Extrapolating from Regularised Solutions for Solving Ill-Conditioned Linear Systems in Machine Learning

DOI10.48550/arXiv.2606.30328MaRDI QIDQ7270372arXivFDO

Authors Jon Cockayne, Chris. J. Oates, Disha Hegde

Publication date 29 June 2026

Published in arXiv

Tldr This paper introduces `autonugget`, a Python package for stable and accurate numerical solution of ill-conditioned linear systems with automatic differentiation support, which improves over traditional Tikhonov-regularised inversion via Richardson extrapolation.

Full work available at URL https://doi.org/10.48550/arXiv.2606.30328

arXiv classification

cs.LGmath.NAstat.ML

Cites work

• A dynamical Tikhonov regularization for solving ill-posed linear algebraic systems
• Accurate inverses for computing eigenvalues of extremely ill-conditioned matrices and differential operators
• An a posteriori parameter choice for ordinary and iterated Tikhonov regularization of ill-posed problems leading to optimal convergence rates
• Approximation of generalized inverses by iterated regularization
• Array programming with NumPy
• Average condition number for solving linear equations
• Compression, inversion, and approximate PCA of dense kernel matrices at near-linear computational complexity
• Extrapolation and the method of regularization for generalized inverses
• Gaussian Processes for Machine Learning
• Learning regularization parameters for general-form Tikhonov
• Learning the optimal Tikhonov regularizer for inverse problems
• Methods of conjugate gradients for solving linear systems
• Numerical Linear Algebra
• Numerical Optimization
• On Krylov projection methods and Tikhonov regularization
• On the choice of the regularization parameter for iterated Tikhonov regularization of ill-posed problems
• On the effectiveness of Richardson extrapolation in data science
• Probabilistic Richardson extrapolation
• Sparse Cholesky factorization for solving nonlinear PDEs via Gaussian processes
• Splines and PDEs: from approximation theory to numerical linear algebra. Lecture notes from the summer school, Cetraro, Italy, July 2--7, 2017
• The approximate arithmetic solution by finite differences of physical problems involving differential equations, with an application to the stresses in a masonry dam.
• The effect of the nugget on Gaussian process emulators of computer models
• The Splitting Extrapolation Method

This page was built for publication: Extrapolating from Regularised Solutions for Solving Ill-Conditioned Linear Systems in Machine Learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q7270372)