Improve gradient performance #53
Merged
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
Should improve performance by factor of ~2-5 in most cases. For now, there are two working solutions: one vectorized and one where one time axis is looped over, similar to the calculation of the filter function.
Codecov Report
@@ Coverage Diff @@
## master #53 +/- ##
==========================================
- Coverage 97.21% 97.04% -0.17%
==========================================
Files 9 9
Lines 2187 2203 +16
Branches 502 499 -3
==========================================
+ Hits 2126 2138 +12
- Misses 25 29 +4
Partials 36 36
Continue to review full report at Codecov.
|
Looped calculation is faster in almost all cases.
liouville_derivative and control_matrix_at_timestep_derivative do not really make sense on their own.
Used in both Basis and PulseSequence constructors
This was referenced Mar 2, 2021
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Outdated suggestions cannot be applied.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.
This is a refactor meant to improve the performance of the gradient calculation and align the API to the main modules.
Currently, there are two solutions implemented for testing:
gradient.calculate_derivative_of_control_matrix_from_scratch, which then usesgradient.control_matrix_at_timestep_derivative. Python loops are mostly avoided in favor of closedeinsumexpressions.gradient.calculate_derivative_of_control_matrix_from_scratch_looploops over one time dimension in a Python loop (similar to the calculation of the control matrix), usesgradient.control_matrix_at_timestep_derivative_loop.Benchmarks indicate that as soon as the dimension
d > 2, the loop version is faster because theeinsumexpressions become too large: