Conference paper · IEEE High Performance Extreme Computing… 2026

Learning to Select Sparse Linear Solvers with Convolutional Neural Networks

Artemis PadosAlan EdelmanEmmanuel LujanDaniel PickardFelipe ToméChristopher Rackauckas

IEEE High Performance Extreme Computing Conference (HPEC), 2026

Program Cite this paper

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.

Published in IEEE High Performance Extreme Computing Conference (HPEC), 2026.

Cite this paper

Download .bib ↓

Artemis Pados, Alan Edelman, Emmanuel Lujan, Daniel Pickard, Felipe Tomé and Christopher Rackauckas, “Learning to Select Sparse Linear Solvers with Convolutional Neural Networks,” IEEE High Performance Extreme Computing Conference (HPEC), 2026.

BibTeX · Conference paper 
@inproceedings{pados2026learning,
  title={Learning to Select Sparse Linear Solvers with Convolutional Neural Networks},
  author={Pados, Artemis and Edelman, Alan and Lujan, Emmanuel and Pickard, Daniel and Tom{\'{e}}, Felipe and Rackauckas, Christopher},
  booktitle={IEEE High Performance Extreme Computing Conference (HPEC)},
  year={2026},
  month={09},
  url={https://ieee-hpec.org/ieee-hpec-2026-agenda/}
}

All publications →