CPLEX giving solution to MILP problem while COUENNE doesn't give any solution to MINLP

[Rupsana Parveen]

Hi all,

 

I am solving a two stage stochastic programming problem where my initial problem was non-convex (MINLP). I converted MINLP to equivalent deterministic MILP and solved it using AMPL CPLEX (output attached).

 

However, when I try to solve MINLP using AMPL COUENNE , I don’t get any solutions (all the objective values, decision variables are 0). Even though the problem is same (i.e.  Parameters, variables, objective function, and constraints are same). Output is attached.

 

I first thought its happening because of time limit. I then increased time limit from 15 minutes to 2 hours in COUENNE . Still same output (no solutions, all are 0).

I checked all of my linearization terms. Seems like they are correct.

 

Can anyone tell me what’s wrong with this MINLP based on output? Thanks in advance.

[Robert Fourer]

All of your Couenne runs indicate "Exiting on maximum time" and "Partial search - best objective 1e+50" which suggests that Couenne could not find a feasible solution before the time limit was reached. Also judging from the timings reported, it appears that you were not successful in increasing the time limit; if you can't figure out how to do this from the Couenne documentation, you might need to write to Couenne support for help.

It does happen sometimes that applying a global solver like Couenne to a MINLP takes a lot longer than applying CPLEX to an equivalent MILP.

Bob Fourer
am...@googlegroups.com

=======

[Pietro Belotti]

The only way to change the time limit is to modify the couenne.opt option file that must reside in the same directory where Couenne is launched. In your case the time limit would be set with the following line:

time_limit 7200

As Bob said, though, Couenne might find it much harder than MILP solvers to find a feasible solution, depending on the problem.

Regards,
Pietro