The revised Simplex

method

Ha Thi Xuan Chi, PhD

1
Outline chapter
Intr

2
Revised Simplex
Solving for Basic Feasible Solution

• The n nonbasic variables from the n+m elements

of are set to zero.
x
A, I   b
x s 

x
x   0
 s

3
Revised Simplex
Solving for Basic Feasible Solution

• The set of m equations with m basic variables denoted by BxB=b

where  xB1  x

 
Vector of basic variables xB=  xB 2  obtained by eliminating the nonbasic variables from x 
 xBm 
 s
 B11 B12 B1m 
Basis matrix B = B B22 B2 m 
obtained by eliminating the columns corresponding to
 21 coefficient of nonbasic variables from A, I 
 Bm1 Bm 2 Bmm 

• To solve BxB = b both side are premultiplied by B-1 B-1BxB=B-1b

xB=B-1b
or
then Z = cBxB= cBB-1b

4
Revised Simplex
Solving for Basic Feasible Solution
 Max Z = 3x1 + 5x2 1 0 1 0 0
[ A, I ]  0 2 0 1 0
 s.t: x ≤4
1
 2x2 ≤ 12 3 2 0 0 1
 3x1 + 2x2 ≤ 18 4
 x1, x2 ≥ 0  x3   
 x  b  12
 x1  xs =
 4 18
In this case x=
x   x5 
c = [3, 5]  2
5
Revised Simplex Solving for Basic Feasible Solution

• Iteration  x3  1 0 0  4 
4
1 0 0
0  x   0 1 0 12  12
0 
B  1 0 =B-1 s  4      
0 0 1  x5  0 0 1 18 18
o
4
 x3  12
x  Z = [0, 0, 0]   =0
 4 18
xB =  x5 

cB = [0, 0, 0]

6
Revised Simplex Solving for Basic
Feasible Solution
• Iteration 1
 x3  1 0 0 1 0 0
xB =  x  B  0 2 0 B-1  0 1 / 2 0
 2
 x5  0 2 1 0  1 1
cB = [0, 5, 0]

 x3  1 0 0  4   4
 x   0 1 / 2 0 12  6
so  2       
 x5  0  1 1 18 6
 4
6 
so Z = [0, 5, 0]   = 30
6

7
Revised Simplex Solving for Basic
Feasible Solution

• Iteration 2

 x3  4  2
 x3  1 0 1 1 1 / 3  1 / 3 1 1 / 3  1 / 3
xB =   B  0 2 0 B-1  0 1 / 2 0  so  x2   0 1 / 2 0  12  6
 
 x2      0  1 / 3 1 / 3 
 
 x1  0 2 3 0  1 / 3 1 / 3   x1   182
cB = [0, 5, 3]  2
6 
so Z = [0, 5, 3]   = 36
2

8
Matrix Notation
Maximize Z= cx  a11 a12  a1n 
Matrix: a a22  a2 n 
subject to A   21
Ax = b    
x0  
 am1 am 2  amn 

The constrains become:

Row vector: c  c1 , c2 ,..., cn 
Column vectors: x  x 
 x   b
A, I x   0
b1   xn 1   s  s
 x1  0 
x  b  0  x 
x   2 b   2 0  xs   n  2 
    
       
 xn  bn  0   xn  m 

9
Simplex tableaux in matrix form

Basic Z Coefficient of: RHS

variable Original Slack
variable variable
Z 1 cB .B-1 .A - c cB .B-1 cB .B-1 .b

xB 0 B-1 .A B-1 B-1 .b

cB : vector whose elements are the objective

function coefficients (including zero for slack variables)
for the corresponding elements of xB
xB : vector of basic variables
B : basis matrix
B-1 : inverse of basis matrix
B : vector of RHS
10
Simplex tableaux in matrix form
Example
Maximize Z = 3x1 + 5x2 Maximize Z = 3x1 + 5x2
Subject to Subject to
x1 ≤4 x1 + s1 =4
2x2 ≤ 12 2x2 + s2 = 12
3x1 + 2x2 ≤ 18 3x1 + 2x2 + s3 = 18
x1 ≥ 0
x1≥ 0, x2 ≥ 0, s1 ≥0, s2 ≥0, s3 ≥0.
x2 ≥ 0.

4  1 0 
 
c  3, 5 b  12  A   0 2
18   3 2 
11
 Simplex tableaux in matrix form
Iteration 0
Basic variables Basis matrix

 s1  1 0 0 
xB   s2  B  0 1 0 
 s3  0 0 1 
The coefficients of basic variables (s1, s2, s3) in constraints
The coefficient of basic variables (s1, s2, s3) in objective function

C B  0 0 0  1 0 0 
B 1  0 1 0 
0 0 1 

12
Simplex tableaux in matrix form
The coefficients of non basic variables (x1, x2) in constraints
1 0 0  1 0  1 0 
B 1. A  0 1 0   0 2   0 2 
0 0 1   3 2   3 2 

The coefficients of basic variables (s1, s2, s3) in constraints

1 0 0   4   4 
B 1.b  0 1 0  12  12
0 0 1  18 18 
13
Simplex tableaux in matrix form

The coefficients of the basis variables (s1, s2, s3) in objective function

1 0 0 
cB .B 1  0 0 0  0 1 0   0 0 0
0 0 1 

14
Simplex tableaux in matrix form

The coefficients of the non basic variables (x1, x2) in objective function
1 0 
cB .B 1. A  c  0 0 0  0 2  3 5  3  5
 3 2 
Negative
4  Solution is not optimal
Z  cB .B 1.b  0 0 0 12   0
18 

15
Simplex tableaux in matrix form
Determine the entering variable :

Selecting the variable whose the coefficient value is the

most negative (maximize problem) and the positive

coefficient (minimize problem) in Z-eq.

The coefficients of the nonbasic variables (x1, x2):

cB .B 1. A  c  0, 0  3,5  3, 5

The most negative

x2 : entering variable
16
Simplex tableaux in matrix form
Determine the leaving variable:
Basic Z Coefficient of: RHS
variable Original Slack
variable variable
Z 1 cB .B-1 .A - c cB .B-1 cB .B-1 .b

xB 0 B-1 .A B-1 B-1 .b

 Entering variable is the original variable:

Selecting the minimum ratio: B-1 .b/B-1 .A

 Entering variable is the slack variable

Selecting the minimum ratio: B-1 .b/ B-1

17
Simplex tableaux in matrix form
x2 : entering variable and original variable

1 0 0  1 0  1 0 
B 1. A  0 1 0   0 2    0 2 
0 0 1   3 2   3 2 
 
Coefficient values of x2
 
 
 12 
Min ratio  4
1 0 0   4   4  2
 8  s2 : leaving variable
B 1.b   0 1 0  12   12   
2 
0 0 1  18  18 
18
 Simplex tableaux in matrix form
Iteration 1
Basic variables Basis matrix
 s1  1 0 0 

xB   x2   B  0 2 0 
 s3  0 2 1 

The coefficient of basic variables (s1, x2, s3) in objective function

cB  0 5 0 
The coefficients of basic variables (s1, x2, s3) in constraints
1 0 0 
 1 
B  0
1
0
 2 
 0 1 1 
 
19
Simplex tableaux in matrix form
The coefficients of non basic variables (x1, s2) in constraints
1 0 0 
  1 0  1 0 
1
1

B .A  0 0  0 2   0 1
 2 
0 1 1   3 2   3 0 
 
The coefficients of basic variables (s1, x2, s3) in constraints

1 0 0 
   4   4
1
1

B .b  0 0  12   6 
 2 
0 1 1  18  6 
  20
Simplex tableaux in matrix form

The coefficients of the basis variables (s1, x2, s3) in objective function

1 0 0 
 1 
 5 
cB .B  0 5 0 0
1
 
0  0 0
 2   2 
 0 1 1 
 

21
Simplex tableaux in matrix form

The coefficients of the non basic variables (x1, s2) in objective function
1 0 
 5 
1
cB .B . A  c  0 0  0 2   3 5  3 5
 2 
 3 2 

Negative
4 
1  5   Solution is not optimal
Z  cB .B .b  0 0  12   30
 2 
18 

22
Simplex tableaux in matrix form

Determine the entering variable :

The coefficients of the non basic variables (x1, s2) in objective function
1 0 
 5 
1
cB .B . A  c  0 0   0 2   3 5  3 5
 2 
 3 2 
The most negative
x1 : entering variable

23
Simplex tableaux in matrix form
Determine the leaving variable:
Basic Z Coefficient of: RHS
variable Original Slack
variable variable
Z 1 cB .B-1 .A - c cB .B-1 cB .B-1 .b

xB 0 B-1 .A B-1 B-1 .b

 If the entering variable is the original variable

Selecting the minimum ratio: B-1 .b/B-1 .A

 If the entering variable is the slack variable

Selecting the minimum ratio: B-1 .b/ B-1

24
Simplex tableaux in matrix form
1 0 0  x2 : entering variable and original variable

  1 0  1 0 
1
1

B .A  0 0  0 2   0 1 
 2 
0 1 1   3 2   3 0 
 

1 0 0  Coefficient values of x1

   4   4
1
1

B .b  0 0  12   6 
 2  4
0 1 1  18  6  1
 
  Min ratio    2  
6 s3 : leaving variable
 
3 25
Simplex tableaux in matrix form

Iteration 2

Basic variables Basis matrix

 s1  1 0 1 
xB   x2  B  0 2 0 
 x1  0 2 3

26
Simplex tableaux in matrix form

The coefficient of basic variables (s1, x2, x1) in objective function

cB  0 5 3
The coefficients of basic variables (s1, x2, x1) in constraints
 1 
 1 0
3
 
1
B 1  0 0
 2 
 
0 1 1 
 3 

27
Simplex tableaux in matrix form

The coefficients of non basic variables (s1, s2) in constraints

 1 
1 0 3 
  1 0   0 0 
1
1 
B .A  0 0  0 2   0 1 
 2 
   3 2  1 0 
 0 1 1 
 3 

28
Simplex tableaux in matrix form

The coefficients of basic variables (s1, x2, x1) in constraints

 1 
1 0 3  4   2
 
1
1 
B .b  0 0  12   6 
 2 
  18   2 
0 1 1 
 3 

29
Simplex tableaux in matrix form
The coefficients of the basis variables (s1, x2, x1) in objective function

 1 
 1 0
3
 
1  3 
cB .B 1  0 5 3 0 0   0 1
 2   2 
 
0 1 1 
 3 
Non-
negative

30
Simplex tableaux in matrix form

The coefficients of the non basic variables (s1, s2) in objective function
1 0 
cB .B 1. A  c  0 5 3  0 2   3 5  0 0
 3 2 

Non- negative
4 
1  3  
Z  cB .B .b  0 1 12   36
 2 
18 
Solution is optimal

31
 Fundamental Insight
Initial Z Decision variables Slack variables RHS
Table
Z x1 x2 s1 s2 s3 rhs

1 -3 -5 0 0 0 0
t= -c 0
0 1 0 1 0 0 4

0 0 2 0 1 0 12
T= A I b
0 3 2 0 0 1 18

t 1 -c 0 0
=
T 0 A I b
1 1 1
1 cb B A  c cb B cb B b
= 1 1
0 B A B B 1b Foundations-32
 Fundamental Insight
Final Z Decision variables Slack variables rhs
Table
Z x1 x2 s1 s2 s3 Rhs
t* = 1 0 0 0 3/2 y* 1 36 Z*
z*-c
0 0 0 1 1/3 -1/3 2

T* = 0 0 A* 1 0 ½ 0 6 b*
S*
0 1 0 0 -1/3 1/3 2

t* 1 z*-c y* Z*
=
T* 0 A* S* b*

1 y*A-c y* y*b
=
0 S*A S* S*b
 Fundamental Insight
Z Decision variables Slack rhs
variables

* * *
1 y A c y yb
* * *
0 SA S Sb
We can use the fundamental insight for sensitivity analysis.
Vector y* plays a very special role. These are shadow prices.

 
Fundamental Insight

Optimal value: m
Z *  y * b   yi *bi
i 1

Old optimal value:

4 
 3    36
Z *  y1b1  y2b2  y3b3   0 1 12 
 2 
18 
New optimal value:
4 
 3 
Z *  0 1 13   37.5
 2 
18 

Z *  1.5 35
36