Linear programming
Mathematical Methods for Business
Objectives
• To represent geometrically a linear inequality in two variables.
• To state a linear programming problem and solve it geometrically.
• To consider situations in which a linear programming problem exists.
• To introduce the simplex method.
• To solve problems using the simplex method.
• To introduce use of artificial variables.
• To learn alteration of objective function.
• To define the dual of a linear programming problem.
Linear Inequalities in 2 Variables
• Linear inequality is written as
ax by c 0 or
ax by c 0 or
ax by c 0 or
ax by c 0
where a, b, c = constants and not both a and b are zero.
• A solid line is included in the solution and a dashed line is not.
Example 1 – Solving a Linear Inequality
Find the region defined by the inequality y ≤ 5.
Solution:
The region consists of the line y = 5
and with the half-plane below it.
Example 3 – Solving a System of Linear Inequalities
Solve the system y 2 x 10
y x 2
Solution:
The solution is the unshaded region.
Linear Programming
• A linear function in x and y has the form
P ax by
• The function to be maximized or minimized is called the
objective function.
Example 1 – Solving a Linear Programming Problem
Maximize the objective function P = 3x + y subject
to the constraints 2 x y 8
2 x 3 y 12
x0
y 0
Solution:
The feasible region is nonempty and bounded.
Evaluating Z at these points, we obtain:
Z A 30 0 0
Z B 34 0 12
Z C 33 2 11
Z D 30 4 4
The maximum value of Z occurs when x = 4 and y = 0.
Example 3 – Unbounded Feasible Region
The information for a produce is summarized as follows:
If the grower wishes to minimize cost while still maintaining the
nutrients required, how many bags of each brand should be bought?
Solution:
Let x = number of bags of Fast Grow bought
y = number of bags of Easy Grow bought
To minimize the cost function C 8 x 6 y
Subject to the constraints
3 x 2y 160
5 x 2y 200
x 2y 80
x0
y 0
There is no maximum value
since the feasible region is unbounded .
Multiple Optimum Solutions
• There may be multiple optimum solutions for an objective function.
Example 1 – Multiple Optimum Solutions
Maximize Z = 2x + 4y subject x 4 y 8
to the constraints x 2y 16
Solution: x, y 0
The region is nonempty and bounded.
Z A 20 42 8
Z B 28 44 32 (max)
Z C 20 48 32 (max)
The Simplex Method
Standard Linear Programming Problem
• Maximize the linear function Z c1x1 c 2 x 2 c n x n
subject to the constraints
a11 x1 a12 x 2 a1n x n b1
a11 x1 a12 x 2 a1n x n b1
a11 x1 a12 x 2 a1n x n b1
.
where x and b are non-negative. .
am1x1 am 2 x 2 amn x n bm
Example 1 – The Simplex Method
x1 x 2 20
Maximize Z 5 x1 4 x2 subject to 2 x1 x 2 35 and x 1, x2 ≥ 0.
3 x x 12
1 2
Solution:
B x1 x2 s1 s2 s3 Z R B x1 x2 s1 s2 s3 Z R
s1 1 1 1 0 0 0 20 x2 0 1 2 1 0 0 5
s2 2 1 0 1 0 0 35 x1 1 0 1 1 0 0 15
s3 3 1 0 0 0 0 12 s3 0 0 5 4 1 0 52
Z 5 4 0 0 0 1 0 Z 0 0 3 1 0 1 95
The maximum value of Z is 95.
7.5 Degeneracy, Unbounded and Multiple Solutions
Degeneracy
• A basic feasible solution (BFS) is degenerate if one of the basic variables
is 0.
Unbounded Solutions
• An unbounded solution occurs when the objective function has no
maximum value.
Multiple Optimum Solutions
• Multiple (optimum) solutions occur when there are different BFS.
Example 1 – Unbounded Solution
5 x1 6 x 2 2 x3 30
Maximize Z x1 4 x 2 x3subject to x1 3 x 2 6 x3 12
x1, x 2 , x3 0
Solution:
B x1 x2 s1 s2 s3 Z R B x1 x2 s1 s2 s3 Z R
s1 - 5 6 - 2 1 0 0 30 s1 - 3 0 - 14 1 - 2 0 6
s2 - 1 3 6 0 1 0 12 x2 - 13 1 2 0 13 0 4
Z - 1 - 4 - 1 0 0 1 0 Z - 73 0 9 0 43 1 16
Since the 1st two rows of the x1-column are negative, no quotients
exist.
Hence, it has an unbounded solution.
Artificial Variables
• Use artificial variables to handle maximization problems that are not of
standard form.
Example 1 – Artificial Variables x x 12
1 2
Maximize Z 2 x1 x 2 subject to x1 2x2 20
x1 x 2 2
Solution:
x1, x 2 0
Add an artificial variable t, x x s 12
1 2 1
x1 2 x 2 s2 20
x1 x 2 s3 t 2
Consider an artificial objective equation,
2 x1 x 2 Mt W 0
Construct a Simplex Table,
B x` x2 s1 s2 s3 t W R B x` x2 s1 s2 s3 Z R
s1 1 1 1 0 0 12
0 0 s1 1 0 1
2 0 1
2 0 5
s2 1 2 0 1 0 0 0 20 0 32 0 1
s 2 0 1 1
2
t 1 1 0 0 1 1 0 2 t 0 1 1
0 12 0 7
2
W 2 M 1 M 0 0 M 0 1 2M
W 0 0 3
2 0 1
2 1 17
All indicators are nonnegative.
The maximum value of Z is 17.
Example 3 – An Empty Feasible Region
Use the simplex method to maximize Z 2 x1 3x2
subject to x x 2
1 2
x1 x2 1
x ,x 0
1 2
Solution:
Construct a Simplex Table,
B x1 x2 s1 s2 t1 W R
t1 1 1 1 0 1 0 2 B x1 x2 s1 s2 t1 W R
s2 1 1 0 1 0 0 1 t1 2 0 1 1 1 0 1
W 2 M 1 M M 0 0 1 2M s2 1 1 0 1 0 0 1
W 1 2M 0 M 1 M 0 1 1 M
The feasible region is empty, hence, no solution exists.
Minimization
• To minimize a function is to maximize the negative of the function.
Example 1 – Minimization
2 x1 x 2 1
Minimize Z x1 2x 2 subject to x1 x2 2
x ,x 0
Solution: 1 2
Construct a Simplex Table,
B x1 x2 s1 s2 t1 t 2 W R B x1 x 2 s1 s2 t1 t2 W R
t1 2 1 1 0 1 0 0 1 t1 1 1 0 1 0 1 0 2
t 2 1 1 0 1 0 1 0 2 t 2 1 0 1 1 1 1 0 1
W 1 3M 2 2M M M 0 0 1 3M W 3 0 0 2 M 2 M 1 4
The minimum value of Z is −(−4) = 4.
The Dual
Example 1 – Finding the Dual of a Maximization Problem
Find the dual of the following: x1 2 x 2 10
Maximize Z 3 x1 4 x2 2x3 subject to 2x 2x x 10
1 2 3
where x1, x 2 , x3 0
Solution: 3 x1 x 2 2
The dual is minimize W 10 y 1 10 y 2subject to x1 x 2 1
4 x x 3
1 2
where y 1, y 2 0
Example 3 – Applying the Simplex Method to the Dual
Use the dual and the simplex method to maximize Z 4 x1 x2 x3
subject to 3 x1 x2 x3 4 where x1, x 2 , x3 0
x1 x 2 x3 2
Solution: 3y1 y 2 4
y y 1
The dual is minimize W 4 y1 2 y2 subject to 1 2
y 1 y 2 1
The final Simplex Table is B y 1 y 2 s1 s2 s3 W R
y2 0 1 41 0 34 0 1
4
s2 0 0 21 1 21 0 5
2
y1 1 0 41 0 41 0 5
4
The minimum value of W is 11/2 . W 0 0 32 0 21 1 11
2
Bibliography and further readings
• Haeussler, E. F., Paul, R. S. & Wood, R. J. (2015). Matemáticas para
administración y economía: Pearson Educación. (Textbook)
• Anderson, D. R., Sweeney, D. J., Williams, T. A., Camm, J. D., &
Cochran, J. J. (2015). Quantitative Methods for Business: Cengage
Learning