idaholab / raven

RAVEN is a flexible and multi-purpose probabilistic risk analysis, validation and uncertainty quantification, parameter optimization, model reduction and data knowledge-discovering framework.
https://raven.inl.gov/
Apache License 2.0
218 stars 133 forks source link

[TASK] Simulated Annealing Discrete Optimization not supported #2305

Closed alfoa closed 4 months ago

alfoa commented 5 months ago

Issue Description

Is your feature request related to a problem? Please describe. Simulated Annealing does not support discrete optimization.

Describe the solution you'd like Possibility to support discrete optimization. This support would require the modification of the method _nextNeighbour. Currently, the method assumes continuous variables only. An approach to support both could resolve around the usage of the distributions linked to the variables for the selection of the "next neighbor" in the "probability space" and then use the "inverseCDF" method to back transform in the manifest space.

Describe alternatives you've considered Use a continuous optimization and then cast the "discrete" variables as "discrete" in an external model. This approach somehow works even if it is not optimal


For Change Control Board: Issue Review

This review should occur before any development is performed as a response to this issue.


For Change Control Board: Issue Closure

This review should occur when the issue is imminently going to be closed.

wangcj05 commented 4 months ago

Issue closure checklist is good.