[SCIP] forcecut in SCIPaddCut()
Benjamin Müller
benjamin.mueller at zib.de
Wed Nov 18 16:55:06 CET 2015
Dear Diego,
I tried both options mentioned in your mail but I could not see the
behaviour you described. Did you implement your own separator and tried
those different options? Are you sure if your code calls SCIPaddCut()?
Can you send me an instance where I can reproduce this behaviour?
By the way,
SCIP_CALL( SCIPsetRealParam(scip, "separating/objparalfac", 0.0) );
works totally fine for me. I guess you just have a typo in objparalfac
(instead of objparalfrac).
Best,
Benny
On 11/13/2015 12:19 PM, dponce at us.es wrote:
>
> Hello list,
>
> I'm missing something about parameter forcecut in function SCIPaddCut()
>
> Taking into acount this previous question
>
> http://listserv.zib.de/pipermail/scip/2013-June/001521.html
>
> and Tobias Achterberg's thesis Section 3.3.8, I though that next two
> options will be similar
>
> Option 1
>
> forcecut = false
>
> SCIP_CALL( SCIPsetIntParam(scip, "separating/maxroundsroot", -1) );
> SCIP_CALL( SCIPsetIntParam(scip, "separating/maxrounds", 0) );
> SCIP_CALL( SCIPsetRealParam(scip, "separating/minefficacyroot", 0.0) );
>
> Option 2
>
> forcecut = true
>
> SCIP_CALL( SCIPsetIntParam(scip, "separating/maxroundsroot", -1) );
> SCIP_CALL( SCIPsetIntParam(scip, "separating/maxrounds", 0) );
>
> But really Option 1 does not add cuts. What's going on? I guessed I
> was not requiring nothing to cuts.
>
> With
>
> SCIP_CALL( SCIPsetRealParam(scip, "separating/minorthoroot", 0.0) );
>
> I have the same. And I cannot change parameter separating/objparalfrac.
>
> Best and thanks.
>
> Diego.
>
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> http://listserv.zib.de/mailman/listinfo/scip
--
______________________________
Benjamin Müller
Zuse Institute Berlin
Takustr. 7, 14195 Berlin
benjamin.mueller at zib.de
+49 30 841 85-195
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20151118/9107a791/attachment.html>
More information about the Scip
mailing list