posted on September 4, 2013
Dr. Matthew G. Knepley, University of Chicago, will present a hands-on tutorial for PETSc at MSI on September 30. Dr. Knepley is one of the authors of the widely used PETSc library for scientific computing from Argonne National Laboratory, and is a principal designer of the PyLith library for the solution of dynamic and quasi-static tectonic deformation problems.
PETSc is a suite of data structures and routines for the scalable parallel solution of scientific applications, often modeled by partial differential equations. It supports MPI, shared memory pthreads, and GPUs, as well as hybrid MPI-shared memory pthreads or MPI-GPU parallelism.
The tutorial will be held on Monday, September 30, from 9 a.m. – 4 p.m. There will be an hour for lunch (12 – 1 p.m.). The tutorial will be held in room 575 Walter. Register on the Tutorial page.
As workstations are limited, participants are invited to bring their own laptops to use during the tutorial. Please visit the PETSc website and install the program. You can install the latest release, but the most up-to-date snapshot is preferred, which uses the “git checkout next” branch; download it on the developers site.
You can also view prior tutorials and a User’s Manual plus extensive additional documentation.
Tutorial Description
This tutorial will cover basic sparse parallel linear algebra, as well as linear and nonlinear algebraic solvers, and timestepping. Students will learn to use the dynamic configuration object system in PETSc, and use it to assemble hierarchical, multilevel solvers for multiphysics problems. The tutorial also includes how to debug and profile the resulting applications.
The morning session will cover basic usage:
- Vectors and Matrices
- Generic Solver Setup
- Dynamic Object Configuration
- Debugging and Profiling
The afternoon session will cover advanced usage:
- Structured Meshes Using DMDA
- Linear Preconditioning
- Structured Linear Multigrid
- Block Preconditioners
- Nonlinear Solvers
- Nonlinear Preconditioners
- Timestepping
Instructor Biography
Matthew G. Knepley received a B.S. in Physics from Case Western Reserve University in 1994, an M.S. in Computer Science from the University of Minnesota in 1996, and a Ph.D. in Computer Science from Purdue University in 2000. He was a Research Scientist at Akamai Technologies in 2000 and 2001. Afterwards, he joined the Mathematics and Computer Science department at Argonne National Laboratory (ANL), where he was an Assistant Computational Mathematician, and a Fellow in the Computation Institute at University of Chicago. In 2009, he joined the Computation Institute as a Senior Research Associate. His research focuses on scientific computation, including fast methods, parallel computing, software development, numerical analysis, and multicore architectures. He is an author of the widely used PETSc library for scientific computing from ANL, and is a principal designer of the PyLith library for the solution of dynamic and quasi-static tectonic deformation problems. He developed the PETSc scalable unstructured mesh support based upon ideas from combinatorial topology. He was a J.T. Oden Faculty Research Fellow at the Institute for Computation Engineering and Sciences, University of Texas (Austin, TX), in 2008, and won the R&D 100 Award in 2009 as part of the PETSc team.