[BUG] Hamiltonian coefficients of CommutingEvolution cannot be differentiated with parameter shift

[astralcai]

Expected behavior

CommutingEvolution can be differentiated properly.

Actual behavior

import pennylane as qml

dev = qml.device('default.qubit')

@qml.qnode(dev, diff_method='parameter-shift')
def circuit(time, coeffs):
    qml.CommutingEvolution(qml.Hamiltonian(coeffs, [qml.X(0), qml.Y(0)]), time, frequencies=(2, 4), shifts=(1, 0.5))
    return qml.expval(qml.PauliZ(0))

time = qml.numpy.array(0.5)
coeffs = qml.numpy.array([1., -1.])

>>> qml.jacobian(circuit)(time, coeffs)
OperatorPropertyUndefined: The operation CommutingEvolution does not have a grad_recipe, parameter_frequencies or a generator defined. No parameter shift rule can be applied.

Additional information

If you only differentiate with respect to time then it works, but it does not work if you differentiate with respect to the coefficients of the Hamiltonian. Strangely enough, the parameter_shift transform is able to do it:

@qml.qnode(dev)
def circuit(time, coeffs):
    qml.CommutingEvolution(qml.Hamiltonian(coeffs, [qml.X(0), qml.Y(0)]), time, frequencies=(2, 4), shifts=(1, 0.5))
    return qml.expval(qml.PauliZ(0))

>>> qml.gradients.param_shift(circuit)(time, coeffs)
(tensor(-1.81859485, requires_grad=True),
 tensor([-0.45464871,  0.45464871], requires_grad=True))

But then adding diff_method="parameter-shift" makes the transform fail as well:

@qml.qnode(dev, diff_method='parameter-shift')
def circuit(time, coeffs):
    qml.CommutingEvolution(qml.Hamiltonian(coeffs, [qml.X(0), qml.Y(0)]), time, frequencies=(2, 4), shifts=(1, 0.5))
    return qml.expval(qml.PauliZ(0))

>>> qml.gradients.param_shift(circuit)(time, coeffs)
OperatorPropertyUndefined: The operation CommutingEvolution does not have a grad_recipe, parameter_frequencies or a generator defined. No parameter shift rule can be applied.

I'm confused 🤷🏻‍♂️

The issue was discovered as part of #6282, and the relevant test in test_commuting_evolution.py is xfailed, when fixed, remove the xfail from the relevant test.

Source code

No response

Tracebacks

No response

System information

Name: PennyLane
Version: 0.39.0.dev10
Summary: PennyLane is a cross-platform Python library for quantum computing, quantum machine learning, and quantum chemistry. Train a quantum computer the same way as a neural network.
Home-page: https://github.com/PennyLaneAI/pennylane
Author:
Author-email:
License: Apache License 2.0
Location: /Users/astral.cai/Workspace/pennylane/venv/lib/python3.10/site-packages
Requires: appdirs, autograd, autoray, cachetools, networkx, numpy, packaging, pennylane-lightning, requests, rustworkx, scipy, toml, typing-extensions
Required-by: PennyLane-Catalyst, PennyLane_Lightning, PennyLane_Lightning_Kokkos

Platform info:           macOS-15.0-arm64-arm-64bit
Python version:          3.10.14
Numpy version:           1.26.4
Scipy version:           1.12.0
Installed devices:
- default.clifford (PennyLane-0.39.0.dev19)
- default.gaussian (PennyLane-0.39.0.dev19)
- default.mixed (PennyLane-0.39.0.dev19)
- default.qubit (PennyLane-0.39.0.dev19)
- default.qutrit (PennyLane-0.39.0.dev19)
- default.qutrit.mixed (PennyLane-0.39.0.dev19)
- default.tensor (PennyLane-0.39.0.dev19)
- null.qubit (PennyLane-0.39.0.dev19)
- reference.qubit (PennyLane-0.39.0.dev19)
- lightning.qubit (PennyLane_Lightning-0.39.0.dev27)
- lightning.kokkos (PennyLane_Lightning_Kokkos-0.38.0)
- nvidia.custatevec (PennyLane-Catalyst-0.9.0.dev15)
- nvidia.cutensornet (PennyLane-Catalyst-0.9.0.dev15)
- oqc.cloud (PennyLane-Catalyst-0.9.0.dev15)
- softwareq.qpp (PennyLane-Catalyst-0.9.0.dev15)

Existing GitHub issues

• I have searched existing GitHub issues to make sure the issue does not already exist.