Welcome to the PROPACK homepage

The software package PROPACK contains a set of functions for computing the singular value decomposition of large and sparse or structured matrices. The SVD routines are based on the Lanczos bidiagonalization algorithm with partial reorthogonalization (BPRO). The Lanczos routines can also be used directly, and form the basis of efficient algorithms for solving linear systems of equations and linear least squares problems, in particular for systems with multiple right-hand sides.

Software

• Fortran version:

  A short overview and installation instructions are given here.

  • Version 2.1 (April 20, 2005): PROPACK77_v2.1.tar.gz. Changelog.
  • Version 2.0 (March 17, 2004): PROPACK77_v2.0.tar.gz. Changelog.
  • Version 1.2 (Jan 30, 2004): PROPACK77_v1.2.tar.gz. Changelog.
  • Version 1.1 (June 13, 2003): PROPACK77_v1.1.tar.gz. Changelog.
  • Version 1.0 (May 1, 2001): PROPACK77_v1.0.tar.gz.

  The Fortran version is currently supported on the following platforms:

  OS	architecture	compiler	OpenMP version
  Linux	ia32	GNU g77	No
  Linux	ia32	Intel ifort	Yes
  Linux	x86_64	GNU g77	No
  Linux	x86_64	Intel ifort	Yes
  Linux	ia64	GNU g77	No
  Linux	ia64	Intel ifort	Yes
  SunOS	Sparc	Sun WorkShop Compilers f90	No
  IRIX	MIPS	MIPSpro f77	Yes
  AIX	Power 4	IBM xlf	Yes
  Mac OS X / Darwin	PPC	IBM xlf	No

• Matlab® version:
  • Version 1.1 (March 2004): PROPACK_1.1_Matlab.tar.gz
  • Version 1.0 (1998): PROPACK.tar.gz

  An overview of the Matlab toolbox is given here. The Matlab routines come with detailed on-line documentation, which can be read by typing "help " at the Matlab prompt. The online documentation for the 4 main routines can be reached via the links below:
  

  • lanbpro.m: Lanczos bidiagonalization with partial reorthogonalization: here.
  • lansvd.m: Singular value decomposition: here.
  • lanpro.m: Hermitian Lanczos with partial reorthogonalization: here.
  • laneig.m: Eigenvalues and eigenvectors of a hermitian matrix: here.

• Test matrices from MatrixMarket: TESTMATRICES.tar.gz.

Documentation

• R. M. Larsen, Lanczos bidiagonalization with partial reorthogonalization, Department of Computer Science, Aarhus University, Technical report, DAIMI PB-357, September 1998. This report describes the initial work that lead to PROPACK.
  Download here in gzipped postscript or pdf format.
• R. M. Larsen, Computing the SVD for Large and Sparse Matrices, SCCM, Stanford University, June 16, 2000. A presentation of PROPACK given on June 16, 2000 at Lawrence Berkeley National Laboratory.
  Download here in gzipped postscript or pdf format.
• R. M. Larsen, Combining implicit restart and partial reorthogonalization in Lanczos bidiagnalization, SCCM, Stanford University, April 2001. A presentation of the implicitly restarted SVD algorithm available in PROPACK version 1.0. This talk was given on April 18, 2001 at UC Berkeley. It includes performance measurements of the Fortran version of PROPACK and comparisons with PLANSO and ARPACK.
  Download here in gzipped postscript or pdf format.