[SCIP] Presolving time-limits

Ambros Gleixner gleixner at zib.de
Fri Jul 3 20:23:36 CEST 2020


Dear Jan,

> Maybe the solutions are only processed before presolving when using
> the shell. With SCIPaddSol() they are not checked before AFAIK.

No, there should be no difference.  Original solutions added by 
SCIPaddSol() are checked during the transforming stage, so right before 
presolving is triggered, see the first for loop in method 
SCIPtransformProb() in scip_solve.c.

If you don't see this happening, but SCIPaddSol() returned with *stored 
== TRUE, then you might have called it after SCIPtransformProb() (or 
before SCIPfreeTransform() in case you are solving iteratively several 
problems)?

If you set display/verblevel to 5 you should also see a line in the log 
file that informs you on how many solutions were accepted as feasible.

Best,
Ambros


Am 03.07.20 um 17:36 schrieb Jan Berling:
> Dear Marc,
> 
> thank you for your answer!
> 
> I will set a timelimit to scip in in my code and then restart the 
> solution without presolvers.
> 
> Maybe the solutions are only processed before presolving when using
> the shell. With SCIPaddSol() they are not checked before AFAIK.
> 
> Best regards and a nice weekend, Jan
> 
> Am Fr., 3. Juli 2020 um 17:29 Uhr schrieb Marc Pfetsch 
> <pfetsch at mathematik.tu-darmstadt.de 
> <mailto:pfetsch at mathematik.tu-darmstadt.de>>:
> 
> 
> 
> Dear Jan!
> 
>> is it possible to set a time limit on presolving?
> 
> There is not separate time limit for presolving.
> 
> However, you can try the following: in the command line, you can try
> to set a time limit, presolve and then extend the time limit again.
> When you continue the optimization, SCIP will finish presolving,
> however. This can be avoided by "set presolving maxrounds 0". If then
> continue, SCIP will finish presolving, but not perform time consuming
> steps.
> 
>> On a related note, solutions in the storage are processed after 
>> presolving. Is it possible to do this before so that SCIP accepts
>> the feasible solutions and sets the upper bound accordingly?
> 
> I thought that stored solutions would also be processed before 
> presolving. However, additional heuristics may only run after 
> presolving.
> 
> Best
> 
> Marc _______________________________________________ Scip mailing
> list Scip at zib.de <mailto: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
> 

-- 
Ambros Gleixner, Research Group Mathematical Optimization Methods at
Zuse Institute Berlin, http://www.zib.de/gleixner


More information about the Scip mailing list