Refactoring gaugeopt.py: maintainability and fidelity as an objective #669
Conversation
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
…two codepaths in _create_objective_fn into their own functions, called _legacy_create_scalar_objective and _legacy_create_least_squares_objective
…s to values based on module-wide or class-wide constants
|
Note: the fact that eigenvalue_entanglement_fidelity in reportables.py rather than optools.py is a weirdness related to issue #526. |
… motivating the current definition.
rileyjmurray
commented
Oct 23, 2025
Comment on lines
+1631
to
+1642
| if is_unitary and is_tp: | ||
| evA = _np.linalg.eigvals(a) | ||
| evB = _np.linalg.eigvals(b) | ||
| _, pairs = _tools.minweight_match(evA, evB, lambda x, y: abs(x - y), | ||
| return_pairs=True) # just to get pairing | ||
| fid = abs(_np.sum([_np.conjugate(evB[j]) * evA[i] for i, j in pairs])) / d2 | ||
| else: | ||
| Ja = _tools.fast_jamiolkowski_iso_std(a, mx_basis) | ||
| Jb = _tools.fast_jamiolkowski_iso_std(b, mx_basis) | ||
| from pygsti.tools.optools import eigenvalue_fidelity | ||
| fid = eigenvalue_fidelity(Ja, Jb, gauge_invariant=True) | ||
| return 1.0 - fid |
There was a problem hiding this comment.
Would be nice to figure out how to do eigenvalue matching in the is_unitary and is_tp codepath to be consistent with the matching in the second codepath.
…lt values to values based on module-wide or class-wide constants" This reverts commit b2b6746. I moved the contents of that commit to another branch.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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 PR is split off of #664, which I've since closed for being too large in scope.
This PR has major backward-compatible changes to the
gaugeopt_to_targetand_create_objective_fnfunctions in gaugeopt.py.Both of these changes are stepping stones so we can introduce new types of gauge optimization in the future. My vision is that one day we'll be able to include user-defined black-box objective functions in gauge optimization suites. This will enable future prototyping without reaching into gauge optimization plumbing.
This PR also changes the definition of fidelity as a gauge optimization objective. It's now defined in such a way that if two models exhibit no relational errors with respect to one another, then it's possible to drive the fidelity-based loss function to zero even if the models are not gauge-equivalent. As part of this effort I introduced a new function in optools.py called eigenvalue_fidelity. This function computes fidelity of two states in a way that pretends they share an eigenbasis. There's an option of ordering eigenvalues in a gauge-invariant way or in a way that's technically gauge-dependent yet still robust.
Small changes to model.py and explicitopmodel.py: type annotations for copy methods.