ThePortable, Extensible Toolkit for Scientific Computation (PETSc, pronounced PET-see; the S is silent), is a suite ofdata structures androutines developed byArgonne National Laboratory for thescalable (parallel) solution ofscientific applications modeled bypartial differential equations. It employs theMessage Passing Interface (MPI) standard for all message-passing communication. PETSc is the world’s most widely used parallel numerical software library forpartial differential equations andsparse matrix computations. PETSc received an R&D 100 Award in 2009.^[1][2][3] The PETSc Core Development Group won the SIAM/ACM Prize in Computational Science and Engineering for 2015.^[4]

PETSc is intended for use inlarge-scale application projects, many ongoing computational science projects are built around the PETSclibraries. Its careful design allows advanced users to have detailed control over the solution process. PETSc includes a large suite of parallellinear andnonlinearequation solvers that are easily used in application codes written inC,C++,Fortran and nowPython. PETSc provides many of the mechanisms needed within parallel application code, such as simple parallelmatrix andvector assembly routines that allow the overlap ofcommunication andcomputation. In addition, PETSc includes support for parallel distributedarrays useful forfinite difference methods.^[5]

PETSc consists of a variety of components consisting of majorclasses and supporting infrastructure. Users typically interact withobjects of the highest level classes relevant to their application, essential lower level objects such as vectors, and may customize or extend any others. All major components of PETSc have an extensible plugin architecture.

PETSc provides many features for parallel computation, broken into several modules:

• Index sets, includingpermutations, forindexing into vectors, renumbering, etc.
• Parallelvectors; andmatrices (generallysparse)
• Scatters (handles communicating ghost point information) andgathers (the opposite of scatters)
• Data management for parallelstructured andunstructured meshes
• Severalsparsestorageformats
• Scalable parallelpreconditioners, includingmultigrid and sparse direct solvers
• Krylov subspace methods
• Parallel nonlinear solvers, such asNewton's method and nonlinear GMRES
• Paralleltime-stepping (ODE andDAE) solvers
• Paralleloptimization solvers, such asBFGS
• Automatic profiling offloating point andmemory usage
• Consistentinterface
• Intensive error checking
• Portable toUNIX,Mac OS X, andWindows

• List of numerical libraries

• PETSc Users Manual, Satish Balay, Shrirang Abhyankar, Mark F. Adams, Jed Brown, Peter Brune, Kris Buschelman, Victor Eijkhout, William D. Gropp, Dinesh Kaushik, Matthew G. Knepley,Lois Curfman McInnes, Karl Rupp, Barry F. Smith, and Hong Zhang, ANL-95/11 Revision 3.5, Argonne National Laboratory, June, 2014.
• Efficient Management of Parallelism in Object Oriented Numerical Software Libraries, Satish Balay, William D. Gropp, Lois Curfman McInnes, Barry F. Smith, Modern Software Tools in Scientific Computing, ed. Bruaset et al., pp. 163–202, 1997.
• Numerical simulation of geodynamic processes with the Portable Extensible Toolkit for Scientific Computation, R.F. Katz, M.G. Knepley, B. Smith, M. Spiegelman, and E.T. Coon, Physics of The Earth and Planetary Interiors, 163, pp. 52-68, 2007.
• Ed Bueler: "PETSc for Partial Differential Equations: Numerical Solutions in C and Python", SIAM,ISBN 978-1-611976-30-4 (2020).

• The Official PETSc web site

• Numerical libraries
• Numerical linear algebra
• Scientific simulation software
• C++ numerical libraries

• CS1 maint: archived copy as title
• CS1 maint: numeric names: authors list