Conference on Fast Direct Solvers

Purdue University, West Lafayette IN

Investigators

Abstract

The 2018 Conference on Fast Direct Solvers will be held at Purdue University in November 9-11, 2018. The purpose of the conference is to bring together researchers from different mathematical and application fields to exchange ideas on the latest developments of fast direct solvers, discuss potential strategies for tackling the current challenges, stimulate new multidisciplinary collaborations, and disseminate relevant ideas to broader communities. Researchers, students, and practitioners from all relevant fields are encouraged to attend the conference and/or to give talks. The conference will provide a multidisciplinary platform for junior and senior researchers, students, and practitioners to discuss and to establish new collaborations. The activities will help to attract students and junior researchers from diverse background to join the field and also stimulate student interest in mathematics. This grant will be used to partially support the travel and lodging of junior researchers and students. In scientific computing, data analysis, and engineering simulations, solving large linear systems is typically a key computational component. Direct solvers do not need iterations, are very robust, and have various other attractive features. However, classical direct solvers are typically expensive for large sparse systems because of the loss of sparsity. In the past 15 years or so, the scientific computing community has witnessed quick developments of the so-called fast direct solvers, which exploit structures and can achieve significantly better efficiency than classical direct solvers, sometimes by orders of magnitude. This conference will bring together researchers, students, and practitioners from diverse fields that actively involve the design, study, and application of fast direct solvers. The invited and contributed talks will help to supply new tools and identify new research directions for direct solvers. The conference is expected to be an effective event to both advance the development of fast direct solvers and benefit more general numerical computations. Website: http://www.math.purdue.edu/~xiaj/FastSolvers2018/ This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.