Module: tfp.experimental.fastgp.linalg

Linear algebra routines to use for preconditioners.

Modules

partial_lanczos module: Run Lanczos for m iterations to get a good preconditioner.

Classes

class Array: Array base class for JAX

Functions

largest_eigenvector(...): Returns the largest (eigenvalue, eigenvector) of M.

make_partial_lanczos(...): Return low rank approximation to M based on the partial Lancozs alg.

make_partial_pivoted_cholesky(...): Return low rank approximation to M based on partial pivoted Cholesky.

make_randomized_truncated_svd(...): Returns approximate SVD for symmetric M.

make_truncated_svd(...): Return low rank approximation to M based on the partial SVD alg.