Download as pdf or txt
Download as pdf or txt
You are on page 1of 3

MODULE 6 – MINIMIZATION LP PROBLEMS’ TECHNIQUES

Objectives : At the end of the module, the students are expected to,
1. Convert inequalities of minimization into equations; and,
2. Solve the LP problems in using minimization techniques.

Topic 1 – Big-M Technique

Discussion. In the example,

Min xo = 20x1 + 25x2


s.t. 50 x1 + 30 x2 ≥ 1500 ⇒ 5 x1 + 3 x2 ≥ 150 ⇒ 5 x1 + 3 x2 – s1 = 150
20 x1 + 30 x2 ≥ 1200 ⇒ 2 x1 + 3 x2 ≥ 120 ⇒ 2 x1 + 3 x2 - s2 = 120
40x2≥ 800 ⇒ x2≥ 20 ⇒ x2 - s3 = 20
x1, x2 ≥ 0 x1, x2 ≥ 0 x1, x2 ≥ 0

x1 x2 s1 s2 s3 RHS
5 3 -1 0 0 150
2 3 0 -1 0 120
0 1 0 0 -1 20

● In the tableau, the identity matrix In is not present necessary for a canonical form.
Introduce an artificial variable for every column of the identity matrix that is missing.

● Introduce y1, y2, y3 ≥ 0

Objective: Drive out all the artificial variables by making them zero.
x1 x2 s1 s2 s3 y1 y2 y3 RHS
5 3 -1 0 0 1 0 0 150
2 3 0 -1 0 0 1 0 120
0 1 0 0 -1 0 0 1 20

NEW Min xo = 20x1 + 25x2 + My1 +My2 +My3 where M is a very, very big number
which make it unattractive.

x1 x2 s1 s2 s3 y1 y2 y3 RHS
y1 5 3 -1 0 0 1 0 0 150
y2 2 3 0 -1 0 0 1 0 120
y3 0 1 0 0 -1 0 0 1 20
Min 20 25 0 0 0 M M M -(150M+120M+20M)=-290M

● But FCF is violated because the objective function has y1, y2, y3 which are basic.
Substitute out basic variables…

y1 = 150 - 5x1 – 3x2 + s1


y2 = 120 – 2 x1 – 3 x2 +s2 from constraints
y3 = 20 – x2 + s3
Substitute to objective function…

Min xo = 20x1 + 25x2 + M(150 - 5x1 – 3x2 + s1) + M(120 – 2 x1 – 3 x2 +s2) + M(20 – x2 + s3) + 290M
Min xo = (20 – 7M)x1 + (25 – 7M)x2 + Ms1 + Ms2 + Ms3 + 290M

● Now acceptable because the artificial variables are already eliminated from the
objective function.

- 290M = - xo + (20 – 7M)x1 + (25 – 7M)x2 + Ms1 + Ms2 + Ms3

● Proceed by following the simplex method in tableau format.


Table 1
leaves enters discarded
x1 x2 s1 s2 s3 y1 y2 y3 RHS
y1 5 3 -1 0 0 1 0 0 150
y2 2 3 0 -1 0 0 1 0 120
y3 0 1 0 0 -1 0 0 1 20
Min 20-7M 25-7M M M M 0 0 0 -290M

Table 2
leaves enters discard
x1 x2 s1 s2 s3 y2 y3 RHS
x1 1 3/5 -1/5 0 0 0 0 30
y2 0 9/5 2/3 -1 0 1 0 60
y3 0 1 0 0 -1 0 1 20
65−4M 20−2M
Min 0 5 5
M M 0 0 - 600 - 80M

Table 3
Leaves enters discard
x1 x2 s1 s2 s3 y2 RHS
x1 1 0 -1/5 0 3/5 0 18
y2 0 0 2/5 -1 9/5 1 24
x2 0 1 0 0 -1 0 20
20−2M 65−9𝑀
Min 0 0 M 0 - 860 - 24M
5 5

Table 4

x1 x2 s1 s2 s3 RHS
x1 1 0 -1/3 1/3 0 18
s3 0 0 2/9 -5/9 1 31/3
x2 0 1 - 2/15 1/3 0 100/3
10 65 1033
Min 0 0 9 9
0 - 3

𝟏𝟎𝟎 𝟔𝟓 𝟏𝟎𝟑𝟑
optimal! Conclusion: (18, 𝟑 , 0, 0, 𝟗 ) with min xo = 𝟑
● If after elimination the tableau is not yet optimal, continue using simplex method till you
got a solution. This is also applicable for maximization using same rules and process.

Topic 2 – Simplex Minimization Method by Dantzig

● ≥ and = are occasionally used in constraints in the maximization process but are more
common in minimization problems.

For ≥, simply subtract a slack variable, S, but add an artificial variable A to avoid a
negative value if other variables become 0.

For =, add an artificial variable A to avoid falsity in equation once the rest of the
variables become 0.

● Steps in solving a minimization problem. (Similar to maximization problem except for 3


processes)
1). The Cj column of the initial table begins with the coefficients of artificial variables
and slack variables in the objective with positive coefficients in the constraints.
2). Instead of looking for the most positive quantity in Cj – Zj row for the optimum
column, look for the most negative entry.
3). The optimum table or final table has entries in the Cj – Zj which are either zero or
positive.
Illustrating example. Min C = 60x + 50y
s.t. 3x + 5y ≤ 15
4x + 4y ≥ 16
x, y ≥ 0
Solution. New program Min C = 60x + 50y + 0s1 +0s2 + 100A1
s.t. 3x + 5y + s1 = 15
4x + 4y – s2 + A1 = 16
x, y, s1, s2 ≥ 0
● Numerical coefficient of A is at 100’s because the objective function’s highest coefficient
is at 10’s. If 100’s, then at 1000’s.

Initial table. Pivotal row


Cj 60 50 0 0 100
Prod Qty x y S1 S2 A1
0 S1 15 3 5 1 0 0
100 A1 16 4 4 0 -1 1
Zj 1600 400 400 0 -100 100
Cj - Zj -340 -350 0 100 0
Optimum column
Table II Pivotal row
Cj 60 50 0 0 100
Prod Qty x y S1 S2 A1
50 y 3 3/5 1 1/5 0 0
100 A1 4 8/5 0 -4/5 -1 1
Zj 550 190 50 -70 -100 100
Cj - Zj -130 0 70 100 0
Optimum column
Table III

Cj 60 50 0 0 100
Prod Qty x y S1 S2 A1
50 y 3/2 0 1 1/2 3/8 -3/8
60 x 5/2 1 0 -1/2 -5/8 5/8
Zj 225 60 50 -5 -75/4 75/4
Cj - Zj 0 0 5 75/4 325/4

Since there is no more negative values in the Cj –Zj row, the solution is optimum.
Conclusion: x = 5/2 s1 = 0 Min Cost = 225
y = 3/2 s2 = 0

Exercises for submission the next meeting.

Use both Big-M Technique and simplex minimization by Dantzig


to determine the optimal solutions.
Min xo = 6x1 + 16x2
s.t. x1 + x2 = 400
x1 ≤ 150
x2 ≥ 200
x1, x2 ≥ 0

--- END OF MODULE 6 ---

You might also like