Check normalization and raise error when not 1#87
Merged
Conversation
Contributor
☂️ Python Coverage
Overall Coverage
New FilesNo new covered files... Modified Files
|
Contributor
|
Contributor
There was a problem hiding this comment.
Pull request overview
This PR adds a runtime normalization check during autoregressive sampling to detect numerical underflow/overflow cases that can produce non-normalized marginals, and raises an error instead of silently returning inconsistent samples.
Changes:
- Extend component sampling to track a maximum normalization deviation during autoregressive evaluation.
- Raise an error from
sample_programwhen deviation exceeds a threshold, and add an integration test covering the failure path. - Minor comment/TODO wording update in exact scalar summation.
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.
| File | Description |
|---|---|
src/tsim/sampler.py |
Adds per-component normalization deviation tracking and raises an error when deviation is too large. |
test/integration/test_sampler.py |
Adds a regression test for the raised error; also introduces a sys.path mutation and updates a __main__ debug snippet. |
src/tsim/core/exact_scalar.py |
Adjusts TODO comment wording related to overflow prevention in summation. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com>
…uting/tsim into rafaelha/underflow_bug
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Due to underflows or overflows in the evaluation routine, it is possible that marginal probabilities are not normalized.
Previously, normalization was not checked, and inconsistent sampling results could silently be returned. With this PR, the batch size is now increased to one by one, and that extra sample is used to verify normalization.