How to improve LP formulation using presolve
I have written an Integer Program using the Gurobi Python API and noticed the following row in the log:
Presolve removed 9331080 rows and 2671 columns - which took about 35 seconds.
I know for sure that I won't be able to change my formulation resulting in 0 presolve time, but I want to reduce it, and I feel that in my case it is possible.
From what I see a *lot* of rows are deleted (almost 9.5M), is there a way to check what rows/columns are removed? I suspect that this will help me identify redundant inequalities.
How many rows and columns does the original model have? Can you post the corresponding log line?
Please sign in to leave a comment.