[Bonmin] understanding bonmin output
Stefan Vigerske
stefan at math.hu-berlin.de
Sat Jul 4 05:05:07 EDT 2009
Hi,
> Hi, I am solving a MINLP with bonmin. I am using B-Hyb
> and the example file which is given with the software is my template.
When I remember right, then you had a nonconvex MINLP. In this case,
using B-BB might be a better choice, or try Couenne...
> The program ends catching TNLPSolver::UnsolvedError
> and below you can see the output from the last two steps. It finally speaks
> about "small primal infeasibility but I know that there is at least one
> "trivial solution"; I mean there is an array which fulfills all
> constraints. Can
> anyone, plese, give me a suggestion to better understand this output?
There could be (at least) two reasons why Ipopt does not solve this NLP
subproblem:
1. The NLP is nonconvex (since your MINLP is nonconvex), so Ipopt cannot
guarantee you that it finds a feasible solution.
2. The NLP that Ipopt tries to solve is generated by fixing some integer
variables. This fixation might yield an infeasible NLP.
> By the way, I put:
>
> print_solution yes
>
> in options file but I can't see any solution array in the output, How can I
> retrieve it?
I guess Bonmin would need to find a feasible solution first before it
can print one ;-).
> Shall I use:
>
> bb.bestSolution();
>
> to retrive the pointer to the first element?
Yes, but check if bb.bestSolution() is not NULL first.
Stefan
More information about the Bonmin
mailing list