Pull Request Test Coverage Report for Build 3329512584

• 5 of 6 (83.33%) changed or added relevant lines in 1 file are covered.
• 2 unchanged lines in 1 file lost coverage.
• Overall coverage decreased (-0.003%) to 84.404%

Totals
Change from base Build 3327479795:	-0.003%
Covered Lines:	61739
Relevant Lines:	73147

💛 - Coveralls

@kdk: yes, if the parameter is complex. From attaching pdb to the traceback, the issue here is that somewhere in the process, the time is getting turned into 0 + 0.23*I, which can't be cast to float. You can see that Scipy actually does try to cast the whole array to float first, and it calls the correct methods in ParameterExpression, which should work:

In [10]: from qiskit.circuit import Parameter
    ...: import numpy as np
    ...:
    ...: t = Parameter("t")
    ...: t.bind({t: 0.25}), np.array([t.bind({t: 0.25})], dtype=float)
Out[10]: (ParameterExpression(0.25), array([0.25]))

Thanks for tracking that down! Another solution would be to leave validate_parameters as is and instead ensure that the time is a float inside the matrix synthesis. That would have the advantage that we can raise a specific error message so I changed it. But I'm also happy to revert if you think it should remain in validate_parameters.

Using the regression test added in this PR, but modified for Qiskit 1.0:

from qiskit.circuit import Parameter, QuantumCircuit
from qiskit.quantum_info import Pauli, Operator
from qiskit.circuit.library import PauliEvolutionGate
from qiskit.synthesis import MatrixExponential

op = Pauli("X")
time = Parameter("t")
evo_gate = PauliEvolutionGate(op, time, synthesis=MatrixExponential())
circuit = QuantumCircuit(1)
circuit.append(evo_gate, [0])
bound = circuit.assign_parameters({time: 0.12})

ref = QuantumCircuit(1)
ref.rx(0.24, 0)

assert Operator(bound).equiv(ref)

this no longer fails, and #7507 is no longer reproducible either, so I'll close this as obsoleted. Feel free to reopen if there's more to be done.