sparsax.sample_gaussian¶
-
sparsax.sample_gaussian(Ai, Aj, Ax, b, z, want_logdet=
False)[source]¶ Draw
eta ~ N(A^{-1} b, A^{-1})from a single factorization ofA.The Gibbs Gaussian-conditional step in one factorization: the posterior mean
A^{-1} band the correlated draw share one Cholesky factor. WithA = P' L L' P, a sample iseta = mean + P' L^{-T} zforz ~ N(0, I), sinceCov(P' L^{-T} z) = A^{-1}. Built onfactor_solve(), so undervmapit factors once per element.