FASTEig

Area: Mathematical libraries

CASS member: FASTMath

Description

FASTEig is a software library designed for the solution of large-scale eigenvalue problems. It provides implementations of a wide range of algorithms for symmetric and nonsymmetric, linear and nonlinear, standard and generalized eigenvalue problems. The library also supports structured eigenvalue problems, including linear response eigenvalue problems that arise in scientific and engineering applications. A distinguishing feature of FASTEig is its use of machine learning techniques to automatically select or adaptively combine appropriate algorithms based on problem characteristics supplied by the user or inferred dynamically during the solution process. In addition, FASTEig can construct effective preconditioners on the fly to accelerate the convergence of iterative methods when no user-supplied preconditioner is available.

Target audience

Eigenvalue problems arise in many science and engineering problems as well as data analysis. The target audience of the FASTEig software is computational and applied mathematicians, computational scientists and data analytic engineers.

License: BSD-3-Clause-LBNL

Availability

FASTEig is working towards its initial public release, expected by summer 2026. This page will be updated with full details of the package at that time. Until then, direct any questions to Chao Yang.