I have an MILP (mixed int linear prog).
There is the comiso library
which does something super naive:
- Solve the problem treating all vars as continuous.
- Pick an int var.
- Round it to the closest int, and keep its value fixed for the next iterations.
- Iterate until all int vars are fixed.
Somehow, this naive method beats gurobi by something like seconds vs minutes. Yes, it doesn't have any guarantees, and sometimes it fails (it doesn't go over all the tree like gurobi), but for my problem it's enough.
I was wondering if gurobi has a similar option that could perform similarly.
Please sign in to leave a comment.