I am trying to find the derivative of a function (that has input as the converged values of Jacobi iterator), with respect to the elements of the matrices used in Jacobi iterations. For example, if f(x) is a function of x, which in turn comes from Ax = b, I am trying to find the derivative of f with respect to element of A and b.
I am using reverseDiff.jl and forwardDiff.jl for this purpose, but all I can get is derivative of f with respect to the converged x, even when I pass b as parameter in the GradientConfig and GradientTape functions.
I have done this in C++ using CoDiPack library, but no success in Julia. Is this even possible in Julia? Any help would be appreciated. Thanks.