pint.models.noise_model.project_basis_covariance

pint.models.noise_model.project_basis_covariance(U: ndarray, Phi: ndarray) → ndarray

Project basis-space covariance to data-space covariance. U should be (N_toa, N_basis_functions) Phi should be (N_basis_functions, N_basis_functions) or (N_basis_functions,). For time domain noise, Phi is diagonal for the RIDGE kernel but 2D for other kernels. For Fourier basis noise, Phi is diagonal for the standard powerlaw bases.