[SCIP] Errors in using GCG version 3.5 for solving some miplib instances.

[Yunzhuang Shen]

Hi Prof. Marco Lübbecke,

Thanks for the timely response.  We are doing research on using machine learning to speed up the process of solving pricing subproblems. An attempt by us is done on the graph coloring problem [1], and we are planning to develop a more generic solution for "unstructured" MIP inputs.

GCG seems to have trouble solving "unstructured" MIP inputs. It raises different types of assertions errors for different instances in MIPLIB with the decomposition tags. It seems that GCG has not been tested for "unstructured"  inputs. I would love to fix these errors and make GCG work on these "unstructured" inputs. However, after some attempts, I feel that debugging GCG code by myself is challenging, as many implementation details are not documented yet.

Lastly, we would love to hear from you any suggestions to proceed in this direction.

Best regards,
Yunzhuang

[1] Shen, Yunzhuang, et al. "Enhancing Column Generation by a Machine-Learning-Based Pricing Heuristic for Graph Coloring." AAAI (2022).
________________________________
From: Marco Lübbecke <marco.luebbecke at rwth-aachen.de>
Sent: Saturday, April 16, 2022 11:38 PM
To: Yunzhuang Shen <s3640365 at student.rmit.edu.au>
Cc: scip at zib.de <scip at zib.de>
Subject: Re: [SCIP] Errors in using GCG version 3.5 for solving some miplib instances.

Hi Yunzhuang,

thanks, we file this as a bug report to look into as soon as we have
capacities; you can write to me directly for any updates on this
issue.

And thanks also for trying GCG on "unstructured" input, if you would
like to share with us your application cases, I am all ears as well.

All the best, Marco

Am Sa., 16. Apr. 2022 um 07:58 Uhr schrieb Yunzhuang Shen
<s3640365 at student.rmit.edu.au>:
>
> Hi SCIP folks,
>
> I am using GCG 3.5 (with SCIP 8.0.0) to solve MIPLIB instances with the decomposition tag.
>
> I find that for some of these instances such as "academictimetablesmall", GCG terminates with the segmentation fault. So I ran the program in debug mode and the attached file is the output. I also tried GCG 3.0.5 (with SCIP 7.0.3) to solve the same instance "academictimetablesmall" and GCG seems to be fine.
>
> Regards,
> Yunzhuang
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://aus01.safelinks.protection.outlook.com/?url=https%3A%2F%2Flistserv.zib.de%2Fmailman%2Flistinfo%2Fscip&data=04%7C01%7Cs3640365%40student.rmit.edu.au%7C2c78f9092e6d4cbda5d108da1fae7b34%7Cd1323671cdbe4417b4d4bdb24b51316b%7C0%7C0%7C637857132336323261%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=Axiq1Z4WM3jUmY4Ms3uNurLuYxDUPTSoFt9Y9rKqA%2BI%3D&reserved=0



--
Prof. Dr. Marco Lübbecke
RWTH Aachen University
Chair of Operations Research
Kackertstrasse 7
D-52072 Aachen
Germany

fon / fax: +49 241 80-93362 / 92369
marco.luebbecke at rwth-aachen.de
https://aus01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwww.or.rwth-aachen.de%2Fluebbecke&data=04%7C01%7Cs3640365%40student.rmit.edu.au%7C2c78f9092e6d4cbda5d108da1fae7b34%7Cd1323671cdbe4417b4d4bdb24b51316b%7C0%7C0%7C637857132336323261%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C2000&sdata=iz%2F2NPRzJyzKB5Yxs68EyFTnhSNgke2UpRdbHYxTddY%3D&reserved=0
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20220418/46899c36/attachment.html>