Graphical Method

Solutions using Geometry
1 The Geometric Approach

In this section, we outline the steps necessary to solve a linear programming
problem graphically. However, we must note that this method is limited to
problems with three variables. When variables exceed 3, the graphical method
becomes hard to implement and follow, and we must then use matrix methods.
Consider the following inequalities a1 x + b1 y + c1 z ≤ d1 , a2 x + b2 y + c2 z ≥
d2 , · · · , an x + bn y + c3 z ≤ dn where ai , bi , ci and di are all constants. These
inequalities may be sketched on the same axes. If di = 0, then they are
inequalities in the x − y plane and can be sketched by hand. To do this, we
sketch the related lines a1 x + b1 y = c1 and a2 x + b2 y = c2 . Note that these
straight lines divide the x − y plane into two halves; one half lies below each
line and the other above the line. We must determine the feasible region of
each inequality by picking a random point either above or below each of the
lines above. If the point satisfies the inequality ai x + bi y = ci , for each i = 1, 2,
then we conclude that that point (and all points in the corresponding half of
the x − y plane) is in the feasible region. So we shade off points which do not
satisfy the inequality. This process is repeated for all the given inequalities.
The intersection of all the feasible regions is referred to as the feasible region
of the entire system of inequalities.
Definition 1. A region R on the x − y plane is called a feasible region of the

system of inequalities
ai x + b i y + c i z ≤ d i , ∀i = 1, 2, . . . , n
if for any point (x, y, z) ∈ R, each of the inequalities in the system holds.
Definition 2. We may also say that a solution (x, y, z) to a linear program-

ming problem is feasible if it satisfies all the constraints. It is called optimal if,
in addition, it gives the desired maximum/minimum value, depending on the
nature of the objective function.
Some problems are not well posed, meaning that they have no feasible
solution. For example, consider the following problem
Example 1. Solve the LP problem below: Maximize 4x1 + 5x2 subject to the
constraints
x1 + x 2 ≤ 2
−2x1 − 2x2 ≤ −9
x1 , x2 ≥ 0.
Kabale University, June 2021: MTH 2205 2
To solve this problem, we sketch the constraints on the same axes. We get
the following diagram From th figure above, we see that there is no feasible
region (since we have shaded the entire x1 − x2 plane). Thus R = ∅, implying

that the LP problem has no solution.
Definition 3. A set X ⊂ R is called convex if for any two points x1 , x2 ∈ X,

the convex combination (the line segment) tx1 +(1−t)x2 ∈ X for any t ∈ [0, 1].
That is, if X is convex, then the line segment joining any two random
points in X lies completely in X. Foe example, the following sets are convex
while the sets below are not convex as we will see in class
Remark 1. The feasible region of an LP problem is always a convex set.
Sometimes, it is not possible to get an optimal solution because the feasible

region is unbounded. For example,
Example 2. Solve the LP problem below: Maximize 4x1 + 5x2 subject to the
constraints
−2x1 + x2 ≤ −1
−x1 − 2x2 ≤ −2
x1 , x2 ≥ 0.
Sketching the constraints on the same axes, we get the following figure We
see that the feasible region R is not empty and it is convex, so the LP problems
has at least one feasible solution. However, the region R is unbounded, so there
is no optimal solution. In this case, we say that the LP problem is unbounded.
Definition 4. A point x ∈ X in a convex set X is said to be an extreme point

if it cannot be expressed as a strict convex combination of two distinct points
in X.
This means that if x is an extreme point, then we cannot pick the parameter
t to satisfy 0 < t < 1.
Remark 2. Geometrically, an extreme point is a corner point of the feasible

region.
Theorem 1. (Dantzig, 1951) Suppose that a feasible set of an LP problem

is non-empty and bounded. Then the LP problem has at least one optimal
solution located at an extreme point.
You may find the proof in some standard Linear Programming textbooks.
Example 3. Solve the LP problem: Maximize 3x + 7y subject to the initial

conditions
x − 2y ≤ 3
5x + 3y ≤ 12
−3x + 8y ≥ 20
x, y ≥ 0.
For this problem, we can sketch the feasible region as follows: We see that
the feasible region R is non-empty and bounded, so there is at least one feasible
solution, and an optimal solution occurs at one of the extreme points. Hence,
determining all the points at the corners, we get
Corner Points Objective function 3x + 7y

(0, 4) 28
(0, 2.5) 17.5
36 136
,
49 49
21.63
Thus, the optimum solution is (0, 4), which gives a maximum value of 28.
Example 4. Solve the linear programming problem below using the graphical
approach: Minimize 3x + 2y subject to the following constraints
1.7x + 2y ≤ 22
x − 3y ≤ 12
5x − y ≥ −10
x + 3y ≥4
x, y ≥ 0.
To solve this problem, we sketch all the inequalities on the same axes as
follows Then, we apply Dantzig’s theorem. The minimum value is expected to
be at the extreme points of th feasible region. So we find all these points. We

see that (0, 4/3), (4, 0), (12, 0) and (0, 10) are extreme points. The other three
extreme points must be calculated by solving simultaneous equations:

2 127
5x − y = −10, 1.7x + 2y = 22 which gives , and
11.7 11.7

88 28
x − 3y = 12, 1.7x + 2y = 22 which gives the point , .
7.1 213
Thus,
Extreme Points Objective function 3x + 2y
(4, 0) 12
(0, 4/3) 2.667
(0, 10) 20
(12, 0) 36
2
, 127
11.7 11.7
22.22
88 28
,
7.1 213
37.45
Therefore, the solution to the minimization problem is (0, 4/3) and the mini-
mum value is 2.67.
The following are the steps required to solve an LP problem graphically:
(i) Sketch all the constraints and determine the feasible set
(ii) Determine all points on the boundary of the feasible region, specifically
those at the corners
(iii) Substitute all the extreme points into the objective function to determine
which of the points gives a maximum/minimum value, depending on the
objective of the research. This point will be the optimal feasible solution.
2 The Standard Form of an LP problem

An LP problem is said to be in standard form if it is of the form
M aximizec1 x1 + c2 x2 + · · · + cn xn subject to the conditions
a11 x1 + a12 x2 + · · · + a1,n xn ≤ b1

a21 x1 + a22 x2 + · · · + a2,n xn ≤ b2
.. .. ..
. . .
am1 x1 + am2 x2 + · · · + am,n xn ≤ bm
x1 , x2 , . . . , xn ≥ 0
In other words, all the constraints must be expressed as ≤. For example,
Example 5. Write the following problem in standard form: Maximize 2x1 +

3x2 − x3 + 2x4 subject to the constraints
−3x1 + 3x2 + x4 ≤ 10
x1 + 3x2 + −5x3 ≥ −8
−5x1 + 2x2 − x3 + 2x4 ≥5
x2 + x4 ≤ 20
x1 , x2 , . . . , xn ≥0

Graphical Method

Uploaded by

Copyright:

Available Formats

Graphical Method

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Graphical Method

Uploaded by

Copyright:

Available Formats

Solutions using Geometry

1 The Geometric Approach

Definition 1. A region R on the x − y plane is called a feasible region of the

Definition 2. We may also say that a solution (x, y, z) to a linear program-

region (since we have shaded the entire x1 − x2 plane). Thus R = ∅, implying

Definition 3. A set X ⊂ R is called convex if for any two points x1 , x2 ∈ X,

Remark 1. The feasible region of an LP problem is always a convex set.

Sometimes, it is not possible to get an optimal solution because the feasible

Definition 4. A point x ∈ X in a convex set X is said to be an extreme point

Remark 2. Geometrically, an extreme point is a corner point of the feasible

Theorem 1. (Dantzig, 1951) Suppose that a feasible set of an LP problem

Example 3. Solve the LP problem: Maximize 3x + 7y subject to the initial

Corner Points Objective function 3x + 7y

be at the extreme points of th feasible region. So we find all these points. We

2 The Standard Form of an LP problem

M aximizec1 x1 + c2 x2 + · · · + cn xn subject to the conditions

a11 x1 + a12 x2 + · · · + a1,n xn ≤ b1

Example 5. Write the following problem in standard form: Maximize 2x1 +

You might also like