-1

Problem with Infesiable result Cplex

Hi, I have a iterable problem using a few mutable param, en some case the first result if infeseable. I have the problem that when the problem made the first iteration it stars a while and the result doesnt change so thtat result doesnt reach a optimal solution. I think probably if i use set the solve cplex in a way that the infeasible solution if not to bad may i can get a better solution that with some iteration reach a optimal solution.

Does someone know which options i can set i cplex solver to get a better solution?

Leonel Henríquez
  • 9
  • 4

1 Answers1

0

I think that you are missing the point here. If CPLEX says that the problem model is infeasible then there is NO solution. It is not the case that the infeasible 'solution' could be better because there is NO solution.

A way to get 'infeasible' solutions is to add some kind of penalty costs/variables to add some slack into your model and allow solutions that should be infeasible to be accepted with some high cost. Then you can try to minimise those penalty costs to get 'nearer' to a properly feasible solution of your original problem.

TimChippingtonDerrick
  • 2,042
  • 1
  • 12
  • 13
  • ok thanks, but wich options can i set in the solver to add that penalty cost/variable?? – Leonel Henríquez May 19 '21 at 21:00
  • If you are using OPL then you can set it so that it will automatically relax constraints if you name those constraints. I don't like doing it that way as you don't hve as much control over when and how things get relaxed. A better way is to explicitly add some slack variables into your model and add terms into your objective that penalise those slack variables. That way you can use your knowledge of the model to allow it to relax the constraints in some meaningful way. Add slack variables like this: a + b <= 10 could become: a + b <= 10 + s and add a term like 1000*s to the objective – TimChippingtonDerrick May 21 '21 at 07:08