[SCIP] Inquiry About SCIP-Exact Solution Discrepancy for MIP Model

[Leon Eifler]

Hi Sepideh,


I am always happy when people want to use exact SCIP! I checked your 
instance, and that is indeed a bug present in the 0.3 tarball.

As also written on GitHub, the most up-to-date version of exact SCIP is 
always available through 
https://github.com/scipopt/scip/tree/exact-rational, where that bug has 
already been fixed.

If you use that, your problem should be solved correctly. If you 
experience any further bugs, please don't hesitate to reach out.


Best regards,

Leon



Am 3/11/2024 um 4:17 PM schrieb Sedghi, Sepideh:
>
> Dear SCIP Development Team,
>
>
> I am a Ph.D. student at Worcester Polytechnic Institute (WPI). I 
> recently installed SCIP-Exact 0.3 for my research to solve MIP models 
> without roundoff errors. Although "|ctest -R MIPEX"|passed all the 
> tests and confirmed a successful installation, I encountered 
> unexpected behavior while solving a basic knapsack problem. With the 
> exact mode disabled, SCIP provided the optimal solution of 14. 
> However, with the exact mode enabled, SCIP returned a feasible but 
> non-optimal solution of 0. I tried with a few other examples as well, 
> and in all the examples SCIP returned a non-optimal solution with 
> exact mode enabled. I have attached my model and a screenshot of the 
> SCIP solution for your reference. Could you kindly provide insights on 
> how to resolve the discrepancy?
>
> Thank you for your time and assistance.
>
> Best regards, Sepideh Sedghi
>
>
> _______________________________________________
> 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/20240311/18199501/attachment.html>