Statistical Properties of the Probabilistic Numeric Linear Solver BayesCG
From MaRDI portal
Publication:6407239
arXiv2208.03885MaRDI QIDQ6407239FDOQ6407239
Authors: Tim W. Reid, Ilse C. F. Ipsen, Jon Cockayne, Chris J. Oates 
Publication date: 7 August 2022
Abstract: We analyse the calibration of BayesCG under the Krylov prior, a probabilistic numeric extension of the Conjugate Gradient (CG) method for solving systems of linear equations with symmetric positive definite coefficient matrix. Calibration refers to the statistical quality of the posterior covariances produced by a solver. Since BayesCG is not calibrated in the strict existing notion, we propose instead two test statistics that are necessary but not sufficient for calibration: the Z-statistic and the new S-statistic. We show analytically and experimentally that under low-rank approximate Krylov posteriors, BayesCG exhibits desirable properties of a calibrated solver, is only slightly optimistic, and is computationally competitive with CG.
Has companion code repository: https://github.com/treid5/ProbNumCG_Supp
Bayesian inference (62F15) Linear equations (linear algebraic aspects) (15A06) Iterative numerical methods for linear systems (65F10)
This page was built for publication: Statistical Properties of the Probabilistic Numeric Linear Solver BayesCG
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6407239)
