Finding Linear programming optimal solution.

266 Views Asked by At

Consider the following linear program:

max $z = 4x_1+x_2+5x_3+3x_4$

subject to
$x_1-x_2-x_3+3x_4 \le 1$

$5x_1+x_2+3x_3+8x_4\le55$

$-x_1+2x_2+3x_3-5x_4\le3$

It is claimed that the solution $x^*=(0, 14, 0, 5)$ is an optimal solution to the problem. Give a proof of the claim.

I have to proof the problem without using the simplex method to solve this problem. How can it be possible?

1

There are 1 best solutions below

6
On BEST ANSWER

Your objectiva value is $29$. Hence if we can exhibit a feasible solution with the same objective value in the dual problem then we are done.

Let's write down the dual:

$$\min p_1+55p_2+3p_3$$

subject to

$$p_1 + 5p_2-p_3 = 4 $$ $$-p_1+p_2+2p_3=1$$

$$-p_1+3p_2+3p_3=5$$

$$3p_1+8p_2-5p_3=3$$

$$p \ge 0$$

Also, from complementary slackness condition we know that $p_2=0$.

We can easily check that the linear system has no feasible solution. Hence the claimed solution can't be optimal.


View the original problem as

$$\max c^Tx$$

subject to $$Ax \le b$$

$A$ is of full row rank and the nullity is $1$, that is there is this vector $d \ne 0$ such that $Ad=0$.

If we compute $c^Td\ne 0$, then we have found a direction that can increase the objective value indefinitely.