I am currently trying to train a quantum circuit on Qiskit using an Azure QPU. I have some parameterized gates that I optimize using ADAMs. However, as the output of the circuit $f(x)$ is not deterministic, the numerical gradient calculated in the optimizer varies for the same input (even after performing 100 shots and calculating the expected value). Is there a fix for this, or is my interpretation of this wrong?
Asked
Active
Viewed 44 times
