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.
The gradient of the parametrized Hermitian operator with respect to its input matrix would need to be obtained. This would likely be the "heavy lifting". Once that's done, evolving the internal states as per the adjoint differentiation logic could yield the gradient of the circuit.
Feature details
https://github.com/PennyLaneAI/pennylane/pull/2543 noted that training
qml.Hermitian
fordiff_method="adjoint"
is not supported. Supporting this feature could still be feasibleImplementation
The gradient of the parametrized Hermitian operator with respect to its input matrix would need to be obtained. This would likely be the "heavy lifting". Once that's done, evolving the internal states as per the adjoint differentiation logic could yield the gradient of the circuit.
How important would you say this feature is?
1: Not important. Would be nice to have.
Additional information
No response