[Scip] Set a primal solution

[Gregor Hendel]

Hi Rotislav,

SCIP restarts the solution process by calling SCIPtransformProb() on
your potentially modified problem. The transformation process
checks all solutions for (in)feasibility and discards the infeasible ones.

I just did so by solving a toy problem and manually changing a variable
bound afterwards to a value different from the optimal solution value.
The interactive shell provides the possibility to change your problem
manually by typing <change> and following the instructions.

Cheers,
Gregor



Am 11.07.2013 13:58, schrieb Rostislav Stanek:
> Dear Gregor,
> thank you for your response. I have then a theoretical question: If I
> add after the freeing process a new global constraint and the solution
> gets infeasible, will it be dropped automatically before starting the
> solving process again? This is not my situation (my solution remains
> feasible), but it could be useful to know it in the future.
> Thank you again.
> Best regards,
> Rostislav Stanek
>
>
> Dne 11.7.2013 13:17, Gregor Hendel napsal(a):
>> Hi Rotislav,
>>
>> when you invoke SCIPfreeTransform(), found solutions are
>> retransformed and thus "lifted" to the original space, so that they
>> survive
>> the freeing process.
>>
>> regards,
>> Gregor
>>
>> Am 11.07.2013 11:58, schrieb Rostislav Stanek:
>>> Dear Michael,
>>> excuse me, that I am writing again, but I have a small additional
>>> question: If I use the second proposed option (I do not want that
>>> the heuristic runs more times) and then call “SCIPfreeTransform” and
>>> after that add some global constraints, do I have to set the
>>> solution again also if I know that the solution remains feasible? Or
>>> do I have to check the solution again via “SCIPtrySol()” /
>>> “SCIPtrySolFree()”?
>>> Thank you again in advice.
>>> Best regards,
>>> Rostislav Stanek
>>>
>>> Dne 10.7.2013 17:30, michael.winkler at zib.de napsal(a):
>>>> Hi Rostislav,
>>>>
>>>> you can write a heuristic plugin setting the timing to
>>>> SCIP_HEURTIMING_BEFOREPRESOL. This heuristic will then be called only
>>>> before presolving of the whole problem (might be multiple times when a
>>>> restart happens).
>>>> Take a look at the heur_trivial.c for further details.
>>>>
>>>> When creating a solution via SCIPcreateSol() the solution vector will
>>>> contain zero values for all variables implicitly.
>>>>
>>>> You then need to use SCIPsetSolVal()/SCIPsetSolVals() to set solution
>>>> values for the variables.
>>>>
>>>> Afterwards please use SCIPtrySol()/SCIPtrySolFree() which will
>>>> check the
>>>> correctness of the solution and if so SCIP will store the solution and
>>>> also update the primalbound and cutoffbound. (So, there is no need
>>>> to add
>>>> you extra "bound constraint" anymore, when adding a solution.)
>>>>
>>>> Please take a look at:
>>>> http://scip.zib.de/doc/html/HEUR.shtml
>>>> http://scip.zib.de/doc/html/scip_8h.shtml#a03da404f4153d8f673e87fb28bc7d552
>>>>
>>>> http://scip.zib.de/doc/html/scip_8h.shtml#adbb9f3c0c0b462481fe5eff677463258
>>>>
>>>> http://scip.zib.de/doc/html/scip_8h.shtml#ab4007453cafad788430c2ad55129479f
>>>>
>>>> http://scip.zib.de/doc/html/scip_8h.shtml#ae4041777d638e0bcbbb592946b520f15
>>>>
>>>> http://scip.zib.de/doc/html/scip_8h.shtml#af00de2379f996445b4497db79e7449f0
>>>>
>>>>
>>>> Best, Michael
>>>>
>>>>> Dear SCIP community,
>>>>> I excuse me for this question which should be clear for me, but how
>>>>> should I correctly set a primal solution before starting the solving
>>>>> process? I solve an ILP using the SCIP solver and before starting the
>>>>> process I would like to set a heuristical primal solution which would
>>>>> also define a bound (this heuristic should not be called during the
>>>>> solving process more). I set the bound as a global constraint, but
>>>>> I do
>>>>> not know how to set the solution. Can I use “SCIPcreateSol”?
>>>>> And what
>>>>> does it mean “creates a primal solution, initialized to zero”?
>>>>> Thank you very much in advance.
>>>>> Yours faithfully,
>>>>> Rostislav Stanek
>>>>> _______________________________________________
>>>>> 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
>>
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://listserv.zib.de/mailman/private/scip/attachments/20130711/9f264b4b/attachment.html