spleaf.Spleaf.cholesky_back#
- Spleaf.cholesky_back()#
Backward propagation of the gradient for the Cholesky decomposition.
Propagate the gradient of a function with respect to D and to the components of \(L\) (U, W, phi, G), to its gradient with respect to the initial components of the matrix (A, U, V, phi, F).
Use
grad_param()
to get the results.