thomasahle
commented
1 month ago Should be easy after adding determinant and matrix inverse functions. Then use these derivative rules:
∂(det(X)) = det(X)Tr(X⁻¹ ∂X)
∂(X⁻¹) = −X⁻¹ (∂X) X⁻¹These functions would also serve as good examples of functions that are applied along two axes instead of the current functions, which are applied along just one.
It would be nice to be able to solve tasks like this: https://math.stackexchange.com/questions/4709572/what-is-the-hessian-of-x-mapsto-log-det-left-at-a-rt-operatornamediag?noredirect=1&lq=1