spleaf.cov.Cov.dotL_back

Cov.dotL_back(grad_y)

Backward propagation of the gradient for dotL().

Propagate the gradient of a function with respect to y, to its gradient with respect to x and to the components of \(L\) (U, W, phi, G).

Use cholesky_back() to propagate the gradient to the initial components of the matrix (A, U, V, phi, F).

Parameters:

grad_y(n,) ndarray

Gradient of the function with respect to y.

Returns:

grad_x(n,) ndarray

Gradient of the function with respect to x.