Small fixes: 'dp' and memory_limit + tensordot axes order #154
Conversation
This file contains 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
removes numpy import and fixes bug with some conversion backens
jcmgray
changed the title
dgasmith
approved these changes
Nov 3, 2020
opt_einsum/contract.py
Outdated
| @@ -569,8 +567,11 @@ def _core_contract(operands, contraction_list, backend='auto', evaluate_constant | |||
| left_pos.append(input_left.find(s)) | |||
| right_pos.append(input_right.find(s)) | |||
|
|
|||
| # Construct the axes tuples in a canonical order | |||
| axes = tuple(zip(*sorted(zip(left_pos, right_pos)))) | |||
There was a problem hiding this comment.
If one of the lists is empty the return is (, ) rather than ((,), (,)) causing the Travis failures.
| @@ -757,7 +758,7 @@ def __call__(self, *arrays, **kwargs): | |||
| try: | |||
| # Check if the backend requires special preparation / calling | |||
| # but also ignore non-numpy arrays -> assume user wants same type back | |||
| if backends.has_backend(backend) and all(isinstance(x, np.ndarray) for x in arrays): | |||
| if backends.has_backend(backend) and all(infer_backend(x) == 'numpy' for x in arrays): | |||
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Outdated suggestions cannot be applied.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.
Description
This PR:
memory_limitmeant the'dp'optimizer would search forever (infinite loop in dynamic-programming with 'max_input' memory limit #153)tensordotaxes in a canonical order (the order they appear on the first operand), so that performance shouldn't change unexpectedly (Performance difference between essentially the same einsum with dummy index interchanging #143)isinstancecall toinfer_backend(is cleaner? and fixes a rare bug when mixing inputs for e.g. jax compiled contraction)Status