Abstract: Tridiagonal systems of equations emerge in a variety of scientific and engineering applications. These systems can be solved using sparse solvers commonly available in most math libraries. However, algorithms that explicitly leverage the tridiagonal structure of the matrices have been shown to exhibit superior execution performance, especially in the context of block tridiagonal systems. This talk will provide a brief overview of these specialized methods, describe a newly developed algorithm with superior parallel performance, discuss its scope, advantages and limitations, and end with a call for audience inputs/insights to address a related open question
Speaker’s Bio: Sudip Seal is a Senior Researcher in the System and Decision Sciences Group (CSMD) specializing in scalable parallel algorithms for large-scale applications of traditional and AI-driven methods in science and engineering.
Last Updated: October 30, 2020 - 2:44 pm