[Cbc] Stopping at a feasible solution

[Stefan Vigerske]

Hi,

On 10/26/2012 08:12 PM, Victor Miller wrote:
> I have a collection of hard MIP's for which I'm only interested in finding
> a feasible solution (I have reason to believe that there is only one
> feasible solution, but I haven't proved it).  Since the objective doesn't
> matter, I have been making it identically 0.  However, after reading some
> recent papers on randomization, I've come to believe that it might be
> better to choose a random objective (say with all the coefficients in
> [0,1]).  However, I'd like CBC to stop as soon as it finds a feasible
> solution, and not continue to try to prove optimality.  I assume that I
> would do this with some call back, but what are the details?

You can also just set the absolute or relative gap tolerances to a very 
large value or the solution limit to 1.

Stefan

>
> Victor
>
>
>
> _______________________________________________
> Cbc mailing list
> Cbc at list.coin-or.org
> http://list.coin-or.org/mailman/listinfo/cbc
>