[SCIP] Saving decomposed model with mps

[Marco Lübbecke]

Hi Zachary,

welcome to the GCG world, glad that you like it so far. The code grew too
fast last year, so documentation is lagging behind -- we are working on
documenting the reporting facilities etc. currently, so any input/wishes
from your point as a user is most appreciated.

It is not the idea to re-arrange the MPS file explicitly, instead, you can
write a second file, the DEC file, that contains the re-arrangement
information. If you plan to analyze the decomposition (automatically), look
at that file. It can also be used to start GCG with a particular
decomposition without detection (just read in MPS and a corresponding DEC
file) and this is also the way to force a particular decomposition that GCG
does not/cannot automatically detect.

You mentioned the "explore" menu which is the place we intended for, well,
exploration. "visualize" is your best friend here, showing the
decomposition as a PDF (you may need to set up your PDF viewer for this).
You may also try to "write report" which produces a LaTeX file with some
(maybe/hopefully) interesting information.

Check section "GCG" in the latest release report
http://www.optimization-online.org/DB_HTML/2018/07/6692.html which does not
describe how to do things but hints at what exists that we find is most
useful for starters. Again. much more documentation is under way, until
this is done, feel free to contact us here on the list or directly any time.

We are also interested in your findings, wishes for extensions etc.

Take care, Marco











Am Do., 21. Feb. 2019 um 04:32 Uhr schrieb Zachary Steever <
zjsteeve at buffalo.edu>:

> Hello,
>
> I am a new user of SCIP and GCG. First, I'd just like to thank you for
> your wonderful product. I have a question regarding GCG. I am looking to:
>  (1) read in a model from an mps file
>  (2) use “detect” to identify block structures
>  (3) decompose/rearrange my model according to one of the identified
> structures
>  (4) simply save the rearranged model back into into an mps fiile
> (maintaining its new structure.
>
> Of note, I am not looking to solve the instance, but rather to save the
> decomposed model so that I may otherwise analyze its new structure.
>
> In the interactive GCG shell, I am able to run "detect", and when I use
> "explore" I see that GCG has detected multiple decompositions. When I use
> "write/problem" and then inspect the resulting mps file, however, I see
> that the original structure still exists. Any help would be greatly
> appreciated (Alternatively, if there is a way to do this in Python with
> pyscipopt, that would also be helpful).
>
> Thank you in advance for your help,
> Zachary Steever
>
>
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
>


-- 
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
www.or.rwth-aachen.de/luebbecke
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.zib.de/pipermail/scip/attachments/20190221/07d63c0e/attachment.html>