tfp.substrates.numpy.math.hpsd_quadratic_form_solve

Computes rhs^T matrix^-1 rhs, where matrix is HPSD.

tfp.substrates.numpy.math.hpsd_quadratic_form_solve(
    matrix, rhs, cholesky_matrix=None
)

Given matrix and rhs, computes rhs^T @ matrix^-1 rhs, where matrix is a Hermitian positive semi-definite matrix.

matrix Floating-point Tensor of shape [..., N, N]. Represents a Hermitian positive semi-definite matrix.
rhs Floating-point Tensor of shape [..., N, K].
cholesky_matrix (Optional) Floating-point Tensor of shape [..., N, N] that represents a Cholesky factor of matrix.

hpsd_quadratic_form_solve Tensor of shape [..., K, K].