[SCIP] Diving Heuristic Implementation

[Marc Pfetsch]

Dear Lars,

you can try to check whether you are in diving in your pricers by using 
SCIPinDive(). Then you would only run the heuristic methods in your 
case. Note that this would effect all diving heuristics, so there is a 
side-effect which you should check. You could, however, store somewhere 
whether you are running your diving heuristic and only do something 
differently in this case.

Best

Marc




On 16/07/2024 14:32, Lars Jaeger wrote:
> Dear SCIP-People,
> 
> I have a branch-and-price algorithm, with 3 different Pricers, 2 
> Heuristics and one exact solving approach.
> 
> I have my own semi assignment branching-rule and everything is working fine.
> 
> Now I want to implement a diving heuristics, fixing one Variable to one 
> (different to my branching rule) and performing Columngeneration but 
> only with the heuristics, till no new column can be found.
> 
> And then fixate a second Variable to one …
> 
> Repeat that till it’s infeasible, or integer feasible.
> 
> I know that I didn’t get a valid lower Bound during the heuristic, but I 
> have reason to expect a good performance in finding a primal solution.
> 
> Is there any implementation, or tips, how to implement such a heuristic?
> 
> I`d assume, this is quite a basic approach, so there should be 
> something, but I didn’t find it so far.
> 
> Please ask, if I forgot to mention some important details.
> 
> Thank you and best wishes,
> 
> Lars Jäger
> 
> ___________________________________________________
> 
> *Lars Jäger, M.Sc.*
> 
> Wissenschaftlicher Mitarbeiter
> 
> Leibniz Universität Hannover
> 
> Institut für Produktionswirtschaft
> 
> Königsworther Platz 1
> 
> D-30167 Hannover
> 
> Büro: I-125
> 
> E-Mail: lars.jaeger at prod.uni-hannover.de 
> <mailto:lars.jaeger at prod.uni-hannover.de>
> 
> Tel: +49 (0)511 762- 14912
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip