Track of evaluations

[javiervinuales]

Hello,

I am using Platypus for optimizing designs using fluid dynamics (CFD). That means that my evaluation function needs to create a folder and run a CFD simulation, from which the objective/constraints can be extracted. Each evaluation is stored in a numeric folder (i=1,2,3...). I am interested in keeping track of how the optimization population (algorithm.result) relates to the original evaluation (i=1,2,3...), as the simulation contains a lot of useful data I need to use. How can I do that?

Thanks!

[jetuk]

Two options I can think of:

1. Use your output folders for the results. I.e. ignore the algorithm.result data. At the end of the simulation read the objectives and constraints you have saved separately (I've done this before with an external database), then perform your own non-dominated sort of the feasible solutions. You then know exactly which solution is which and have access to all the corresponding meta-data.

2. Subclass Problem and implement your own evaluate method. This method calls your CFD model, but also assign an unique attribute (e.g. .id) to the solution that corresponds to your run folder name/UUID. Hopefully that attribute would then be available on each of the solutions in algorithm.result at the end of the simulation.

[github-actions[bot]]

This issue is stale and will be closed soon. If you feel this issue is still relevant, please comment to keep it active. Please also consider working on a fix and submitting a PR.

[debpal]

See the answer of issue #210