[SCIP] Pricer loop doesn't stop iterating
witzig at zib.de
Fri Dec 11 10:02:17 CET 2015
I cannot remember any changes in the according methods between 3.1.1 and
3.2.0 that could cause such a behaviour. Have you used SCIP 3.1.1 before
or do you mean with "previous version" an earlier SCIP release?
Are you sure that you are in SCIP_DECL_PRICERREDCOST and not in
SCIP_DECL_PRICERFARKAS? Because after your pricer runs and returns with
the result = SCIPDIDNOTRUN the SCIPpriceLoop should terminate because of:
*aborted = ( (*aborted) || (result == SCIP_DIDNOTRUN) );
Have you tried to set the bool pointer stopearly to TRUE? Afterwards,
SCIP should stop pricing if this flag is set by the first called pricer.
The parameter "maxpricingrounds" is only available in
branch_strongcoloring.c in the coloring application
(your-scip/applications/Coloring). SCIP calls SCIPpriceLoop with
maxpricerounds = -1 which means no limit.
Am 11.12.2015 um 08:37 schrieb Rossinelli Philippe Sakari:
> Dear all,
> I am currently working on the same optimization problem as Jan Berling
> who posted on this mailing list some while ago. My issue is the
> following. We recently decided to update SCIP, MUMPS, METIS and Ipopt to
> more recent versions (SCIP v3.2.0, MUMPS 4.10.0, METIS v4.0.3, Ipopt
> 3.12.4). We managed to install all the components, and the whole program
> seems to work very well when no pricer is added.
> The catch comes when we try to use our implemented pricer, it doesn’t
> stop iterating (while when using the previous version, it would stop
> after 3 iterations). After 3 iterations, no more variables are added to
> the problem though.
> I tried to solve the problem by implementing in the
> SCIP_DECL_PRICERREDCOST the following lines
> if(pricerData->countIteration > 5)
> (*result) = SCIP_DIDNOTRUN;
> But it doesn’t make any difference, it would not stop.
> I also saw in the solver that there is parameter maxpricerounds that
> could be set, but it doesn’t appear in the list of “all scip parameters”
> in here : http://scip.zib.de/doc/html/PARAMETERS.php .
> I would appreciate any idea!
> Philippe Rossinelli
> Scip mailing list
> Scip at zib.de
Informationstechnik Berlin (ZIB)
Division Mathematical Optimization and Scientific Information
Research Group Mathematical Optimization Methods
Tel. : +49 (0)30 84185-416
Fax : +49 (0)30 84185-269
email: witzig at zib.de
More information about the Scip