"MIP start did not produce a new incumbent solution"

[Rodrigo Castro]

Hi guys, I have modeled an IP using a LP file and I know an initial solution:

* If I put the solution I know in the "subject to" section, it works.
* If I put the solution I know in a .mst, Gurobi tells me it is not feasible.

Without fixing (I always "fix" using the "subject to" section), some of my variables only remain in the "lazy constraints" section. It seems that if I leave those out of the .mst, it works.
Any clues?

[Rodrigo Castro]

Hi again, I think I found the problem (I am the culprit, but it is not the kind of error one would expect). I was generating the .mst files with a wrong syntax:

VARIABLE = VALUE

while I should have used

VARIABLE VALUE

However, some examples still ran (without error) with the first syntax; in any case, I would have expected Gurobi to tell me the syntax was wrong (in all cases): "error while loading .mst file" message, probably.
Cheers!