturing problems and create different designs. It
helps increase the production levels for maximizing profit in particular conditions. It also
helps solve problems related to design and manufacturing. Likewise, it assists to increase
efficient production and maximize profit.

7 Conclusions
LP is a branch of mathematics and statistics that permits researchers to find out solutions
to the problems of optimization. It is broadly used to determine the optimum value in
several real-life applications. It can be used in several fields of real-life situations such
as agriculture, transportation, industries, engineering, etc. So we use different methods
to find the solutions to different types of problems such as the simplex method, steepest
descent method, interior point method, and graphical method.

Acknowledgements
We would like to thank: Prof. Madhu Belur for his unwavering support and guidance.

References
[1] Sharma, J. Operations research theory and applications. Sixth Edition. Laxmi Pub-
lications Pvt. Ltd, 2017.

[2] Bixby, Robert E. Solving real-world linear programs: A decade and more of progress.
Operations research. 2002; 50(1): 3–15.

[3] Dantzig G B, Thapa MN. Linear programming 1:Introduction. Springer-Verlag New

York, Inc, 1997.

[4] Jyothimon C, M.Tech Technology Management University of Kerala

[5] EJ-MATH, European Journal of Mathematics and Statistics ISSN: 2736-5484

10