cupynumeric.linalg.pinv#

cupynumeric.linalg.pinv(
a: ndarray,
rtol: float = 1e-05,
) → ndarray#

Compute the (Moore-Penrose) pseudo-inverse of a matrix.

Parameters:

  • a ((M, N) array_like) – Matrix to be pseudo-inverted.

  • rtol (float, optional) – Cutoff for small singular values relative to the largest singular value. Singular values less than or equal to rtol * largest_singular_value are set to zero. Default: 1e-5.

Returns:

B – The pseudo-inverse of a.

Return type:

(N, M) ndarray

Raises:

  • NotImplementedError: – If the dimension of the array is greater than 2 or if M < N for two-dimensional arrays

  • LinAlgError: – If the dimension of the array is less than 2

See also

numpy.linalg.pinv

Similar function in NumPy

Availability:

Single GPU, Single CPU

Notes

  • The SVD part of the computation supports only single process execution

  • Does not support batched operations

  • Does not support rcond parameter (use rtol instead)

  • Does not support hermitian parameter (will be supported once cupynumeric.svd supports hermitian option)

On this page