Constraint violation modification + direct evolution of parameters #22
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
There was a problem hiding this comment.
Pull Request Overview
This PR updates the evolutionary algorithm to incorporate constraint violation as a pseudo‐objective and adds a new direct prescriptor for evolving parameters directly. Key changes include:
- Modifying evaluator functions to return metrics together with a constraint violation value (cv) across evaluators and tests.
- Updating CSV output and corresponding test assertions to account for the new cv field.
- Introducing the DirectPrescriptor and DirectFactory classes for direct parameter evolution.
Reviewed Changes
Copilot reviewed 18 out of 18 changed files in this pull request and generated no comments.
Show a summary per file
| File | Description |
|---|---|
| tests/test_parallel.py | Updated test to account for the new tuple output in evaluate_candidate. |
| tests/test_cv.py | Added tests for constraint violation prioritization in domination. |
| tests/test_cartpole.py | Modified CSV reading and assertions to validate new output shape and cv. |
| tests/dummy_evaluator.py | Updated evaluator to return a tuple (metrics, cv). |
| presp/prescriptor/direct.py | Added DirectPrescriptor and DirectFactory for direct parameter evolution. |
| presp/prescriptor/base.py | Introduced the cv attribute in the base prescriptor. |
| presp/prescriptor/init.py | Imported DirectPrescriptor and DirectFactory. |
| presp/nsga2_utils.py | Updated domination logic to account for constraint violations. |
| presp/evolution.py | Updated result recording now including cv and refined CSV output. |
| presp/evaluator/base.py | Adjusted evaluation functions to return (metrics, cv). |
| examples/cartpole/direct_evaluator.py | Evaluator now returns tuple (metrics, cv) for CartPole. |
| examples/bnh/demo.ipynb | Updated demo notebook to reflect changes in CSV output and prescriptor usage. |
| examples/bnh/config.yml | Added configuration for evolution and prescriptor parameters. |
| examples/bnh/bnh_evaluator.py | New evaluator implementation that computes constraint violations. |
danyoungday
commented
Apr 25, 2025
| c6 = self.c6(x2) | ||
|
|
||
| g = np.array([c1, c2, c3, c4, c5, c6]) | ||
| g = np.sum(np.maximum(0, g)) |
Contributor
Author
There was a problem hiding this comment.
Constraints are handled within the evaluator, evolution just needs to know the final number
danyoungday
commented
Apr 25, 2025
| for candidate, metrics in zip(pop_subset, pop_results): | ||
| for candidate, (metrics, cv) in zip(pop_subset, pop_results): | ||
| candidate.metrics = metrics | ||
| candidate.cv = cv |
Contributor
Author
There was a problem hiding this comment.
We must set the cv parameter now
danyoungday
commented
Apr 25, 2025
| if candidate1.cv > candidate2.cv: | ||
| return False | ||
|
|
||
| # If both candidates are feasible or have the same constraint violation, we compare metrics |
Contributor
Author
There was a problem hiding this comment.
Domination starts with constraint violation. If they are tied, we go into rank and distance.
danyoungday
commented
Apr 25, 2025
| self.cand_id = "" | ||
| self.parents = None | ||
| self.metrics = None | ||
| self.cv = 0 |
Contributor
Author
There was a problem hiding this comment.
Default CV is 0
danyoungday
commented
Apr 25, 2025
| Creates a randomly initialized vector of floats between xl and xu uniformly. | ||
| """ | ||
| genome = np.random.rand(*self.xl.shape) | ||
| genome = genome * (self.xu - self.xl) + self.xl |
Contributor
Author
There was a problem hiding this comment.
Random uniform initialization between xl and xu
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
We modify our evolution to softly handle constraint violation by using it as a pseudo-objective that is prioritized first.
We also create a direct prescriptor class that directly evolves parameters within a given range.
Finally, we combine the two to optimize a multi-objective problem with constraints.