perf(csa): reuse compiled structured geometry in banking - #104
Merged
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
This avoids repeatedly validating and encoding stable bank entries. The derived geometry view is not shared between studies, persisted in checkpoints, or exposed as mutable optimizer state.
Performance
Same-process benchmarks used 600 CSA steps, a bank capacity of 24, and seven repetitions. Compiled and fallback runs produced identical final states.
Instrumentation on a 300-step integer-array run recorded 5,880 batched distance calculations instead of 9,016 public metric calls. Validation calls remained unchanged at 604, while traced peak allocation decreased from 232,101 to 169,333 bytes. The instrumented runs were used for call and allocation attribution, not for the wall-time claim above.
Correctness
The test coverage exercises append, replacement, removal, growth, clustering, equality-hostile candidates, stale and foreign views, invalid indices, cutoff boundaries, pickle round trips, and checkpoint continuation. A 48-step compiled-versus-fallback trajectory test compares every resulting optimizer state.
No public symbols, dependency requirements, or checkpoint schema were added or changed.
Verification
uv run pytest -q(1598 passed)231 passed)uv run ruff format --check src tests .github/scriptsuv run ruff check .uv run basedpyright src tests .github/scripts(0 errors, 0 warnings, 0 notes)uv run mkdocs build --strict