Qiskit / qiskit-ibm-runtime / 38712388204

91%

main: 91%

push

github

Fix FoldRzz pass to respect target (#3489)

<!--
⚠️ The pull request title should be short detailed and understandable
for all.
⚠️ Also, please add a release note file using towncrier if the change
needs to
  be documented in the release notes.
⚠️ If your pull request fixes an open issue, please link to the issue.

✅ I have added the tests to cover my changes.
✅ I have updated the documentation accordingly.
✅ I have read the CONTRIBUTING document.
-->

### Summary

The `FoldRZZ` pass and by extension the `rzz_wrapper()` function were
previously incorrectly appending a global phase gate to the circuits
they output. No IBM backend (both current and historical) supports a
`GlobalPhaseGate` in the ISA and as a target aware pass used for gate
translation it was invalid to inject the gate not supported by the
target. Running this pass in any compilation targeting an IBM backend
results in an invalid compilation that could not be run on IBM hardware.
Adding a `GlobalPhase` gate was presumably done to track the global
phase adjustment needed by the angle wrapping, however there is a
simpler way this could have been accomplished by just adjust the
`global_phase` attribute of the output circuit. This commit makes this
adjustment to all the places in the `FoldRZZ` path that were previously
using `GlobalPhaseGate` and changes it to adjust the global phase
directly, avoiding inserting invalid gates into the circuit (and also
speeding up the pass).

At the same time this commit updates the `rzz_wrapper()` function, which
is registered in the default `WrapAngleRegistry` to avoid needlessly
instantiating the `FoldRZZ` pass class. The function is solely calling
static methods defined on the class that don't require an instance of
the pass. We can just call the method on the class and avoid the
overhead of create and destroying an object on every call. This is
technically an unrelated change and can be split out into a separate PR
if desired, but since it was a small c... (continued)

6 of 10 new or added lines in 2 files covered. (60.0%)

8649 of 9495 relevant lines covered (91.09%)

9.95 hits per line