Other than the two famous parameter-shift rule and finite difference method, is there any other (perhaps simple) methods to estimate gradients of a cost function in variational quantum algorithms?
$\begingroup$
$\endgroup$
Add a comment
|
1 Answer
$\begingroup$
$\endgroup$
One scheme that's been used quite successfully in practice is SPSA. If you're using a simulator and keeping track of the entire state vector, another really efficient method is adjoint differentiation.
