SIAM/ACM Prize in Computational Science and Engineering
Argonne National Laboratory United States – 2014
CITATION

The PETSc Core Development Team: Satish Balay, Jed Brown, William Gropp, Matthew Knepley, Lois Curfman McInnes, Barry Smith, and Hong Zhang

Press Release

BACKGROUND
 
PETSc, an initiative of Argonne National Laboratory, where it continues to be developed,
employs the Message Passing Interface (MPI), a standardized, portable message-passing system
used by modern computer software on a wide variety of parallel computers. PETSc, which is
easy to use for beginners, is carefully designed to allow advanced users to have detailed control
over the solution process.
 
The project is a collaboration among William Gropp of the University of Illinois; Lois Curfman
McInnes, Satish Balay, Jed Brown, Barry Smith, and Hong Zhang of Argonne National
Laboratory; and Matt Knepley of the University of Chicago.
 
The prize is being awarded to Satish BalayJed BrownWilliam GroppMatthew KnepleyLois
Curfman McInnesBarry Smith, and Hong Zhang for their collaborative work in developing the
PETSc software package, which has transformed the way large-scale software libraries are
developed, supported, and used within the CS&E community. The creation of this innovative and
seminal numerical software package provides the scientific and engineering community with
robust, efficient, scalable, and extensible tools that are the backbone of numerous high
performance applications. The sustained impact of this work has been felt worldwide.
 
The team's innovations have had a powerful and positive effect on the high performance
computing community, have improved support for users, and helped to foster deep partnerships
within scientific teams.
 
PETSc includes a large suite of parallel linear, nonlinear equation solvers, and ordinary
differential equation (ODE) integrators that are usable for application codes written in C, C++,
Fortran and Python. These tools enable scalable solutions of scientific applications that are
modeled by partial differential equations. Designed using an object-oriented architecture, PETSc
is built on a concept that allows many people to contribute and maintain their own libraries in a
distributed fashion.