Repository navigation
Conversation
This branch has not been deployed
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fetching a VAMP or TICA model during
partial_fitdiscards its covariance accumulator. The next batch silently starts a new estimate. For two simple batches, VAMP reports covariance 1.25 instead of 26.25 when a model is fetched between them. Accessing.modelor callingtransformalso triggers this.Keep the accumulated covariances and update the decomposed model when those covariances change. Repeated model reads keep the same model, previously fetched snapshots stay unchanged, and explicit fits still reset the stream. Changing lagtime also starts a fresh covariance estimate before adding data, so statistics from different lags are not mixed. Regression tests cover VAMP/TICA, weighted batches, all three access paths, and explicit-fit resets.
Validation on macOS/Python 3.12:
Numerical shard, base interfaces and doctests: 521 passed, with 12 optional Torch collection-skip events.
Independent lifecycle and lag-change checks: 70 passed. The original streaming regressions fail on the base; the lag-change regression prevents mixing old and new lags.
Current-base native extensions built locally; no native source changes. Full suite and other platform/Python matrices were not run.
Include tests for the change
Add yourself to
AUTHORSPrepared with AI assistance.