You are on page 1of 30

Problem 5

Maximize

Problem Set 3.4B Pages 101-102

z 2 x1 2 x2 4 x3

Subject to the constraints

2 x1 x2 x3 2
3 x1 4 x2 2 x3 8
x1 , x2 , x3 0
We shall solve this problem by two phase
method.

Phase I:
Minimize

r R2

Subject to the constraints

2 x1 x2 x3 s1
3 x1 4 x2 2 x3

s2 R2 8

x1 , x2 , x3 , s1 , s2 , R2 0
Here s1 is a slack variable, s2 is a surplus
variable and R2 artificial variable.

Basic r x1 x2
3
4
r 1 0
0

x3
2
0

s1 s2
0 -1
0 0

R2
0
-1

Sol.
8
0

s1

0 2

R2
r

0 3
1 0

4
0

2
0

0
0

-1
0

1
-1

8
0

s1

0 5/4

1/2

1/4

-1/4

x2

0 3/4

1/2

0 -1/4

1/4

Phase I has ended. We now start Phase II.

Basic z x1 x2
-1/2 0
z
1 -2 -2

x3
-3
-4

s1 s2
0 -1/2
0
0

R2

Sol.
4
0

s1

0 5/4

1/2

1/4

x2

0 3/4

1/2

0 -1/4

z
x3

1 7
0 5/2

0
0

0
1

6
2

1
1/2

4
0

x2

0 -1/2

-1 -1/2

This is the optimal tableau.


Opt.Sol.: x1 = 0, x2 = 2, x3 = 0. Opt.Value: z = 4

We saw in Phase I, in the starting


iteration, there was a tie for the leaving
variable: namely s1 and R2. We rightly
chose R2 to leave the basis. Now we shall
see what happens if s1 was allowed to
leave the basis.

Basic r x1 x2
3
4
r
1 0
0

x3
2
0

s1 s2
0 -1
0 0

R2
0
-1

Sol.
8
0

s1

0 2

R2

0 3

-1

1 -5

-2

-4

-1

x2

0 2

R2

0 -5

-2

-4

-1

Phase I has ended. But we find the artificial


variable R2 is still in the basis. Since it is at
0 level, we now start Phase II having the
starting BFS as x2 and R2. (That is we have
to keep the R2 column in Phase II also.)

Basic z x1 x2
2 0
z
1 -2 -2

x3
-2
-4

s1 s2
2 0
0
0

R2
0
0

Sol.
4
0

x2

0 2

R2

0 -5

-2

-4

-1

z
x2

1 7
0 -1/2

0
1

0
0

6 1
-1 -1/2

-1
1/2

4
2

x3

0 5/2

-1/2

1/2

We now make R2 leave the basis. Remember that even


though its coefficient in the pivot column is negative, it does
not spoil the feasibility as RHS value is zero.

As R2 has left the basis, we now forget the


R2 column. The remaining tableau is
optimal. Hence the optimal solution is:
x1 = 0, x2 = 2, x3 = 0.
And the optimum value: z = 4 (as before.)

Now we shall look at four special cases


that arise in the application of Simplex
method.
Degeneracy
Alternative optimum solutions
Unbounded solutions
Infeasible solutions

Degeneracy
If in a basic feasible solution (BFS) one of
the basic variables is present at 0 level (i.e.
is zero) then we say we have a degenerate
BFS. This usually happens when there is a
tie for the leaving variable (as we saw in
the previous example). There is nothing
alarming in having a degenerate BFS
except that in some cases it may lead to

Cycling : a phenomenon in which the a


leaving variable again enters the basis and
again leaves repeatedly without optimality
being reached. The Optimization people
claim that this is a rare phenomenon and
normally optimality is reached in a finite
number of steps.
Degeneracy can be permanent as in the
previous example or can be temporary as
the following example shows.

Problem 2 Problem Set 3.5A Page 105


Maximize

z 3 x1 2 x2

Subject to

4 x1 x2 8
4 x1 3 x2 12
4 x1 x2 8
x1 , x2 0

We solve it by Simplex method.

Basic z
z
1

x1
-3

x2
-2

s1
0

s2
0

s3
0

Sol.
0

s1

-1

s2

12

s3
z

0
1

4
0

1
-11/4

0
3/4

0
0

1
0

8
6

x1

-1/4

1/4

s2

-1

s3

-1

Note that the BFS obtained is degenerate as s3 = 0. (We also


observe that there was a tie for the leaving variable.)

Basic z
z
1

x1
0

x2
0

s1
-5/8

s2
0

s3
11/8

Sol.
6

x1

1/8

1/8

s2

-2

x2
z

0
1

0
0

1
0

1/2
1/8

0
17/2

x1

-1/8

3/8

3/2

s1

-2

x2

1/2

-3/2

-1/2 0
0 5/8

This is the optimal tableau. Note that the optimal BFS


obtained is NOT degenerate.

Alternative Optimal Solutions


When the objective function is parallel to a
binding constraint (i.e. a constraint that is
satisfied as an equation), then the objective
function will assume the same optimal value
at more than one corner point. We say the
problem has alternative optimal solutions. In
this case there also will be an infinity of
(non-basic) feasible solutions where the
objective function has the same optimum
value.

Criterion for the existence of alternative


optimal solutions from the optimal tableau:
If ,in the optimal tableau, the coefficient of
a non-basic variable is zero, then that can
enter the basis and the objective function
will not change (why?). Thus we shall have
alternative optimal solutions as the following
example illustrates.

Maximize

z x1 x2

Subject to the constraints


x1 x2 6
x1 2 x2 10
x1 , x2 0

We shall solve this problem by the Simplex


method.

Basic z
z
1

x1
-1

x2
-1

s1
0

s2
0

Sol.
0

s1

s2

10

z
x1
s2

1
0
0

0
1
0

0
1
1

1
1
-1

0
0
1

6
6
4

This is the optimal


tableau. Since
coefficient of the
nonbasic variable
x2=0, alt. opt. soln
exists.

z
1
0
0
1
0
6
x1
0
1
0
2 -1
2
x2
0
0
1
-1
1
4
Note that we have got an alt. opt. soln. We understand this
graphically also in the next slide.

(0,6)
(0,5)

Optimal BFSs
(2,4)

(6,0)

All points on this line


segment (except the
endpoints) are nonbasic
optimal feasible
solutions.
(10,0)

Note that the objective function line is parallel to the first


(binding) constraint line.

Maximize z 2 x1 x2
Subject to the constraints
x1 x2 10
2 x1 x2 40
x1 , x2 0

We shall solve this problem by the Simplex


method.

Basic z
z
1

x1
-2

x2
1

s1
0

s2
0

Sol.
0

s1

-1

10

s2

-1

40

z
x1
s2

1
0
0

0
1
0

-1
-1
1

2
1
-2

0
0
1

20
10
20

z
1
0
0
0
1
40
x1
0
1
0
-1
1
30
x2
0
0
1
-2
1
20
Note that we have got the optimal tableau.

Since the coefficient of the nonbasic variable s1


is zero in the objective function row, we expect
an alternative optimal BFS to exist. We see that
the problem indeed has an infinite number of
alternative feasible solutions which are NOT
basic.
Could you guess them?
Answer:
x1= 30+t; x2 = 20+2t, where t is any
nonnegative number.
We can easily see this graphically also.

(30,20)
All points on this line from
(30,20) onwards are
Optimal nonbasic solutions.

Problem 2
Maximize

Problem Set 3.5B Page 108

z 2 x1 x2 3 x3

Subject to the constraints


x1 x2 5 x3 10
2 x1 x2 3 x3 40
x1 , x2 , x3 0
Show that the above LPP has an infinite number
of Optimal (nonbasic) feasible solutions.

Unbounded solutions
In some LPPs, the values of some decision
variables can be increased indefinitely
without violating any of the constraints,
meaning that the solution space is unbounded
in at least one direction. As a result the
objective value may increase (maximization
case) or decrease (minimization case)
indefinitely. Thus both the feasible solution
space and the objective function are
unbounded.

The solution space is unbounded if in a


Simplex iteration, all the entries in a column
corresponding to a non-basic variable are
negative (or zero) in all the constraint rows
and in the objective function row, the entry is:
(a) negative (or zero) in the case of a
maximization problem
(b) positive (or zero) in the case of a
minimization problem
We illustrate by an example.

Problem 2

Problem Set 3.5C Page 110

Maximize z 20 x1 10 x2 x3
Subject to the constraints
3 x1 3 x2 5 x3 50
x1
x3 10
x1 x2 4 x3 20
x1 , x2 , x3 0
It is clear that the solution space is unbounded
in the x2 direction.

Thus the variable x2 can be increased


indefinitely without violating any of the
constraints and thus the solution space
and the objective function are both
unbounded. (You can form the Simplex
tableau and verify that in x2 column all
the entries in the constraint rows are
negative or zero and that in the z-Row, it
is negative.)

An example where cycling happens


3
1
Maximize z x1 20 x2 x3 6 x4
4
2
Subject to the constraints
1
x1 8 x2 x3 9 x4 0
4
1
1
x1 12 x2 x3 3x4 0
2
2
x3
1
x1 , x2 , x3 , x4 0

You might also like