Originally posted by: tazingerurl

Hi,

how is it possible to fullfill the following constraint:

(x <= 5 && x > 0) => y == 1;

(x > 5) => y == 2;

Considering the operator x > 0, the program says, that this operator is not available.

Could you give me an alternative way to solve my problem?

Edit: The dvar is a float, so the program says, that (x <= 5 && x != 0) => y == 1; is not possible as well.

King regards

Hi,

could

dvar float x;
 dvar float y;
 
 float epsilon=0.0000001;
 
 subject to
 {
  (x <= 5 && x >= epsilon) => (y == 1);

(x >= 5+epsilon) => (y == 2);

}

help ?

regards

Originally posted by: JorisK

Hi,

Basically what Alex is saying is that you cannot do '>', '<', or '<>' in linear programming. Instead, use something like '>= value + epsilon' to obtain '> value'. For the remainder of your question, have a look at piecewise linear functions and how you can model them. 

What I often do is refer to the AIMMS modeling guide as it has quite a lot modelling tricks and examples, including some examples on how to do piecewise linear functions:

http://www.aimms.com/downloads/manuals/optimization-modeling

Originally posted by: tazingerurl

Alright,

i tried it by adding a small value, i guess it works.

Thanks to both of you, much appreciated!