[Scip] solving pseudo boolean optimization problem

[michael.winkler at zib.de]

Hi,

it seems that you already posted here:
https://www.or-exchange.org/questions/11617/pseudo-boolean-optimization-suites

and I think you got good answers. It seems that the cutting planes are
helping, but doesn't work well within SCIP (maybe try different zerohalf
and clique cuts settings).

Michael

P.S. You can send your model(s) to the pbeva.computational-logic.org
committee and see who will perform best on your instance(s).

> Dear Chivalry,
>
> there is a paragraph in the FAQ <http://scip.zib.de/doc/html/FAQ.php>
> about how to change the behavior of SCIP. I would go for aggressive
> heuristics and leave the rest unchanged for a first shot. Keep in mind
> that changing the settings for proving optimality, i.e. reducing the
> effort in keeping the search tree size small, might make it harder or
> even impossible to succeed on the primal solution side.
>
> I am not sure whether there are other techniques specific for
> pseudo-boolean optimization,
> maybe the experts for PBO on this mailing list can give further advice.
>
> Kind regards,
> Gregor
>
>
>
>
>
> Am 10.03.2015 um 23:24 schrieb key01023 at gmail.com:
>> Hi,
>>
>> Basically I am trying to solve this problem in pseudo boolean
>> optimization format (.opb) with 18 binary variables,
>>
>>
>>
>> The problem is, no matter what setting I used (maximize heuristics
>> frequency etc.) I always arrive at primal optimum (not proved yet) at
>> 20s and used another 40s+ to prove it
 I am wondering whether this is
>> the state of art speed already

>>
>> On the second hand, we have the objective of solving much larger
>> systems, say, such system with 100 variables+. We know that it will
>> take forever to prove optimality, but would there be some settings I
>> could use in SCIP, so that SCIP would simply use all its time trying
>> to find better solution rather than proving optimality? Thank you:D
>>
>> With regards,
>> Chivalry
>>
>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> http://listserv.zib.de/mailman/listinfo/scip
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
>