cs.LGNov 9, 2026

Learning to Select Sparse Linear Solvers with Convolutional Neural Networks

Authors: Artemis Pados, Alan Edelman, Emmanuel Lujan, Daniel Pickard, Felipe Tome, Christopher Rackauckas

Organizations: Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, USA. · Computer Science & Artificial Intelligence Laboratory, Massachusetts Institute of Technology, USA. · Department of Aeronautics and Astronautics, Massachusetts Institute of Technology, USA.

Abstract

We investigate convolutional-neural-network (CNN) based automatic solver selection for sparse linear systems, a problem of practical importance where solver performance varies dramatically with matrix sparsity structure. A CNN is trained to regress log timing ratios between candidate solvers using spy plot images of matrix nonzero structure as input, achieving strong prediction accuracy on both held-out SuiteSparse matrices and real-world matrices from high-performance simulation codes.

Figures & tables

Explore similar work

CardsList
  1. Learning to Discover Iterative Spectral Algorithms

    Feb 10, 2026Zihang Liu, Oleg Balabanov, Yaoqing Yang +1Automated Algorithm DiscoveryLearning to Optimize

  2. Estimating Condition Number with Graph Neural Networks

    Mar 10, 2026Erin Carson, Xinye Chen

  3. RINS: Residual-Image Neural Subspace Solvers for Large Sparse Linear Systems

    Sep 9, 2026Zhongyan Ouyang, Weixin Liao, Mingquan Feng +2Neural PDE SolversLearning to Optimize