0

enter image description here

This problem is from the book Luenberger "Linear and Non Linear Optimization". I am facing difficulty with this problem. I am trying to follow this logic -

Let $t = \max (c_1^Tx+d_1,....,c_p^Tx+d_p)$. Then as $t$ is the maximum value could I re-write the constraints as this ?

$\min t$

subject to

$c_1^Tx+d_1 \geq t, ....., c_p^Tx+d_p \geq t, Ax = b , x \geq 0 $

Is this correct. or else please tell me where I am going wrong.

roni
  • 197

1 Answers1

2

You are almost correct.

It should have been than $$c_i^Tx+d_i \leq t$$

as we want $t$ to be an upper bound of $c_i^Tx+d_i$ and the maximal value will be attained as we minimize $t$.

Siong Thye Goh
  • 149,520
  • 20
  • 88
  • 149
  • Dear me ! That was quite silly of me. I mistakenly gave the sign wrong. By the way could you please kindly answer this one too. http://mathoverflow.net/questions/244261/convert-absolute-form-into-linear-programming-problem – roni Jul 13 '16 at 18:55