NA Digest Sunday, August 20, 1995 Volume 95 : Issue 33

Today's Editor:
Cleve Moler
The MathWorks, Inc.

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to

URL for the World Wide Web: -------------------------------------------------------

From: Chris Bischof <>
Date: Mon, 14 Aug 1995 14:56:26 -0500
Subject: Automatic Differentiation Tools

Automatic Differentiation (AD) is a technique for augmenting computer programs
with derivative computations.

A compilation of available automatic differentiation software
available at

It provides short highlights of Fortran 77, Fortran 90, ANSI-C and C++
AD tools, as well as of system integrating AD such as MAPLE or AMPL, and
AD support libraries.

For each contribution, a brief description of its functionality, highlights,
as well as pointers to developers and informational material are provided.

Chris Bischof
Mathematics and Computer Science Division
Argonne National Laboratory


From: ADIFOR Project <>
Date: Wed, 16 Aug 1995 10:39:50 -0500
Subject: ADIFOR 2.0 is Now Available

The ADIFOR 2.0 automatic differentiation system is now available.
We are proud to mention that ADIFOR 2.0 recently won the 1995 Wilkinson
Prize for Numerical Software at the Int. Conf. on Industrial and Applied
Mathematics (ICIAM 95) in Hamburg, Germany.

To find out more about ADIFOR, in particular how to obtain the ADIFOR 2.0
system or documentation and reports pertaining to it, visit the ADIFOR web
sites at
or (note the tilde!)

ADIFOR is not in the public domain, but it is available for non-commercial
use or evaluation for commercial purposes free of charge.

The ADIFOR 2.0 system is a tool for the automatic differentiation of
Fortran 77 programs. Given a Fortran 77 source code and a user's
specification of dependent and independent variables, ADIFOR will generate
an augmented derivative code that computes the partial derivatives of all
of the specified dependent variables with respect to all of the specified
independent variables in addition to the original result.

The ADIFOR 2.0 system consists of the ADIFOR 2.0 preprocessor, the
ADIntrinsics 1.0 intrinsic handler system, and the SparsLinC 1.0
(Sparse Linear Combination) library. It offers full Fortran 77
support, flexible handling of Fortran intrinsics, and transparent
sparsity support.

Christian Bischof Alan Carle
Math & Comp. Sc. Div. Center for Research on Parallel Comp.
Argonne National Lab. Rice University


From: Hans Mittelmann <>
Date: Thu, 17 Aug 1995 09:22:46 -0700
Subject: Decision Tree for Optimization Software

Decision Tree for Optimization Software

An HTML-version of the decision tree and guide recently announced
in NA-digest 95-31 and the sci.math.num-analysis and sci.op-research
newsgroups is now accessible at

Hans D. Mittelmann
Arizona State University Phone: (602) 965-6595
Department of Mathematics Fax: (602) 965-0461
Tempe, AZ 85287-1804 email:


From: Bas Braams <>
Date: Fri, 18 Aug 1995 11:12:55 GMT
Subject: Matrix Re-ordering for ILU-CG

Matrix re-ordering strategies are important in connection with direct
solution methods for sparse equations; e.g. /1/. What about matrix
re-ordering in connection with ILU-based iterative methods? I would be
interested to know of studies comparing the effectiveness of various
matrix orderings for the (non-symmetric) ILU-CG solution of sparse
systems. I am particularly interested in systems that come from the
discretization of an advection-diffusion equation.
/1/ I. S. Duff, A. M. Erisman and J. K. Reid: Direct Methods for Sparse
Matrices, Clarendon Press, Oxford, 1986.
-- Bas Braams (


From: Roger Ghanem <>
Date: Mon, 14 Aug 1995 17:26:51 -0400
Subject: Change of Address for Roger Ghanem

New address is:

Roger Ghanem,
201 Latrobe,
The Johns Hopkins University,
3400 N. Charles St.,
Baltimore, MD 21218.
Tel: (410) 516-7647
Fax: (410) 516-7473


From: Panos Pardalos <>
Date: Mon, 14 Aug 1995 11:41:22 -0400
Subject: New Books on Nonconvex Optimization and its Applications

Nonconvex Optimization and its Applications
Kluwer Academic Publishers

Title: Advances in Optimization and Approximation/ edited by
D.-Z. Du and J. Sun
Series Name: Nonconvex optimization and its applications ; vol. 1
Publication Info: Dordrecht ; Boston : Kluwer Academic Publishers, 1994.
ISBN: 0-7923-2785-3

Title: Handbook of Global Optimization / edited by Reiner Horst
and Panos M. Pardalos.
Series Name: Nonconvex optimization and its applications ; vol. 2
Publication Info: Dordrecht ; Boston : Kluwer Academic Publishers, 1995.
ISBN: 0-7923-3120-6

Title: Introduction to global optimization / by Reiner Horst,
Panos M. Pardalos, and Nguyen V. Thoai.
Series Name: Nonconvex optimization and its applications ; v. 3
Publication Info: Dordrecht ; Boston : Kluwer Academic Publishers, 1995.
ISBN: 0-7923-3556-2 (also available in paperback)

Title: Minimax and applications / edited by Ding-Zhu Du and
Panos M. Pardalos.
Series Name: Nonconvex optimization and its applications ; v. 4
Publication Info: Boston : Kluwer Academic Publishers, 1995.
ISBN: 0-7923-3615-1

How to order the books:
Fax Kluwer at 617-871-6528.
Phone Kluwer at 617-871-6600.


From: Dugald Duncan <>
Date: Fri, 18 Aug 1995 17:54:27 +0100
Subject: Scottish Computational Maths Symposium

10am-5pm Wednesday 20th September
Hume Tower, University of Edinburgh, Scotland.

organised by D.B. Duncan (Heriot-Watt University) and D.M. Sloan
(University of Strathclyde) with the aim of bringing together
mathematicians and others who develop and/or use computer algorithms to
solve mathematical problems. The meetings are open to everyone
interested and this one is supported by a grant from the London
Mathematical Society.


* M. Berzins (University of Leeds)
"Adaptive Solution Methods for Combustion and Atmospheric
Dispersion Problems using the Method of Lines"

* D.F. Griffiths (University of Dundee)
To be announced.

* J.C. Mason (University of Huddersfield)
"Approximation with Neural Networks"

* A. Ramage (University of Strathclyde)
"Iterative Methods for a Class of Indefinite Linear Systems"

* J.C. Eilbeck/D.B. Duncan (Heriot-Watt University)
"Mathematics on the Internet".


Please register as soon as possible (deadline September 8th) by sending
a cheque for 20 pounds (15 for students) payable to Heriot-Watt
University. Include your address so that we can contact you. The fee
includes tea, coffee and lunch.

To register or to obtain further information please contact:

SCMS95, Department of Mathematics,
Heriot-Watt University, Edinburgh, EH14 4AS.

WWW Information:


From: Yousef Saad <>
Date: Wed, 16 Aug 1995 10:30:49 -0500 (CDT)
Subject: Postdoctoral Position at University of Minnesota

University of Minnesota
Department of Computer Science

Position available: Postdoctoral Associate

A Postdoctoral Associate position is available to work on a project
related to parallel sparse matrix computations. Applicants must have
a strong background in numerical linear algebra. Good knowledge of
partial differential equations, domain decomposition techniques,
sparse matrix techniques, and iterative methods, is desirable. A
Ph.D. in Computer Science or related field is required as well as an
extensive experience in parallel programming.

The project, supported by ARPA, involves the development of a library
for sparse matrix computations called P_SPARSLIB. A first version of
the library is currently available and part of the work consists of
adding functionality to the library, as well as improving performance
on specific platforms.

The university of Minnesota provides a state-of-the-art environment
for supercomputing and parallel computing. In addition to hardware
available to U of M researchers at the Minnesota Supercomputer Center,
the department of Computer Science also houses a cluster of SGI
workstations and a cluster of IBM workstations and is soon to receive
an IBM SP2.

Interested applicants should send a resume and cover letter *as soon
as possible* to:

Professor: Yousef Saad
Dept. of Computer Science
University of Minnesota
4-192 EE/CSci Building
200 Union Street S.E.
Minneapolis, MN 55455


End of NA Digest