[Scip] Capturing multiple optimal solutions to a ILP
Rich Grenyer
grenyer at imperial.ac.uk
Tue Apr 28 19:54:34 MEST 2009
Hi -
I've been experimenting with SCIP+SoPlex over the last couple of days
and I would really value some help getting started. My previous
experience has been with the GLPK, so the open complexity of SCIP is a
little daunting!
I have an problem expressed as a ILP. The problem is mid-sized -
approximately 17000 binary variables and 5000 constraints, and has a
large number of optimal solutions. I wish to examine a sample of the
optimal solutions because I'm interested in the redundancy of
individual variables.
As an ideal, I would like to collect a random sample of, say,
1000-10000 solutions from the set of optimal solutions to the ILP. Is
this a feasible (meaningful?) goal, and where might I start looking
for ways to implement it?
Simply presolving and optimising my program generates (extremely
quickly!) 8 optimal solutions, and describes one of them. Should I
start looking at ways to 'design' a search strategy to sample a wider
range of decisions? Or is there a simple solution I've missed?
Many thanks in advance...
Rich
More information about the Scip
mailing list