Speaker: Dr. Sanjukta Bhowmick, Applied Physics and Applied Mathematics, Columbia University
The solution of large sparse linear systems is a fundamental problem in scientific computing. A variety of solution schemes are available reflecting a wide range of performance and quality trade-offs. The ``best'' solution method can vary across application domains and often even across different phases in a single application. We can tap the benefits of this variety of sparse linear solution techniques by either selecting solvers to match application characteristics and/or by using multimethod solvers,solvers that use more than one solution scheme. I will discuss how machine learning techniques can be used in solver selection and also talk about the construction and performance of multimethod solvers.