\documentstyle{article}
\oddsidemargin 0.0in
\evensidemargin 0.0in
\textwidth 6.5in
\begin{document}
\centerline{\bf CENTER FOR NUMERICAL ANALYSIS}
\centerline{\bf ITPACK SOFTWARE}
\bigskip
The ITPACK software packages are designed to solve the linear system
$Au=b$ by various iterative techniques where $A$ is a large sparse
matrix. A brief description of each version of ITPACK is described below.
\begin{description}
\item[ITPACK 2C:] This package assumes that $A$ is symmetric and
positive definite (SPD) or nearly so. The matrix is stored in the
{\tt A-JA-IA} sparse matrix format as in the Yale Sparse Matrix
Package (YSMP). With this format, the nonzeros of $A$ are stored by
rows in consecutive locations. This package uses either conjugate
gradient or Chebyshev acceleration applied to the Jacobi, SSOR
(Symmetric Successive Overrelaxation), or RS (Reduced System) basic
iterative methods. In addition, SOR (Successive Overrelaxation) is
supplied without acceleration. Adaptive procedures can be used for
iteration parameters such as $\omega$ and for eigenvalue estimates.
This package is not intended for nonsymmetric or symmetric
indefinite systems. A double precision version of this code is also
available for short-word-length computers.
\item[ITPACKV 2D:] This package is a vectorized version of the ITPACK
2C package. The algorithms used are exactly the same as those in
ITPACK 2C. The sparse matrix storage scheme was changed to another
format which allowed certain computational kernels to vectorize.
This storage format is the {\tt COEF-JCOEF} sparse matrix format
as in the ELLPACK package from Purdue for solving elliptic PDEs.
This version of ITPACKV 2D is intended for the Cray Y-MP and other
vector computers.
\item[ITPACK 3A:] This package contains acceleration routines and
basic iterative methods for both symmetric and nonsymmetric systems.
The accelerators for symmetric systems are conjugate gradient and
Chebyshev acceleration. The accelerators for nonsymmetric systems
include ORTHOMIN(s), ORTHODIR(s), ORTHORES(s), Lanczos/ORTHOMIN
(Biconjugate Gradient), Lanczos/ORTHODIR, Lanczos/ORTHORES, and
nonsymmetric Chebyshev acceleration. The matrix is stored in the
Yale Sparse Matrix format.
\item[ITPACK 3B:] This package contains basically the same algorithms
as ITPACK 3A. It is written with an ELLPACK-style preprocessor
that allows the user more flexibility in designing the iterative
method. A high-level language is used to describe the matrix problem
and the desired iterative solution technique. The preprocessor
constructs a Fortran program with the appropriate calls to the ITPACK
3B library from the high-level description.
\item[NSPCG:] This package is similar to ITPACK 2C and 3A in that it
is Fortran-callable and contains acceleration schemes and basic
iterative methods intended for both symmetric and nonsymmetric
matrix problems. It has a large variety of basic iterative methods
and acceleration schemes, and the matrix can be represented in any
one of several sparse matrix formats plus a matrix-free format in
which the user supplies the matrix-vector routines. As with ITPACKV 2D,
it has been vectorized for vector computers such as the Cray Y-MP.
\end{description}
A limited number of copies of these packages will be made available
on request with the understanding that they are intended as research
tools and may undergo further development. The interested reader
should write to the address below for additional information.
\bigskip
\begin{tabular}{l}
Center for Numerical Analysis \\
RLM Bldg. 13.150 \\
University of Texas at Austin \\
Austin, Texas \ \ 78713-8510 \\
\\
Ph: (512) 471-1242
\end{tabular}
\bigskip
The Center for Numerical Analysis distributes these software
packages on a tape. The ITPACK tape contains several versions of
ITPACK: ITPACK 2C, double precision ITPACK 2C, ITPACKV 2D,
ITPACK 3A, ITPACK 3B, and NSPCG. The tape also
contains sample testing programs for each package, and User's Guides
for these packages are mailed along with the tape. Reports of
difficulties encountered in using these packages plus comments and
suggestions for improving them are welcome. An invoice of \$50 is
included to cover the costs of tape, reports, postage, and handling.
\bigskip
\centerline{\bf Disclaimer}
\bigskip
The University of Texas at Austin and the Center for Numerical Analysis
disclaim all warranties with regard to these software packages,
including all warranties of merchantability and fitness, and any stated
express warranties are in lieu of all obligations of liability on the
part of The University or the CNA for damages, including, but not limited
to, special, indirect, or consequential damages arising out of or in
connection with the use or performance of them. In no event will The
University or the CNA be liable for any direct, indirect, special,
incidental, or consequential damages arising in connection with use of
or inability to use these packages or their documentation. It should be
emphasized that these packages are preliminary, incomplete, and subject
to change.
\newpage
\centerline{\bf CNA Software on NETLIB}
\bigskip
The Fortran source codes for the ITPACK 2C, ITPACKV 2D, and NSPCG
software packages are available on {\bf netlib}. ITPACK 2C is available
in single and double precisions, but ITPACKV 2D and NSPCG are available
only in single precision. Testing routines and User's Guides (formated
using \LaTeX) for these packages are also available. To receive the
index of ITPACK files, send a computer mail message to
\bigskip
netlib@ornl.gov
\bigskip
\noindent
In the body of the message, type
\bigskip
send index from itpack
\bigskip
\noindent
You will receive an index of files that can be obtained in subsequent
messages using the syntax
\bigskip
send {\em file} from itpack
\bigskip
\noindent
where {\em file} is the name of a file from the index. The list of
available files is given below.
\bigskip
\begin{tabular}{ll}
dsrc2c.f & double precision ITPACK 2C source code \\
dtst2c.f & testing routine for double precision ITPACK 2C \\
index & an index to the itpack directory on netlib \\
info.tex & this file that you are reading (formatted with \LaTeX) \\
nspcg1.f & NSPCG source code, part 1 \\
nspcg2.f & NSPCG source code, part 2 \\
nspcg3.f & NSPCG source code, part 3 \\
nspcg4.f & NSPCG source code, part 4 \\
nspcg5.f & NSPCG source code, part 5 \\
quick.tex & Quick Reference Guide to NSPCG \\
src2c.f & single precision ITPACK 2C source code \\
srcv2d.f & ITPACKV 2D source code (vectorized for Cray Y-MP) \\
tst2c.f & testing routine for single precision ITPACK 2C \\
tstnsp1.f & testing program 1 for NSPCG \\
tstnsp2.f & testing program 2 for NSPCG \\
tstnsp3.f & testing program 3 for NSPCG \\
tstnsp4.f & testing program 4 for NSPCG \\
tstnsp5.f & testing program 5 for NSPCG \\
tstv2d.f & testing routine for ITPACKV 2D \\
user2c.tex & ITPACK 2C User's Guide (formated with \LaTeX) \\
usernsp.tex & NSPCG User's Guide (formated with \LaTeX) \\
userv2d.tex & ITPACKV 2D User's Guide (formated with \LaTeX)
\end{tabular}
\bigskip
Machine-dependent constants and the timing function may need to be
modified by the user when installing the packages for a particular
computer. They are located in routines DFAULT and TIMER in
all three packages.
Comments and suggestions are welcome and should be addressed to
the Center for Numerical Analysis.
\end{document}