Re: [Help-glpk] Solution not found on larger problem

[Andrew Makhorin]

On Sun, 2011-09-11 at 13:50 +0200, Chris Bielow wrote:
> Dear glpk team,
> 
> I want to use glpk to solve some ILP instances, and for the "average" 
> problem the following works quite well:
> 
> glpsol.exe  --cuts --max   p2.mps -o p2.solution
> 
> This works for Problems p2 and p3.
> Unfortunately for p1, the solver seems not be able to compute a solution 
> in reasonable time (I aborted it after 24h).
> 
> Using CoinMP, I have the solution in <0.5 seconds with some cuts and 
> heuristics enabled.
> I tried to play with some flags of glpsol.exe, but no luck.
> 
> Do you have any idea of what I can try to get the solution for p1?
> 
> Thx in advance.
> 
> cheers
> Chris
> 
> 

Sorry, p1.mps is hard for glpk because of its size and combinatorial
structure.

I used glpsol 4.47 with options --max --pcost and --clique to solve
p1.mps and it found an integer feasible solution with mipgap = 30% for
about 15 minutes. However, solving to optimality may take too long time.


Andrew Makhorin