Artificial Variable Techniques Big M-method

Lecture 6 Abstract If in a starting simplex tableau, we dont have an identity submatrix (i.e. an obvious starting BFS), then we introduce artificial variables to have a starting BFS. This is known as artificial variable technique. There are two methods to find the starting BFS and solve the problem the Big M method and two-phase method. In this lecture, we discuss the Big M method.

Suppose a constraint equation i does not have a slack variable. i.e. there is no ith unit vector column in the LHS of the constraint equations. (This happens for example when the ith constraint in the original LPP is either or = .) Then we augment the equation with an artificial variable Ri to form the ith unit vector column. However as the artificial variable is extraneous to the given LPP,

we use a feedback mechanism in which the optimization process automatically attempts to force these variables to zero level. This is achieved by giving a large penalty to the coefficient of the artificial variable in the objective function as follows:
Artificial variable objective coefficient

= - M in a maximization problem, = M in a minimization problem where M is a very large positive number.

Consider the LPP:

Minimize

z 2 x1 x2

Subject to the constraints

3 x1 x2 9 x1 x2 6 x1 , x2 0

Putting this in the standard form, the LPP is:

Minimize

z 2 x1 x2
9

Subject to the constraints

3 x1 x2 s1 x1 x2 x1 , x2 , s1 , s2 0

s2 6

Here s1, s2 are surplus variables. Note that we do NOT have a 2x2 identity submatrix in the LHS.

Introducing the artificial variables, R1, R2, the LPP is modified as follows: Minimize

z 2 x1 x2 MR1 MR2

Subject to the constraints 3 x1 x2 s1 R1 9 x1 x2 s2 R2 6 x1 , x2 , s1 , s2 , R1 , R2 0

Note that we now have a 2x2 identity submatrix in the coefficient matrix of the constraint equations.

Now we solve the above LPP by the Simplex method.

Basic z

x1

x2

s1

s2

R1

R2 0 -M

Sol. 15M 0

-2+4M -1+2M -M -2 -1 0

-M 0 0 -M

R1
R2 z x1

0
0 1 0

3
1 0 1

-1

0
1 0 0

9
6 6+3M 3

1 0 -1 0 -1/3+ -2/3+ -M 2/32M/3 M/3 4M/3 1/3 -1/3 0 1/3

R2 z
x1 x2

0 1
0 0

0 0
1 0

2/3 0
0 1

1/3

-1

-1/3

-1/2 -1/2 1/2M 1/2-M 15/2

-1/2 1/2 1/2 1/2 -1/2 3/2 3/2 9/2

-3/2 -1/2

Note that we have got the optimal solution to the given problem as

3 9 x1 , x2 2 2 15 Optimal z = Minimum z 2
It is illuminating to look at the graphical solution also.

(0.9)

SF
(0,6)

3 9 ( , ) 2 2

(0,0)

(3,0)

(6,0)

z minimum at

3 9 ( , ) 2 2

Problem 5(a) Problem Set 3.4A Page 97

Maximize

z 2 x1 3x2 5 x3

Subject to the constraints

x1 x2 x3 7 2 x1 5 x2 x3 10 x1 , x2 , x3 0

Introducing surplus and artificial variables, s2, R1and R2, the LPP is modified as follows: Maximize

z 2 x1 3 x2 5 x3 MR1 MR2
Subject to the constraints x1 x2 x3 R1
2 x1 5 x2 x3 s2 x1 , x2 , x3 , s2 , R1 , R2 0 7 R2 10

Now we solve the above LPP by the Simplex method.

Basic z

x1

x2

x3

s2

R1 0 M 1

R2 0 M 0 1 1+ 3M/2 -1/2 1/2

Sol. -17M 0 7 10 10 2M 2 5 102/7 4/7 45/7

z R1 R2 z

1 0 0 1

-2-3M -3+4M 5-2M M -2 -3 5 0 1 2 1 -5 -8 7M/2 7/2 -5/2 0 1 0 1 1 6M/2 1/2 1/2 50/7 1/7 6/7 0

0 0 1
0 0 1

R1 x1
z x2 x1

0 0
1 0 0

-1 0 -1 - 0 M/2 1/2 1 -1/2 0

1/7 16/7 + -1/7 M +M 1/7 2/7 -1/7 -1/7 5/7 1/7

The optimum (Maximum) value of

z = 102/7
and it occurs at x1 = 45/7, x2 = 4/7, x3 = 0

Remarks

If in any iteration, there is a tie for entering variable between an artificial variable and other variable (decision, surplus or slack), we must prefer the non-artificial variable to enter the basis. If in any iteration, there is a tie for leaving variable between an artificial variable and other variable (decision, surplus or slack), we must prefer the artificial variable to leave the basis.

 If in the final optimal tableau, an artificial variable is present in the basis at a non-zero level, this means our original problem has no feasible solution.

Problem 4(a) Problem Set 3.4A Page 97

Maximize

z 5 x1 6 x2

Subject to the constraints

2 x1 3 x2 3 x1 2 x2 5 6 x1 7 x2 3 x1 , x2 0

Introducing slack and artificial variables, s2, s3, and R1, the LPP is modified as follows:
Maximize

z 5 x1 6 x2 MR1
3 5 s3 3

Subject to the constraints

2 x1 3 x2 R1 x1 2 x2 6 x1 7 x2 s2

x1 , x2 , R1 , s2 , s3 0

Basic z

x1

x2

R1 0 M

s2 0 0

s3 0 0

Sol -3M 0

-5+2M -6-3M -5 -6

R1
s2

0
0

-2
1

3
2

1
0

0
1

0
0

3
5

s3
z R1 s2

0
1 0 0

6
1/7+ 32M/7 -32/7 -12/7

7
0 0 0

0
0 1 0

0
0 0 1

6/7+ 18/73M/7 12M/7 -3/7 12/7 -2/7 29/7

x2

6/7

1/7

3/7

This is the optimal tableau. As R1 is not zero, there is NO feasible

Problem 4(d) Problem Set 3.4A Page 97

Minimize

z 4 x1 6 x2

Subject to the constraints

2 x1 3 x2 3 4 x1 5 x2 10 4 x1 8 x2 5 x1 , x2 0

Introducing the surplus and artificial variables, R1, R2, the LPP is modified as follows:
Minimize z 4 x1 6 x2 M R1 M R2 M R3

Subject to the constraints 2 x1 3 x2 R1 3 4 x1 5 x2 s2 R2 10 4 x1 8 x2 s3 R3 5 x1 , x2 , s2 , s3 , R1 , R2 , R3 0

Basic z

x1

x2

s2

s3 -M 0

R1 0 -M

R2 0 -M

R3 Sol. 0 18M -M 0

-4+6M -6+16M -M -4 -6 0

R1
R2

0
0

-2
4

3
5

0
-1

0
0

1
0

0
1

0
0

3
10

R3
z R1 R2

0
0 0 0

4
-1-2M -7/2 3/2

8
0 0 0

0
-M 0 -1

-1

0
0 0 1

-3/4 0 +M 3/8 5/8 1 0

3/4 15/4 -2M +8M -3/8 9/8 -5/8 55/8

x2

1/2

-1/8

1/8 5/8

Basic z

x1

x2

s2

s3

R1

R2

R3 Sol.

z R1
R2

1 0
0

-1-2M -7/2
3/2

0 0
0

-M 0
-1

-3/4 0 +M 3/8
5/8

0 0
1

3/4 15/4 -2M +8M -3/8 9/8

-5/8 55/8

1
0

x2
z s3

0
1 0

1/2
-8 + 22M/3 -28/3

1
0 0

0
-M 0

-1/8
0 1

1/8 5/8
-M -1 6 +5M 3

2 0 -8M/3 8/3 0

R2 x2

0 0

22/3 -2/3

0 1

-1 0

0 0

-5/3 1/3

1 0

0 0

5 1

Basic z

x1

x2

s2

s3

R1

R2

R3 Sol.

z
s3 R2 x2 z s3

1
0 0 0 1 0

-8 + 22M/3
-28/3 22/3 -2/3 0 0

0
0 0 1 0 0

-M
0 -1 0 -6/11

0
1 0 0 0

2 0 -8M/3
8/3 -5/3 1/3 0 1 0

-M
-1 0 0

6 +5M
3 5 1
126 11
103 11 15 22

-14/11 1

2/11 12/11 -M - M -M 6/11 14/11 -1

x1
x2

0
0

1
0

0
1

-3/22
-1/11

0
0

-5/22 3/22 0
2/11 1/11 0

16 11

This is the optimal tableau. The Optimal solution is:

15 16 x1 , x2 22 11
126 And Optimal z = Min z = 11