Optimization Tool - Size of Problem/Speed Issues

[paigens]

I'm working on building an optimizer macro within a batch tool.   I'm running through groups of potential options, and using the optimizer to identify the best candidates within each group.

I think the issue that I'm having is one of constraints. Because of the nature of my data the constraints are pair wise, so if I have options A B and C I can have A and B or A and C but B and C I can't have at the same time. (Except you know, for 300 items, which means thousands of constraints, I have minimal control over how large the group sizes get.)

The constraints are based on a spatial overlap percentage. And so far the only way I can think to look at that is pairwise but I'm open to suggestions.

I left one of my groups running over the weekend and it never finished. I was able to run a smaller set this morning through the same workflow in a matter of seconds.

Is there any way to speed up the process/improve it for larger groups?  I am already using the SLAM matrix as an input format. 

My current plan is to use an iterative macro to slowly narrow the rules around the pairwise constraints to where they actually need to be according to my requirements, and run the batch on the ever increasing constraints. But that seems like a less than ideal solution.  Thoughts?

[paigens]

Just bumping this. I've tried reducing the overall size of the problem set but that doesn't seem to be helping at all. It just runs and never gives me a result even after 2 or 3 days.  If anyone has any insights into detecting what might be making this problem unsolvable I'd appreciate it!

[paigens · Accepted solution]

Posting my solution here in case anyone else needs it.

I switched the solver from glpk to Symphony, and now what was taking hours is taking minutes.

Let that be a lesson to me to flip all the switches first!