Module svd 

SVD decomposition for tensors.

Provides svd and svd_with for computing truncated SVD of TensorDynLen values. The tensor is unfolded into a matrix by splitting its indices into left and right groups, then the standard matrix SVD is computed and truncated according to SvdOptions.

This module works with concrete types (DynIndex, TensorDynLen) only.

Structs

SvdOptions

Options for SVD decomposition with truncation control.

Enums

SvdError

Error type for SVD operations in tensor4all-linalg.

Functions

default_svd_truncation_policy

Get the global default truncation policy for SVD.

set_default_svd_truncation_policy

Set the global default truncation policy for SVD.

svd

Compute SVD decomposition of a tensor with arbitrary rank, returning (U, S, V).

svd_with

Compute SVD decomposition of a tensor with arbitrary rank, returning (U, S, V).