Fix unboxing of numpy boolean scalars in C-backend#1790
Merged
ricardoV94 merged 1 commit intopymc-devs:mainfrom Jan 5, 2026
Merged
Fix unboxing of numpy boolean scalars in C-backend#1790ricardoV94 merged 1 commit intopymc-devs:mainfrom
ricardoV94 merged 1 commit intopymc-devs:mainfrom
Conversation
ricardoV94
force-pushed
the
fix_c_scalar_bool_unbox
branch
from
0d583ca to
6b1c93e
Compare
There was a problem hiding this comment.
Pull request overview
This PR fixes a bug in the C-backend where numpy boolean scalars were not being properly unboxed by the ScalarFromTensor operation. The issue caused np.array(True) to evaluate as False after going through scalar_from_tensor, requiring a workaround using .item() in some places like conv.py.
Key changes:
- Replaced deprecated
PyArrayScalar_NewandPyArrayScalar_ASSIGNAPIs withPyArray_Scalarin the C-backend scalar synchronization code - Removed the
.item()workaround from the convolution operation that is no longer needed - Added test coverage for boolean scalar conversion to prevent regression
Reviewed changes
Copilot reviewed 3 out of 3 changed files in this pull request and generated 1 comment.
| File | Description |
|---|---|
pytensor/scalar/basic.py |
Refactored c_sync method to use PyArray_Scalar with proper descriptor handling instead of deprecated PyArrayScalar_* APIs, fixing boolean scalar unboxing. Incremented C code cache version. |
pytensor/tensor/signal/conv.py |
Removed .item() workaround from Convolve2d.perform method that is no longer needed with the C-backend fix. |
tests/tensor/test_basic.py |
Added test_bool_scalar_from_tensor to verify boolean scalars are correctly converted through scalar_from_tensor without losing their truth value. |
ricardoV94
changed the title
Member
Author
|
Failing test is a precision issue with the new conv2D tests that were merged without a seed, and having too strict tolerance. That's fixed by #811 |
jessegrabowski
force-pushed
the
fix_c_scalar_bool_unbox
branch
from
6b1c93e to
e88a658
Compare
ricardoV94
force-pushed
the
fix_c_scalar_bool_unbox
branch
from
e88a658 to
97d0afe
Compare
ricardoV94
commented
Jan 5, 2026
| def perform(self, node, inputs, outputs): | ||
| in1, in2, full_mode = inputs | ||
|
|
||
| if isinstance(full_mode, np.bool): |
Member
Author
There was a problem hiding this comment.
you accidentally force-pushed away this change. Re-added it
jessegrabowski
approved these changes
Jan 5, 2026
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.
We were seeing weird things where an output of ScalarFromTensor for a np.array(True), would evaluate to False. See this old comment here:
pytensor/pytensor/tensor/signal/conv.py
Lines 258 to 260 in d8501d1
And this snippet would fail:
I tracked it down to the
c_syncmethod for ScalarType, which is called when the C code has to generate a Python object as an output (or intermediate computation if going into python mode). Basically unboxing of the internal representation of scalars into numpy scalar arrays.