Customized Sparse Eigenvalue Solutions in Lighthouse

Graduate Thesis Or Dissertation

Citations

Citeable URL: https://scholar.colorado.edu/concern/graduate_thesis_or_dissertations/s4655h01m

Abstract

We come across sparse eigenvalue problems in many areas of scientific computing. A variety of high-performance numerical software packages with many different eigensolvers are available to solve such problems. Two main challenges arise while solving these problems: 1) finding the eigensolver routine that can correctly solve the problem and 2) implementing the desired solution accurately and efficiently using the appropriate software package. Our work overcomes these two challenges by intelligently identifying the sparse eigensolvers that are likely to perform the best for the given user-specified input characteristics and auto generating customized code template that uses the identified best solver. Lighthouse Taxonomy is a novel interface and search platform for users seeking software solutions to linear algebra problems. For this thesis, we integrated SLEPc into Lighthouse to support sparse eigenvalue solutions.

Creator