Issues with Model Optimization Time in Gurobi
Awaiting user input
Hello everyone,
I'm currently working on a linear programming model in Gurobi but I am having trouble with slow optimization times, even for 'basic' problems. I have tested out various solvers and tried to tune the parameters but Gurobi seems to be performing slower than expected. I have a couple of hundred variables and constraints, but I am taking a long time to converge.
I looked at a basic level of making sure the constraints are correct and not redundant. I also have contemplated parameter settings like MIPGap and TimeLimit, but I think there could be a bigger issue at play. I've heard scaling and numerical performance can affect performance but have no idea how to deal with this issue.
Has anyone had similar problems with slow optimization times? Any suggestions or good practice points for speeding performance up or diagnosing model potential bottlenecks would be great.
Thanks for you help!
Best,
Megancharlotte
-
-
Gurobi Staff
Hi Megan,
I'm currently working on a linear programming model in Gurobi
Do you mean mixed integer linear programming model? MIPGap would not make sense in the context of a LP.
It is possible to have small MILP models that are hard to solve. In the MIPLIB 2017 benchmark set we find a model called markshare2 in the "hard" set. This is a tiny MILP model with 74 variables and 7 constraints and it takes Gurobi (and other solvers) are very long time to terminate with default MIP gap settings.
The first place to start with your model would be to inspect the log. Can you share that please?
- Riley
0 -
Please sign in to leave a comment.
Comments
1 comment