CS 6220
Data-Sparse Matrix Computations
Matrices and linear systems can be data-sparse in a wide variety of ways, and we can often leverage such underlying structure to perform matrix computations efficiently. This course will discuss several varieties of structured problems and associated algorithms. Example topics include randomized algorithms for numerical linear algebra, Krylov subspace methods, sparse recovery, and assorted matrix factorizations. Students must have a strong background in linear algebra, programming experience, and prior exposure to numerical methods.
Topics
- Students will be able to describe the key ideas behind fast algorithms for rank-structured matrices.
- Students will compare different methods in randomized numerical linear algebra and argue which techniques are most applicable for certain problem settings.
- Students will be able to develop and analyze specialized algorithms for data-sparse matrices.
- Students will be able to evaluate the performance of algorithms for data-sparse matrices.
- Students will be able to read, synthesize, and summarize current research in the field.
- Students will be able to select appropriate methods for numerical linear algebra problems arising in their application areas of interest.
Official roster learning outcomes
Sources
- Course catalogue ↗2026-2027