[SCIP] SCIP speed-up

Ambros Gleixner gleixner at zib.de
Thu May 17 10:52:59 CEST 2018


Dear Marcus,


Am 15.05.2018 um 15:22 schrieb Marcus Garvie:
> Dear Sir or Madam.
> 
> I recently downloaded a free version if SCIP to calculate all feasible 
> solutions of a large binary linear system of equations with a dummy 
> objective function. I am following the instructions described at part 4. of
> 
> http://www.aiexp.info/calculating-all-feasible-solutions-of-ilp.html
> 
> Can you please provide me with some advice as to how I can speed-up the 
> solution process?

I guess you saw the other conversation on counting.  Maybe try the same 
thing, activating separation after switching to counting mode.  In the 
shell, this is set sepa emph ...

Otherwise, counting is rather tedious and may naturally be slow.  It is 
difficult, maybe not possible, to speed it up significantly.  (Of 
course, there may be more clever mathematical algorithms to do so. 
Here, I am only talking about the branch-and-bound enumeration that 
SCIP's counting is based on.)


> I notice there is a lot of output displayed during the 
> solution process. Is there a ‘quiet’ mode I can use that will speed 
> things up? What else can I do to speed things up?

I don't think that output creates a performance issue.  But you can play 
with "set display verblevel" and "set display freq".

Best,
Ambros


> I have attached a typical LP file that I solve.
> 
> Thank you,
> 
> Marcus.
> 
> PS The attached model takes me about 5 minutes to solve, but I’m wanting 
> to solve much larger problems too.
> 
> 
> 
> ________________________________
> Marcus R Garvie
> Associate Professor
> Rm 552 MacNaughton Bldg
> Dept. of Math & Stats
> University of Guelph
> Guelph, ON Canada N1G 2W1
> Tel. 519-824-4120 ext 53409
> Email. _mgarvie at uoguelph.ca_ <mailto:mgarvie at uoguelph.ca>
> 
> 
> 
> 
> 
> _______________________________________________
> Scip mailing list
> Scip at zib.de
> https://listserv.zib.de/mailman/listinfo/scip
> 

-- 
Ambros Gleixner, Research Group Mathematical Optimization Methods at 
Zuse Institute Berlin, http://www.zib.de/gleixner


More information about the Scip mailing list