Refactor _evaluate_gradient for param_shift
#5666
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
As name says. --------- Co-authored-by: Astral Cai <astral.cai@xanadu.ai>
**Context:** Some newly added test cases failing with legacy opmath **Description of the Change:** Skip these tests if new opmath is disabled: `test_measure_identity_no_wires` in `tests/devices/qubit/test_measure.py` `test_identity_on_no_wires_with_other_observables` in `tests/devices/qubit/test_sampling.py` `test_observables_on_no_wires_coeffs` in `tests/pauli/grouping/test_pauli_group_observables.py` `test_multiple_hamiltonian_expansion_finite_shots` in `tests/test_qnode_legacy.py` **Related Shortcut Issues:** [sc-62144] [sc-62143]
**Context:** 1. `Hamiltonian` does not remove ops from queue if `simplify=True` 2. `TrotterProduct` does not remove `Hamiltonian` **Description of the Change:** Manually dequeue said ops **Related Shortcut Issues:** [sc-62142] --------- Co-authored-by: Mudit Pandey <mudit.pandey@xanadu.ai>
### Before submitting
Please complete the following checklist when submitting a PR:
- [x] All new features must include a unit test.
If you've fixed a bug or added code that should be tested, add a test to
the
test directory!
- [x] All new functions and code must be clearly commented and
documented.
If you do make documentation changes, make sure that the docs build and
render correctly by running `make docs`.
- [x] Ensure that the test suite passes, by running `make test`.
- [x] Add a new entry to the `doc/releases/changelog-dev.md` file,
summarizing the
change, and including a link back to the PR.
- [x] The PennyLane source code conforms to
[PEP8 standards](https://www.python.org/dev/peps/pep-0008/).
We check all of our code against [Pylint](https://www.pylint.org/).
To lint modified files, simply `pip install pylint`, and then
run `pylint pennylane/path/to/file.py`.
When all the above are checked, delete everything above the dashed
line and fill in the pull request template.
------------------------------------------------------------------------------------------------------------
**Context:**
`qml.counts` does not return the same `keys` with `dynamic_one_shot` and
`defer_measurements`.
**Description of the Change:**
Cast `keys` to the type produced by `concretize`.
**Benefits:**
Consistent `keys`.
**Possible Drawbacks:**
Slight (likely unnoticeable) performance decrease since `NDArray` object
cannot be preallocated.
**Related GitHub Issues:**
#5566
---------
Co-authored-by: David Wierichs <david.wierichs@xanadu.ai>
- [x] Fix last open bug or open issue **Context:** Docs get out of sync over time. In particular, the output shape of gradient transforms applied to QNodes was changed in #4945 but not updated in the docs. **Description of the Change:** Update examples in docs of `qml.gradients` module. Also updates the `qml.kernels` docs, with a few very small changes. **Benefits:** **Possible Drawbacks:** **Related GitHub Issues:**
2 very minor corrections to documentation added in this release
dwierichs
commented
May 7, 2024
dwierichs
commented
May 7, 2024
dwierichs
commented
May 7, 2024
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #5666 +/- ##
==========================================
- Coverage 99.68% 99.67% -0.01%
==========================================
Files 416 416
Lines 39055 38757 -298
==========================================
- Hits 38931 38632 -299
- Misses 124 125 +1 ☔ View full report in Codecov by Sentry. |
dwierichs
added
the
review-ready 👌
astralcai
approved these changes
May 16, 2024
dwierichs
commented
May 16, 2024
dwierichs
commented
May 16, 2024
albi3ro
approved these changes
May 22, 2024
dwierichs
added
merge-ready ✔️
dwierichs
added a commit
that referenced
this pull request
Jun 14, 2024
…oadcast=True)` (#5667) **Context:** Shot vectors and multiple measurements are currently explicitly not supported when using `param_shift` with the `broadcast=True` option. In #5666 , a crucial subroutine `_evaluat_gradient` is being prepared to allow for support of these scenarios. **Description of the Change:** Allow for shot vectors and/or multiple measurements in `param_shift(..., broadcast=true)`. Due to the changes in #5666 including unit tests for the relevant subroutines, this PR mostly is concerned with integration tests. **Benefits:** Feature parity with `broadcast=False`, allowing faster gradient evaluation. **Possible Drawbacks:** **Related GitHub Issues:** fixes #5598 [sc-62283] --------- Co-authored-by: Mudit Pandey <mudit.pandey@xanadu.ai> Co-authored-by: Astral Cai <astral.cai@xanadu.ai> Co-authored-by: Vincent Michaud-Rioux <vincentm@nanoacademic.com> Co-authored-by: lillian542 <38584660+lillian542@users.noreply.github.com>
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.
Context:
param_shiftuses the internal method_evaluate_gradient, which mostly consists of logic to take the contraction of tape execution results with the parameter-shift rule coefficients and map it over tuple axes. It also needs to respect batching of execution results ifbroadcast=Trueis used inparam_shift.Description of the Change:
This PR cleans up
_evaluate_gradient, extends it to multi-measurement and shot vector scenarios when broadcasting is used, and recycles helper methods fromgradient_transform.pyto reduce the code.We also add unit tests for this method, allowing to reduce integration test count in the future.
Benefits:
Prepare
_evaluate_gradientfor multi-measurement and shot vector support withbroadcast=True.Improve testing and code quality.
Possible Drawbacks:
Related GitHub Issues:
prepares a bug fix for #5598
[sc-62283]