CODEWITHSUNDEEP
optimizationgurobimixed-integer-programming
1Pre
1Pre

Reputation: 37

Gurobi Reducing Binary Variable Makes Solve Time Longer

I'm solving MILPS with one binary variable. I have written an algorithm to reduce the binary variables. It works and gives the same solution. The unreduced binary variables is 33 variables. The reduced one is 19 variables. I found that running with 33 binary variables takes about 11 seconds while the reduced one takes about 12.5-13 seconds.

I'm wondering how this is possible since the expectation was that reducing the binary variable would makes things much faster as less options have to be explored.

What would be the reason for this to be the other way around?

Upvotes: 0

Views: 294

Answers (1)

mattmilten
mattmilten

Reputation: 6716

It could be that Gurobi automatically finds a similar reduction for the original model, and the different runtimes are just due to performance variability. You should check with multiple random seeds to get a feeling for the performance variability.

It is hard to say more with such limited information on the model.

Upvotes: 1

Related Questions