[Scip] More solutions
Julio Rojas
jcredberry at gmail.com
Thu Nov 19 16:12:49 MET 2009
Thank you. I have never written a piece of C code and I'm in a tight
schedule right now. I wanted to prove some convexity condition I thinnk is
hidden behind all possible solutions this problem has. I know all solutions
have the same value, but this is because this is a simplified version of a
much more complex problem. In the more complex problem, these solutions have
a different value and I don't know if the first solution obtained in the
simpler model is the best one. I have made an heuristic algorithm that
solves this problem, but I would really like to see if by picking a
different solution (not the first one) I converge faster to the point I want
to get. I think this will have to wait until I have the time to look for
help in C coding.
Sorry for all the bothering and thank you very much for all your help and
for an excelent program. Best regards.
-------------------------------------------------
Julio Rojas
jcredberry at gmail.com
On Thu, Nov 19, 2009 at 3:55 PM, Tobias Achterberg <achterberg at zib.de>wrote:
> First, SCIP does not have a "populate" feature like CPLEX. That means, if
> the model is solved to optimality, you cannot get more solutions
> (alternative optima or inferior solutions).
>
> SCIP collects all solutions that it finds during the solving process. With
> the "write solution" command, you can only write the current incumbent
> (i.e., best known solution).
>
> Using the callable library, you could also access the other solutions, but
> I think that this is not possible with the interactive version of SCIP.
>
> If you are not afraid of implementing stuff on your own, it is very easy to
> extend SCIP to do what you want, i.e., to add an interactive command like
> "write allsolutions". You just need to write a dialog handler plugin (look
> at src/scip/dialog_default.c how to do this). The SCIP team can certainly
> help you with this if you have any questions.
>
>
> Tobias
>
>
> Julio Rojas wrote:
>
>> Maybe I'm doing something wrong. Let me write the commands I'm issuing:
>>
>> read problem.lp
>> set limits solutions 1
>> optimize
>> write solution sol1.txt
>> read problem.lp
>> set limits solution 2
>> optimize
>> write solution sol2.txt
>>
>> sol1.txt and sol2.txt should be different, as the optimizer is telling
>> me it have found two solutions, but that's not the case as both
>> solutions are the same (diff sol1.txt sol2.txt).
>>
>> If I do this:
>>
>> read problem.lp
>> set limits solutions 1
>> optimize
>> write solution sol1.txt
>> set limits solution 2
>> optimize
>>
>> Then the solver tells me the model is already solved and the same
>> solution appears.
>>
>> So, maybe there is just one solution even if the solver says there are
>> two? Or I'm not doing the correct procedure to get the second solution.
>>
>> Thank you so much for your time and patience.
>> -------------------------------------------------
>> Julio Rojas
>> jcredberry at gmail.com <mailto:jcredberry at gmail.com>
>>
>>
>>
>> On Thu, Nov 19, 2009 at 2:45 PM, Stefan Vigerske
>> <stefan at math.hu-berlin.de <mailto:stefan at math.hu-berlin.de>> wrote:
>>
>> Hi,
>>
>> I think the idea was to set the limit to one before starting
>> optimization. Then it should stop when the limit is reached, and so
>> on...
>>
>> Stefan
>>
>> Am 19.11.2009 14:22, schrieb Julio Rojas:
>> > I tried to do what you suggest me. In one problem, after
>> optimization, the
>> > report is the following:
>> >
>> > SCIP Status : problem is solved [optimal solution found]
>> > Solving Time (sec) : 0.07
>> > Solving Nodes : 1
>> > Primal Bound : +5.20000000000000e+01 (2 solutions)
>> > Dual Bound : +5.20000000000000e+01
>> > Gap : 0.00 %
>> >
>> > So, in theory, I have two solutions. If I check the value of the
>> solution
>> > limit it shows:
>> > SCIP> set limits solution
>> > current value: -1, new value [-1,2147483647]: 2
>> > parameter <limits/solutions> set to 2
>> >
>> > It doesn't matter if I do it before or after optimization, when I
>> run
>> > optimize again it tells me the problem has already been solved
>> and no new
>> > solution is presented.
>> >
>> > Thanks for your kind help.
>> > -------------------------------------------------
>> > Julio Rojas
>> > jcredberry at gmail.com <mailto:jcredberry at gmail.com>
>> >
>> >
>> >
>> >
>> > _______________________________________________
>> > Scip mailing list
>> > Scip at zib.de <mailto: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
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.zib.de/mailman/private/scip/attachments/20091119/ac70a550/attachment.html
More information about the Scip
mailing list