Deprecate MeasurementProcess.return_type and ObservableReturnTypes
#6841
Conversation
|
Hello. You may have forgotten to update the changelog!
|
JerryChen97
added
the
deprecation 🗑️
This reverts commit 9bcec8f.
…-and-`ObservableReturnTypes`
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #6841 +/- ##
=======================================
Coverage 99.60% 99.60%
=======================================
Files 476 476
Lines 45194 45185 -9
=======================================
- Hits 45015 45008 -7
+ Misses 179 177 -2 ☔ View full report in Codecov by Sentry. |
…-and-`ObservableReturnTypes`
| @@ -56,27 +56,27 @@ def test_serialize_numeric_arguments(self, queue, observable_queue, expected_str | |||
| observable3 = qml.prod(qml.PauliZ(0), qml.PauliZ(1)) | |||
|
|
|||
| numeric_observable_queue = [ | |||
| (returntype1, observable1, "|||ObservableReturnTypes.Expectation!PauliZ[0]"), | |||
| (returntype1, observable1, "|||ExpectationMP!PauliZ[0]"), | |||
There was a problem hiding this comment.
QQ: after deprecating return type and replacing occurrences with isinstance, we might have to replace the expected output texts in circuit_graph with the exact class names. Is this a harmless change here? @albi3ro @astralcai
Also, in this module do we have any secondary applications from these return type information?
There was a problem hiding this comment.
Another question: is doc/development/guide/architecture.rst also within the scope of deprecations/removals? It defines MP.return_type there as well
There was a problem hiding this comment.
QQ: after deprecating return type and replacing occurrences with isinstance, we might have to replace the expected output texts in
circuit_graphwith the exact class names. Is this a harmless change here? @albi3ro @astralcai Also, in this module do we have any secondary applications from these return type information?
I see another usecase in drawer: tape_text.py:_add_measurement and _add_cwire_measurement
There was a problem hiding this comment.
Also, return_type has been built into the repr of MP...
There was a problem hiding this comment.
Also,
return_typehas been built into thereprof MP...
solution: make repr equivalents individually for each subclass of MP.
reason: we definitely want to get rid of the Enum of obs short name list, but on the other hand we should not touch any public interfaces that use these pieces of information.
…None, this should be a QuantumFunctionError, no?
Context:
The
MeasurementProcess.return_typeis only used by the legacy device in a chain ofif mp.return_type.name is ...checks that can easily be replaced withisinstancechecks.In PL0.40,
return_typeis a property ofMeasurementProcesswhich returns None only. Meanwhile,ObservableReturnTypesis an Enum that defines and wraps the name strings, redundantly.Description of the Change:
return_type.return_typewith directisinstancecheckObservableReturnTypeswith direct definitions.return_type: ExpectationMP, ProbabilityMP, SampleMP, StateMP, CountsMP and VarianceMPBenefits:
Less redundancies.
Possible Drawbacks:
any downstreaming usage of the deprecations:
_adjoint_jacobian_base.py,lightning_base.pyandtest_measurements.py. Occurrencies here actually block this PRRelated GitHub Issues:
[sc-71892]