Revised Simplex Method

Original simplex method calculates and stores all

numbers in the tableau many are not needed.
Revised Simplex Method (more efficient for
computing)
Used in all commercially available packages. (e.g.
IBM MPSX, CDC APEX III)
Max
subject to

Z = cx
Ax b
x 0

Initially constraints become (standard form):

x
xs
xs

= slack variables

Basis matrix: columns relating to basic variables.

B11 . . . . . . . . . . B1m
.. .
.
. ..
.. .
.
. ..
.
.
.
. .
Bm1 . . . . . . . . . . Bmm

(Initially B = I )
Basic variable values:

xB =

xB1
..
..
..
xBm

At any iteration non-basic variables = 0

Therefore

B xB = b
xB = B-1b

B-1 inverse matrix.

At any iteration, given the original b vector and

the inverse matrix, xB (current R.H.S.) can be
calculated.
Z = cBxB where cB = objective coefficients of
basic variables

Step in the Revised Simplex Method

1. Determine entering variable, Xj , with associated
vector Pj .

- compute Y = cBB-1
- compute zj cj = Y Pj - cj for all non-basic
variables.
Choose largest negative value (maximisation).
If none, stop.
2. Determine leaving variable, Xr , with associated
vector Pr .

- compute xB = B-1b (current R.H.S.)

- compute current constraint coefficients of
entering variable:
j = B-1Pj
Xr is associated with
= Min { (xB )k / jk , jk > 0}
k

(minimum ratio rule)

3. Determine next basis i.e. calculate B-1

Go to step 1.

Example:
Max Z = 3X1 + 5X2

4
2X2 12
3X1 + 2X2 18
X1 , X2 0

s.t.

X1

Standard form of constraints:+ S1

= 4
2X2
+ S2
= 12
+ S3 = 18
3X1 + 2X2
X1 , X2 , S1 , S2 , S3 0
X1

xB = B-1b

cB =

Z =

= 1 0 0
0 1 0
0 0 1
0
0

4
12
18

4
= 12
18

0
0

4
12
18

= 0

First iteration
Step 1
Determine entering variable, Xj , with associated
vector Pj .
- compute Y = cBB-1
- compute zj cj = Y Pj - cj for all non-basic
variables.
Y =

z1 c 1 = 0 0 0

1
0
0

0 0
1 0
0 1
1
0
3

- 3

and similarly for z2 c2 = - 5

Therefore X2 is entering variable.

0 0

= -3

Step 2
Determine leaving variable, Xr , with associated
vector Pr .
- compute xB = B-1b (current R.H.S.)
- compute current constraint coefficients of
entering variable:
j = B-1Pj
Xr is associated with
= Min { (xB )k / jk , jk > 0}
k

xB =

4
12
18

2 =

1 0 0
0 1 0
0 0 1

0
0
2 = 2
2
2

= Min { - , 12/2 , 18/2 }

= 12/2
therefore S2 leaves the basis.

Step 3
Determine new B-1
B =

B-1 = 1 0 0
0 0
0 -1 1

1 0 0
0 2 0
0 2 1

Solution after one iteration:

xB = B-1b
= 1 0 0
0 0
0 -1 1
Go to step 1

4
4
12 = 6
18
6

Step 1 (second iteration)

Compute Y = cBB-1
Y =

1
0
0

-1

0
0
1

5/2

- compute zj cj = Y Pj - cj for all non-basic

variables (X1 and S2):X1: z1 c1 = 0 5/2 0

1
0
3

- 3

= -3

S2: z4 c4 = 0 5/2 0

0
1
0

- 0

= 5/2

Therefore X1 enters the basis.

Step 2
Determine leaving variable.
xB =

4
6
6

1 =

1 0 0
0 0
0 -1 1

1
1
0 = 0
3
3

= Min { 4/1 , - , 6/3 }

= 6/3
therefore S3 leaves the basis.

Step 3
Determine new B-1
B =

1 0 1
0 2 0
0 2 3

B-1 = 1
0
0

1/3 -1/3
1/2
0
-1/3
1/3

Solution after two iterations:

xB = B-1b
= 1 1/3
0 1/2
0 -1/3
Go to step 1

-1/3
0
1/3

4
12
18

2
= 6
2

Step 1
- compute Y = cBB-1
Y =

B-1

0 3/2 1

- compute zj cj = Y Pj - cj for all non-basic

variables (S2 and S3):S2: z4 c4 = 0

3/2

0
1
0

- 0

= 3/2

S3: z5 c5 = 0

3/2

0
0
1

- 0

No negatives. Therefore stop.

Optimal solution:
S1* = 2
X2* = 6
X1* = 2
Z* = cB xB =

0 5 3

2
6
2

= 36