Fast Linear Solvers via AI-Tuned Markov Chain Monte Carlo-based Matrix Inversion

ePubs

The open archive for STFC research publications

Persistent URL	http://purl.org/net/epubs/work/63565151
Record Status	Checked
Record Id	63565151
Title	Fast Linear Solvers via AI-Tuned Markov Chain Monte Carlo-based Matrix Inversion
Contributors	A Lebedev (STFC Daresbury Lab.) (Pr.Au.), W Lee (STFC Daresbury Lab.) (Pr.Au.), S Ghosh (IBM Research Yorktown Heights), O Ivanyshyn Yaman (STFC Daresbury Lab.), V Kalantz (IBM Research Yorktown Heights), Y Lu (IBM Research Yorktown Heights), T Nowicki (IBM Research Yorktown Heights), S Ubaru (IBM Research Yorktown Heights), L Horesh (IBM Research Yorktown Heights), V Alexandrov (STFC Daresbury Lab.)
Abstract	A presentation of our development of a recommendation system using graph neural networks to optimise continuously the parameters used in the Markov Chain Monte Carlo for Matrix Inversion (MCMCMI) methods accompanying the linked conference paper. We demonstrate the approach taken, issues observed and results obtained.
Organisation	STFC , HC
Keywords	numerical linear algebra , recommendation systems , Markov Chain Monte Carlo , AI , MCMCMI
Funding Information	Hartree National Centre for Digital Innovation, Explore (HT07066)
Related Research Object(s):	63563870
Licence Information:	Creative Commons Attribution 4.0 International (CC BY 4.0)
Language	English (EN)

Type	Details	URI(s)	Local file(s)	Year
Presentation	Presented at Supercomputing 25 (SC25), St. Louis, United States of America, 16-21 Nov 2025.		ScalAH25_SC25_pre…tation_16Nov2025.pdf	2025