Integration testing for the cut_circuit_mc transform
#2407
Conversation
Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
…ithub.com:PennyLaneAI/pennylane into qcut-sample-subgraphs-to-fragment-tape-conversion
Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
…I/pennylane into qcut-cut-circuit-mc-transform
Codecov Report
@@ Coverage Diff @@
## master #2407 +/- ##
=======================================
Coverage 99.45% 99.45%
=======================================
Files 244 244
Lines 18978 18978
=======================================
Hits 18874 18874
Misses 104 104
Continue to review full report at Codecov.
|
|
[sc-16860] |
tests/transforms/test_qcut.py
Outdated
| shots = 10 | ||
| dev = qml.device("default.qubit", wires=2, shots=shots) | ||
|
|
||
| @qml.cut_circuit_mc(classical_processing_fn=fn) |
There was a problem hiding this comment.
How could we make it so that @qml.cut_circuit_mc(fn) is valid?
Perhaps we could make classical_processing_fn and max_depth (the two arguments we expect users to maybe set at QNode level) be the first and second arguments (after tape), respectively.
There was a problem hiding this comment.
to make sure I've understood, do you mean like this?:
def cut_circuit_mc(
tape: QuantumTape,
classical_processing_fn: Optional[callable],
max_depth: int,
shots: Optional[int] = None,
device_wires: Optional[Wires] = None,
) There was a problem hiding this comment.
I've made the change here a1adee8 however, I'm not totally conviced it's doing what we want. When passing a classical processing function it's happy, but if no function is passed an error is given e.g:
@qml.cut_circuit_mc(fn, 1)
works well, but
@qml.cut_circuit_mc(None, 1)
causes an error in _cut_circuit_mc_expand() which has also been updated to accommodate the new order. In the tests this is solved by using
@qml.cut_circuit_mc(classical_processing_fn=None, max_depth=1)
which actually seems worse than the original.
If we want to keep these changes I'll update the docstrings. But maybe there's something I missed?
There was a problem hiding this comment.
Thanks @anthayes92! Left a comment here.
Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
…/pennylane into qcut-extend-mc-cutting-tests
* Add summary of sample-based cutting additions to changelog, add example * remove draw and draw_mpl from transforms __init__ * add draw imports back to trasnforms __init__ file * Update * Increase shot number * Apply suggestions from code review Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com> * fix indentation Co-authored-by: trbromley <brotho02@gmail.com> Co-authored-by: Tom Bromley <49409390+trbromley@users.noreply.github.com>
Context:
Integration testing for the
cut_circuit_mctransform.Description of the Change:
Additional unit tests for mid circuit measurements, circuits with disconnected components, circuits with trivial wire cuts, and
circuits with complicated cuts