[SCIP] Long runtimes of VRP solver
Gerald Gamrath
gamrath at zib.de
Thu Jan 31 09:49:53 CET 2019
Dear Shen,
don't worry, nothing went wrong, this is similar to the behavior I observed.
Note that the VRP example is just a very basic example for how a pricer
can be implemented. It does not focus on providing good performance
since this would complicate the code. If you want to implement a
branch-and-price code, I would always recommend to look at the
binpacking example or the coloring application. These two provide a
better overview of what is normally needed for branch-and-price code,
including problem data and a dedicated branching rule.
Best regards,
Gerald
On 30.01.19 11:51, Shen Li wrote:
> Dear SCIP mailing list
>
> I just compiled the VRP example of the SCIP package, and I obtained
> the following output for eil22.vrp (part of the SCIP/VRP data files):
>
> (…)
> time | node | left |LP iter|LP it/n| mem |mdpt |frac |vars |cons
> |cols |rows |cuts |confs|strbr| dualbound | primalbound | gap
> (…)
> 428s| 300 | 15 |554079 |1689.8 | 256M| 20 | 29 | 37k| 273 |
> 13k| 273 | 2 | 0 | 421 | 3.687105e+02 | 3.750000e+02 | 1.71%
>
> SCIP Status : problem is solved [optimal solution found]
> Solving Time (sec) : 476.24
> Solving Nodes : 359
> Primal Bound : +3.75000000000000e+02 (23 solutions)
> Dual Bound : +3.75000000000000e+02
> Gap : 0.00 %
> SCIP Status : problem is solved [optimal solution found]
> Total Time : 476.24
> solving : 476.24
> presolving : 0.00 (included in solving)
> reading : 0.00
> copying : 0.06 (3 #copies) (minimal 0.02, maximal
> 0.03, average 0.02)
> (…)
>
> I found it surprising that it took 428 seconds to solve this instance,
> which is typically solved within fractions of a second. Is this is
> similar to the behaviour you have observed? If not, any pointers to
> where I may have gone wrong would be greatly appreciated!
>
> Many thanks
>
> Shen
>
> _______________________________________________
> 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/20190131/8a236399/attachment.html>
More information about the Scip
mailing list