Piecewise optimization gurobi

Question

Following is a piecewise function. I need to optimize it. q is the decision variable while the objective is to minimize the overall function subject to the constraint that it never gets zero.

How can I optimize this piecewise. all other variables other than q are constant values.

score 0 · Answer 1 · answered Jun 12 '18 at 12:31

0

Using y(t,m,i) = q(t,m,i) + eps/MI, and ln(a/b)=ln(a)-ln(b) there are two terms to worry about:

 f(y(t,m,i)) = y(t,m,i) ln y(t,m,i)

and

 g(y(t,m,i),y(t-1,m,i)) =  y(t,m,i) ln y(t-1,m,i)

The first one is easy. The second one not so much, see link for some notes.

answered Jun 12 '18 at 12:31

Erwin Kalvelagen

15,677
2
14
39

How is the first one easy?? – Danial Jalil Sep 02 '18 at 16:34
I assumed it was somewhat obvious that piecewise approximation of f(x,y) is much more difficult than of f(x). – Erwin Kalvelagen Sep 02 '18 at 17:30
I am sorry to say, that is quite a nonsensical piece of code. `setPWLOb` is not supposed to return anything. Please consult the Gurobi documentation and the examples. I think you are missing some concepts here. It may be a good idea to talk to your teacher. – Erwin Kalvelagen Sep 03 '18 at 02:11
how does a the setPWLObj is linked with overall cost? – Danial Jalil Sep 03 '18 at 10:41
Can you give me some comments like the ones mentioned in the answer of this link . https://groups.google.com/forum/#!topic/gurobi/YUWqUfk3PSc – Danial Jalil Sep 03 '18 at 11:14

Piecewise optimization gurobi

1 Answers1