[SCIP] Separator not called

[Gerald Gamrath]

Dear Emanuel,

another guess: did you define your variables to be integral and does 
SCIP actually do any branching? If you did not define your variables to 
be integral and also did not add a constraint handler to check solutions 
(e.g., for integrality in the original space), your LP solution after 
pricing might just be declared optimal. If your separator should ensure 
a restriction on your feasible space, it should better be a constraint 
handler. A separator should only generate cuts to improve the LP 
relaxation, but should not be part of the model definition, i.e., should 
not change feasibility of any (integer) feasible solution.

Best,
Gerald


On 14.09.2017 09:23, Gregor Hendel wrote:
> Dear Emanuel,
>
> I am sorry for your troubles. A good first step to debugging 
> unexpected plugin behavior is to
> add the define
>
> #define SCIP_DEBUG
>
> at the very top of the file src/scip/sepa.c. After recompiling SCIP 
> and inputting a problem to optimize,
> you should see a lot of prose explaining why SCIP decides to run (or 
> not to run) your plugin.
>
> If you need further assistance, please add information about concrete 
> parameter values for priority, maxrounds, frequency, and also the 
> DELAYED flag.
>
> Kind regards,
> Gregor
>
>
>
>
> Am 13.09.2017 um 17:29 schrieb Emanuel Delgadillo:
>> Hi,
>>
>> I have implemented a branch-price with SCIP and I want to add a 
>> separator. I coded it with C++ extending ObjSepa class and included 
>> it to scip, but it is never called.
>> I have output messages in it, and I have checked maxrounds(root) and 
>> maxcuts(root) are not 0 and also tried with different values for 
>> priority and frequency. I have my own version of 
>> includeDefaultPlugins, which doesn't include any other default separator.
>> I have already checked documentation, examples and previous similar 
>> threads, but I couldn't solve it. I also checked some files of source 
>> code (solve.c) to find what I'm doing wrong, without success.
>> Any ideas?
>>
>> Thanks in advance,
>> Emanuel
>>
>>
>>
>>
>>
>> _______________________________________________
>> Scip mailing list
>> Scip at zib.de
>> https://listserv.zib.de/mailman/listinfo/scip
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20170914/c08ade6b/attachment.html>