Implement custom comparison for ControlledSequence

[EmilianoG-byte]

Context:

Currently, qml.equal does not have a registered function that compares two ControlledSequence objects correctly. This PR addresses this issue.

Description of the Change:

Registers a new overloaded implementation of _equal for arguments of type ControlledSequence. For two objects of this type, qml.equal compares if:

• their control properties are the same
• their base's are the same to the requested tolerances and parameter checks.

For example:

op1 = qml.ControlledSequence(qml.PauliX(0), control=1)
op2 = qml.ControlledSequence(qml.PauliX(0), control=1)
qml.equal(op1, op2)

>>> True

Benefits:

Correct comparison of two ControlledSequence objects is now possible.

Possible Drawbacks:

None.

Related GitHub Issues:

Closes: #4797

[EmilianoG-byte]

Hi!

I have based this PR on an older commit 970d30c (from 26.10) since the newest commit was throwing some import errors when I pulled from master, and this was one of the last ones were the CI checks were passing. I see the only conflict is with the changelog, so hopefully doing it this way was ok :)

Let me know what you think about my changes!

[rmoyard]

Hello @EmilianoG-byte , thanks for your contribution! I will give an initial review of your PR this afternoon.

[rmoyard]

Thank you @EmilianoG-byte your PR looks good 💯 I have left some comments.

Concerning your import issue, I think it might be that we added pytest-benchmark mandatory to run the tests suite, you should pip install the package and let us know if the issue persists.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Comparison is base (02295e4) 99.66% compared to head (a4744cb) 99.65%.
Report is 1 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #4829      +/-   ##
==========================================
- Coverage   99.66%   99.65%   -0.01%     
==========================================
  Files         382      382              
  Lines       34465    34211     -254     
==========================================
- Hits        34348    34093     -255     
- Misses        117      118       +1     

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

[lillian542]

This looks good @EmilianoG-byte, thanks for the thorough tests! 🚀

[EmilianoG-byte]

This looks good @EmilianoG-byte, thanks for the thorough tests! 🚀

Thank you 😃 @lillian542 !!

[EmilianoG-byte]

Hi @rmoyard, I made the changes we discussed 😊
Let me know if some of them were still up for discussion or if they are okay!

[rmoyard]

Thanks for the PR 💯

[EmilianoG-byte]

thank you @rmoyard and @lillian542!

Two final things:
1- Regarding the issue I mentioned, it turns out it was a problem on my side. I was using an environment with python 3.8, so I upgraded to 3.11 and everything was fine! The pipeline on the master branch was failing in most recent commits tho, is that okay?

2- Any thoughts on my comment regarding the parametrisation of the first functions? In general, I wanted to know if using pytest parametrize like that was a standard/acceptable practice. I have never seen it being used like that, i.e. having the parameter be the function to test:

@pytest.mark.parametrize("test_class", [Controlled, ControlledSequence])

[rmoyard]

Hi @EmilianoG-byte, I am glad you solved the issue! For the second part, in general you are welcome to use it but I also like when it is tested separated. It makes it easier to refactor if needed.

[Alex-Preciado]

Awesome! Thank you so much @EmilianoG-byte. Merging this now 😎

[EmilianoG-byte]

Awesome! Thank you so much @EmilianoG-byte. Merging this now 😎

Thank you @Alex-Preciado 😄