Implement qml.equal comparison for BasisRotation

[albi3ro]

The BasisRotation template cannot currently be compared with qml.equal or == due the presence of a numpy array as a hyperparameter.

For example, if we run:

weights = np.array(
    [
        [-0.618452, -0.68369054 - 0.38740723j],
        [-0.78582258, 0.53807284 + 0.30489424j],
    ]
)
op = qml.BasisRotation(wires=range(2), unitary_matrix=weights)
op2 = qml.BasisRotation(wires=range(2), unitary_matrix=np.array(weights))
qml.equal(op, op2)

File /pennylane/ops/functions/equal.py:209, in _equal_operators(op1, op2, check_interface, check_trainability, rtol, atol)
    206 if op1.wires != op2.wires:
    207     return False
--> 209 if op1.hyperparameters != op2.hyperparameters:
    210     return False
    212 if check_trainability:

ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()

To fix this issue, we need to register a custom comparison with the single dispatch function _equal defined here

pennylane/pennylane/ops/functions/equal.py

Line 160 in 9781f11

def _equal(

Examples of custom kernels can be found in that file. Ths op.hyperparameters['unitary_matrix'] hyperparameter should be compared with qml.math.allclose to the requested relative and absolute tolerances (atol, rtol).

The code for BasisRotation can be found:

pennylane/pennylane/templates/subroutines/basis_rotation.py

Line 25 in 9781f11

class BasisRotation(Operation):

Once custom comparison is implemented, the xfail mark can be removed from this test that runs qml.ops.functions.assert_valid:

pennylane/tests/templates/test_subroutines/test_basis_rotation.py

Line 23 in 9781f11

@pytest.mark.xfail # to be fixed by shortcut story 49160

If the standard validity checks still fail after custom comparison is implemented, we can reassess any further steps.

A PR for this issue should:

1. Add a kernel to _equal for BasisRotation
2. Remove the xfail mark from the standard validity test
3. Add a changelog entry to docs/releases/changelog-dev.md

[minhtriet]

I would like to work on this if you don't mind

[albi3ro]

I've assigned you the issue @minhtriet . Feel free to combine this and #4894 if that is easiest for you.

[albi3ro]

Closed by #4919