[SCIP] Strange behavior of PyscipOpt

[Abbas Omidi]

Dear support team,

I am trying to solve a problem by testing some of SCIP's internal options
in order to compare the solving process. The strange thing is that when I
set the following options the solving process seems to get stuck and
restart again and the B&B tree leads to a sub-optimal solution.

> m.setPresolve(SCIP_PARAMSETTING.AGGRESSIVE)
> m.setHeuristics(SCIP_PARAMSETTING.AGGRESSIVE)
> m.setSeparating(SCIP_PARAMSETTING.AGGRESSIVE)


Actually, the correct solution would be 17 while the solver is showing 18.
[image: image.png]

I think the first line in the above snippet of the log shows something
unusual. I should note that the model without using any options can turn
out the optimal solution. Also, I test the same problem with the console
runner and the run time is so decreasing by using these options as I
expected it to be with the correct optimal solution. Please, see an MRE here
<https://colab.research.google.com/drive/1Z4TrjnpYulPO8Up4hYQrz9A4kYSbdfGY?usp=sharing>.
Also, the related thread could already be found here
<https://github.com/scipopt/PySCIPOpt/issues/680>.


I was wondering if, how can I fix that?

Best regards
Abbas
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20230522/23472907/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image.png
Type: image/png
Size: 91956 bytes
Desc: not available
URL: <http://listserv.zib.de/pipermail/scip/attachments/20230522/23472907/attachment.png>