Fix CircuitDrawer single wire multiple observables

[antalszava]

Context
QNodes can have multiple measurements on the same wire so long as they are part of qubit-wise commuting observables. Drawing such QNodes results in an error, unfortunately.

This boils down to how the CircuitGraph processes observables. It assumes that there is a single observable for every QNode or alternatively it merges observables for separate return measurement processes. When there are more than one observable per wire, it causes an error

Changes

Changes the logic in the greedy_layers method of the CircuitGraph class to account for multiple cases where there was a wire that was measured more than once.

Note: in such cases the circuit includes diagonalizations and the observables will be drawn as PauliZ operators.

Ralated Issues
Closes #1326

[github-actions]

Hello. You may have forgotten to update the changelog!
Please edit .github/CHANGELOG.md with:

• A one-to-two sentence description of the change. You may include a small working example for new features.
• A link back to this PR.
• Your name (or GitHub username) in the contributors section.

[codecov]

Codecov Report

Merging #1353 (f22d7ac) into master (305279a) will increase coverage by 0.00%.
The diff coverage is 100.00%.

@@           Coverage Diff           @@
##           master    #1353   +/-   ##
=======================================
  Coverage   98.27%   98.28%           
=======================================
  Files         150      150           
  Lines       11262    11285   +23     
=======================================
+ Hits        11068    11091   +23     
  Misses        194      194           

Impacted Files	Coverage Δ
pennylane/circuit_graph.py	93.30% <100.00%> (+0.76%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 305279a...f22d7ac. Read the comment docs.

[josh146]

Thanks @antalszava! Have left some relatively minor thoughts and questions.

[josh146]

Is there a reason this helper function is nested/using closure? If not, you could just define it top level at the top of the module file (_is_returned_obs) to avoid the overhead of creating the function within the loop.

[antalszava]

Right! Moved it to the top.

[josh146]

Two thoughts when reading this:

• I would prefer a shorter name for this method; the compute seems a bit redundant. Maybe something like max_concurrent_measurements? Measurements is more general than observable.

• Maybe the summary docstring could be made a bit more clear, it took me a couple of times to read it to realise what it meant

[antalszava]

Thanks! Went with max_measure_per_wire, how would that sound? Also adjusted the docstring.

[josh146]

To be honest, I find the 'per wire' a bit confusing, since it sounds like it will be returning a list (with each element corresponding to a wire).

Something like 'max_simultaneous_measurements` is much more identifiable to me?

[antalszava]

Sounds good, changed to that! Thanks :)

[josh146]

Thanks Antal! Happy to approve now, but recommend renaming the method to avoid the 'per_wire' terminology