Cleans up code and refactors tests for single qubit unitary decompositions

[astralcai]

Context:

As I am implementing the XZX single qubit decomposition (#4862), I noticed some issues with the the code and the tests for single qubit unitary decompositions.

Description of the Change:

Cleans up transforms/decompositions/single_qubit_unitary.py.

• Removes duplicate code,
• Fixes doc examples that were untrue
• Adds support for returning global phase for the rot decomposition.
• Normalize all rotation angles to the range $[0, 4\pi]$

Refactors tests/transforms/test_decompositions.py.

• Generalizes logic for obtaining decompositions and asserting their correctness such that all types of decompositions share the same assertion logic. Removes duplicate code.
• Adds testing for when return_global_phase=False as well as when return_global_phase=True
• Update test cases to use analytical instead of numerical values when possible.

Benefits:

Cleaner code.

Related GitHub Issues:

#4868
#4862

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (1e6c11d) 99.65% compared to head (83bea0f) 99.64%.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4869      +/-   ##
==========================================
- Coverage   99.65%   99.64%   -0.02%     
==========================================
  Files         383      383              
  Lines       34571    34314     -257     
==========================================
- Hits        34451    34191     -260     
- Misses        120      123       +3     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[rmoyard]

Hi @astralcai thanks a lot for this follow up PR. I am not sure it is possible to change the base of this PR because you are external, but that would be nice. Currently it is hard to distinguish your two PRs. If it is not possible, I will wait for the first one to be merged before reviewing.

[astralcai]

Hi @astralcai thanks a lot for this follow up PR. I am not sure it is possible to change the base of this PR because you are external, but that would be nice. Currently it is hard to distinguish your two PRs. If it is not possible, I will wait for the first one to be merged before reviewing.

@rmoyard Hi, I have made another update to this PR, now normalizing all rotation angles to $[-2\pi, 2\pi]$ for consistency and easy of use (previously different types decompositions deal with this differently). I have also updated all test cases for single-qubit unitary decompositions to use analytical values as expected parameters when possible.

[rmoyard]

Hi @astralcai ! Once you solve the conflicts I will review this PR. One thing about angles for rotation, our canonical range is [0, 4pi], same as Numpy. So it would be best to remove your angle change from this PR. Thanks 👍

[astralcai]

Hi @astralcai ! Once you solve the conflicts I will review this PR. One thing about angles for rotation, our canonical range is [0, 4pi], same as Numpy. So it would be best to remove your angle change from this PR. Thanks 👍

@rmoyard In this case, I will update the code for the rot decomposition to normalize its rotation angles to the [0, 4pi] range instead.

Can I ask why this range is chosen?

[astralcai]

Hi @astralcai ! Once you solve the conflicts I will review this PR. One thing about angles for rotation, our canonical range is [0, 4pi], same as Numpy. So it would be best to remove your angle change from this PR. Thanks 👍

@rmoyard In this case, I will update the code for the rot decomposition to normalize its rotation angles to the [0, 4pi] range instead.

Can I ask why this range is chosen?

@rmoyard I have updated the PR to remove the angle change. It is ready for review now.

[rmoyard]

Thank you I will take a look between today and tomorrow. From what I remember, we wanted to follow the convention by numpy.mod and % where the rest is not negative. We thought it was a better software friendly convention compared to the other one which comes more from physics.

[rmoyard]

Thanks for cleaning up the tests and the changes to rot. It looks great, I have left some comments. Happy to approve after that 💯

[lillian542]

I don't have anything to add to @rmoyard's feedback. Thanks @astralcai, this looks great! 🚀

[rmoyard]

Thanks @astralcai, feel free to merge the PR 💯

[astralcai]

Thanks @astralcai, feel free to merge the PR 💯

Hi, thank you. I don't think I have permission to merge this PR.