I. Title
SIAM Short Course on Using the Scalable PETSc Solvers

II. Organizer

Barry Smith
Argonne National Laboratory

III. Associated SIAM Conference:
SIAM Conference on Parallel Processing for Scientific Computing
February 22-24, 2006
San Francisco, California
Sir Francis Drake Hotel
(Short course to be held at the same hotel on February 21, 2006)

IV. Rationale

Many large-scale numerical simulations require the parallel generation and solution of algebraic systems of equations. There are a large number of available algorithms and software packages for solving these systems. The Portable Extensible Toolkit for Scientific computation (PETSc) provides a common interface for many of these solvers.

V. Instructors

Barry Smith, Argonne National Laboratory, is one of the two original developers of PETSc. He has led the project for the past ten years.

Other PETSc developers as appropriate.

VI. Course Description

We will teach how one defines and solves large scale linear and nonlinear algebraic equations arising from the discretization of PDEs using PETSc.

VII. Level of the Material

60% beginner
30% intermediate
10% advanced

VIII. Target Audience

Graduate students and researchers in computational science.

IX. Recommended Background
Experience with Unix and C, C++ or Fortran programming.
Elementary understanding of linear algebra and partial differential equations.
Experience with MPI or parallel computing is not assumed.

X. Course Outline

All attendees should bring a laptop (with compilers)

Introductory example (15 minutes)
Hands-on: Installing PETSc (25 minutes)
Using the solvers for structured grids (15 minutes)
Vectors and Matrices (30 minutes)
Using the solvers for unstructured grids (30 minutes)
Available preconditioners (30 minutes)
Hands-on: Experimenting with preconditioners (15 minutes)
Available Krylov methods (15 minutes)
Hands-on: Experimenting with Krylov (10 minutes)
Multigrid methods (30 minutes)
Hands-on: Experimenting with multigrid (15 minutes)
Working with Python (15 minutes)
The TOPS Solver Component (15 minutes)