No start values specified in MIP start
Awaiting user input1- I am wondering as to why Gurobi reject the provided mip-start values in my model. The model is two-stage and the second stage only have a slightly different objective.
While the first stage generates a couple of solutions and no constraint is violated by the given start values, the MIP starts are rejected right away with the message:
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
No start values specified in MIP start
User MIP start did not produce a new incumbent solution
2- In my investigation, I set the upper and lower bounds of the variables in the .sol file to the MIP-start values and reran the second stage. Surprisingly, it then found an optimal solution with the MIP-start values.
Presolve removed 53086 rows and 7223 columns
Presolve time: 0.03s
Presolve: All rows and columns removed
Explored 0 nodes (0 simplex iterations) in 0.09 seconds (0.09 work units)
Thread count was 1 (of 16 available processors)
Solution count 1: 5.78938e+06
Optimal solution found (tolerance 0.00e+00)
Warning: max constraint violation (2.7081e-01) exceeds tolerance
Warning: max general constraint violation (2.7081e-01) exceeds tolerance
(possibly due to large matrix coefficient range)
3- I also tried setting SubMIPNodes to some high values like 200000 but it didn't help.
Any insights into why this discrepancy occurs would be greatly appreciated.
-
The message
No start values specified in MIP start
implies that something is not working correctly with your MIP start. I assume you provide the start values via a file, is this correct? Do you see any log output around the line
\(Read MIP start from file\)
that tells what might be wrong?In your second approach, the output shows that the solution violates a general constraint and is actually not feasible:
Warning: max general constraint violation (2.7081e-01) exceeds tolerance (possibly due to large matrix coefficient range)
It also points to possible numerical issues (large matrix coefficient range).
Our Guidelines for Numerical Issues might be helpful.0
Please sign in to leave a comment.
Comments
1 comment