NA Digest Sunday, January 8, 1995 Volume 95 : Issue 02

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

Submissions for NA Digest:

Mail to

Information about NA-NET:

Mail to


From: Gene Golub <golub@sccm.Stanford.EDU>
Date: Sun, 1 Jan 95 15:10:19 PST
Subject: Congratulations

Belated best wishes to Ake Bjorck and Mike Osborne on the occasion of their
60th birthday. Welcome to the club guys!



From: Iain Duff <>
Date: Fri, 6 Jan 95 16:38:00 GMT
Subject: John Rollett

John Rollett

I am sorry to relate the news from the Oxford University Computing Laboratory
that John Rollett died on January 3rd. Apparently he passed away peacefully
in his sleep whilst on holiday in Spain. I know that many of you will
remember the energy and enthusiasm of John at Oxford. At the moment, I
don't have any details of the contact address or funeral arrangements
but suggest that you contact Joan Himpson at
for further information.


From: John C. Nash <>
Date: Mon, 2 Jan 1995 15:47:58 -0500 (EST)
Subject: New Book, Scientific Computing with PCs

Announcing the publication of

Scientific Computing with PCs

by John C. Nash and Mary M. Nash

Copyright (C) 1994 J C and M M Nash

Available by FTP from the experimental server

There is a $10 licence fee for printing this book.

The table of contents, the preface and a sample chapter may be
downloaded and printed freely without violating the copyright.
This sample file is named SCPCDOC.PS.

The complete 200-page book is in four PostScript files SCPC-1.PS, ...
SCPC-4.PS. For MS-DOS users, there is a ZIP file that reduces the
size of the transmission considerably. The layout is designed to use
paper efficiently (a traditional layout took 300 pages).

The book is a condensation and overview of the extensive experience
of the authors with personal computers in scientific applications.
The intent is to address "what to" and "why to" calculate rather than
"how to".

Some of the examples -- not published elsewhere -- are likely to be
of special interest to the numerical analysis community. For
example, one chapter addresses the issue of comparing implementations
of numerical algorithms for efficiency. The results -- spanning
almost two decades of testing -- provide a unique and possibly
surprising panorama of the difficulties of predicting program
performance on even the simplest of codes. The example programs are
three variants of the well-known Cholesky decomposition.

Another chapter discusses the use of statistical graphics in
the comparison on program performance, this time using published
results for the performance of optimization codes.

Note: The files are .PS (PostScript) because of the graphics and
display needed for a book of this type. We have already had a
couple of requests for an ASCII form of the book, which we are
not prepared to offer at this time. Those without PostScript
printing may want to investigate the GhostScript program from
gnu. This is available from many software archives, including, where you may need to get the fonts from the
unix/gnu/ etc. subdirectories while the program for PCs comes
from the msdos section of the archive.

John C. Nash, Professor of Management, Faculty of Administration
136 Jean-Jacques Lussier Private, University of Ottawa, Ottawa
Ontario, K1N 6N5 Canada fax 613 564 6518


From: Najib Abboud <>
Date: Tue, 3 Jan 1995 18:02:34 -0500
Subject: Out-of-core Symmetric Banded Solver

I would appreciate any help in locating a public-domain
fortran out-of-core solver for a symmetric banded matrix.
I would be interested in a wavefront solver but I would like
to find out about alternatives if any.

I thank you in advance for any pointers, comments ...

Najib N. Abboud


From: Jos van Dorsselaer <>
Date: Thu, 5 Jan 1995 14:22:36 +0100
Subject: New Address for Jos van Dorsselaer

Dear colleagues,

I moved from Leiden University to Utrecht University. My new
address is:

Department of Mathematics
Utrecht University
P.O. Box 80010
3508 TA Utrecht
The Netherlands

New email-address:

Jos van Dorsselaer


From: Tony Garratt <>
Date: Thu, 05 Jan 1995 12:04:34 GMT
Subject: Positions at AspenTech in Cambridge, UK



Aspen Technology is the world leader in providing computer modelling and
simulation tools to the process industries. Our customers include most of the
world's largest chemical companies. Due to expansion at our Cambridge office we
currently have 2 positions to join our mathematics development team.

The team is responsible for the development of numerical methods needed for the
robust solution of a range of problem types in the process simulator SPEEDUP.
We require a well qualified numerical mathematician and an experienced
mathematical programmer.

The numerical mathematician requires experience in one or more of the following

Numerical solution of DAEs
Numerical solution of PDEs
Numerical linear and non-linear algebra
Statistics / Parameter Estimation / Data Reconciation
Dynamic optimisation / Optimal Control
Homotopy methods
Manipulation of State-Space Models

The mathematical programmer requires experience in one or more of the following

Code performance analysis
Experience in software development and object oriented design techniques.
Design and writing of robust code for complex numerical methods
Parallel computing

Skills and experience pertinent to both positions are:

Working on a range of computer platforms: DEC openVMS, UNIX (various), Windows
on machines ranging from PCs to CRAY supercomputers.

Strong programming skills: FORTRAN essential with C or C++ an advantage.

The ability to locate faults in numerical codes quickly and to resolve such
problems efficiently.

Good communication skills both written and verbal.

Ability to work within a team.

Up-to-date with the recently published material.

The ability to adapt and pick up new challenges quickly.

To apply, please send your CV to:

Ms Pauline Cheeseman,
AspenTech UK Ltd.,
Sheraton House,
Castle Park,
CB3 0AX.


From: SIAM <>
Date: Wed, 04 Jan 95 14:07:28 EST
Subject: Contents, SIAM Control and Optimization

SIAM Journal on Control and Optimization
MARCH 1995 Volume 33, Number 2

Jack Warga: In Appreciation

H^infinity Optimal Sensitivity for a Class of Infinite-Dimensional Systems
Hong Yang

Zeros of Spectral Factors, the Geometry of Splitting Subspaces, and the
Algebraic Riccati Inequality
Anders Lindquist, Gyorgy Michaletzky, and Giorgio Picci

A Globally Convergent Successive Approximation Method for Severely
Nonsmooth Equations
Liqun Qi and Xiaojun Chen

Optimal Supervisory Control of Discrete Event Dynamical Systems
Ratnesh Kumar and Vijay K. Garg

Uniform Stabilization of a Hybrid System of Elasticity
Bopeng Rao

System Equivalence for Periodic Models and Systems
Osvaldo M. Grasselli, Sauro Longhi, and Antonio Tornambe

Supervisory Control of Nondeterministic Systems with Driven Events via
Prioritized Synchronization and Trajectory Models
Mark A. Shayman and Ratnesh Kumar

Risk-Sensitive Production Planning of Stochastic Manufacturing Systems:
A Singular Perturbation Approach
Qing Zhang

Multilevel Hierarchical Decision Making in Stochastic Marketing-Production
S. P. Sethi and Qing Zhang

On Bang-Bang Constrained Solutions of a Control System
Raphael Cerf and Carlo Mariconda

Dealing with Integral State Constraints in Boundary Control Problems of
Quasilinear Elliptic Equations
Eduardo Casas and Luis A. Fernandez

The Stochastic Maximum Principle for Linear, Convex Optimal Control with
Random Coefficients
Abel Cadenillas and Ioannis Karatzas

Persistency of Excitation in Identification Using Radial Basis Function
A. J. Kurdila, Francis J. Narcowich, and Joseph D. Ward

Maximizing Robustness in Nonlinear Illposed Inverse Problems
Kazufumi Ito and Karl Kunisch

Corrigendum: Lagrange Multipliers in Stochastic Programming
Sjur Didrik Flam


End of NA Digest