Beginning with Gurobi 9.0, log files contain a model fingerprint:
Model fingerprint: 0x6775726f
Two models have the same fingerprint if and only if they are exactly the same. If you solve two models with the same model fingerprint on the same platform using the same Gurobi parameter settings (and if applicable, using the same start solution), then the solution path will be the same. This means that Gurobi will explore the same number of nodes, use the same number of simplex iterations, etc. Note that for two models to be identical, the constraints and variables must be in the same order.
The fingerprint can be used to confirm that you are solving the same model. If two models have different fingerprints, the models are different in some way. This can occur if the code that creates the model is non-deterministic. Common sources of non-determinism in Python code are (i) Python's built-in sets and (ii) certain pandas operations.
Comments
0 comments
Article is closed for comments.