NA Digest Saturday, September 7, 2002 Volume 02 : Issue 36

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

Submissions for NA Digest:

Mail to

Information via e-mail about NA-NET: Mail to


From: Gaston Gonnet <>
Date: Fri, 6 Sep 2002 17:51:02 +0200 (MET DST)
Subject: Testing Functions Accuracy

This is a relatively demanding test for the accuracy of mathematical system
functions like, sin, exp, lgamma, etc. It consists of a group of programs,
one for each function, which can be run on a target machine and will produce
a short report about the precision of the function in that machine.

Everything can be found under

There are two main goals for producing this software: first to understand how
reliable the numerical evaluations of the system you use are, and secondly to
provide some encouragement to the developers to improve some areas with

We have run these tests for 4 different architectures/operating systems. The
tests also reveal some properties of the hardware. There are lots of extremely
interesting and surprising results.

Each function is tested over various inputs, for which we have precomputed the
result to higher precision. These inputs are difficult, in various senses.
First, a list of known problematic values are tried. Secondly, values which
will evaluate to problematic values are also tried. Finally, ranges of values
are explored randomly, but for each value, nasty neighbours are computed. A
nasty value is one for which the exact result will be extremely close to 0.5
ulp. This is done with a cute application of the LLL lattice reduction
algorithm. By testing with these numbers we often find the weakest

Best wishes, Gaston Gonnet.


From: Vadim Olshevsky <>
Date: Tue, 3 Sep 2002 21:08:45 -0400
Subject: Change of Address for Vadim Olshevsky

Dear colleagues and friends,

I have joined the Department of Mathematics of
The University of Connecticut. My new address is:

Vadim Olshevsky
Department of Mathematics
University of Connecticut
196 Auditorium Road, U-9
Storrs, CT 06269




From: Ruediger Seydel <R.Seydel@Uni-Koeln.DE>
Date: Tue, 3 Sep 2002 13:10:41 +0200 (MET DST)
Subject: New Book, Tools for Computational Finance

A new book is available:

by Rudiger Seydel

This book provides a practical introduction to Computational Finance,
formulating methods and algorithms that can be implemented and used.
The topics cover tools for modeling, stochastic differential
equations, and the PDE context of the Black-Scholes world.
Applications are mainly to options, including exotic options.
Numerous figures and many examples illustrate the concepts.
An extensive appendix provides additional material for readers with
little background. The book is written with a minimum of formalism
and focussing on readability.

The book grew out of courses about this topic. In the experience of
the author, the topic of Computational Finance is highly attractive
to students. In this area one finds exciting applications in the
financial press and is tempted to apply the algorithms to the daily
financial world. The topic can be recommended to be added to any
NA syllabus. The full table of contents is available at

The book was published by Springer, Berlin 2002;
Universitext, 224 pages; ISBN 3-540-43609-X


From: Eilish Hathaway <>
Date: Thu, 5 Sep 2002 12:33:47 -0700
Subject: Semester at UCLA IPAM on Nanoscale Science

Program in Mathematics in Nanoscale Science and Engineering at the
Institute for Pure and Applied Mathematics (IPAM): There will be a
semester long program in Mathematics in Nanoscale Science and
Engineering from September 16 - December 13, 2002.
Details of the program, an application form and other information are at
The program will take place at the IPAM building on the campus of UCLA
in Los Angeles. The program is organized by Russel Caflisch (UCLA),
James Heath (UCLA), Mitchell Luskin (University of Minnesota), Antonio
Redondo (Los Alamos National Laboratory), and Peter Shor (AT&T Labs-Research).
Funding is available for graduate students, post-docs and others who wish to
attend the entire program. Online application form is on our website.
Funding is also available for shorter stays.

Mathematical and computational methods are expected to play a major role
in nanoscience and nanoengineering. Mathematics and computation can
provide effective theory and simulations for analysis and interpretation
of experimental results, model-based prediction of nanoscale phenomena,
and design and control of nanoscale systems. Computational methods, such
as density functional theory (DFT) and kinetic Monte Carlo (KMC), have
already had major success in nanoscience, and there are still many
opportunities for further involvement of mathematics and computation in
nanoscience. Exploiting these opportunities will require collaboration
between mathematical scientists and nanosystems researchers.


Tutorials: September 17-20, 2002

Workshop I Alternative Computing: September 30- October 4, 2002
Organizing Committee: Russel Caflisch (UCLA), James Heath (UCLA), Vwani
P. Roychowdhury (UCLA)

Workshop II Joint IPAM/MSRI Workshop on Quantum Computing: October
21-25, 2002. Organizing Committee: Peter Shor (AT&T), David Di Vincenzo

Workshop III Data Analysis and Imaging: November 4-8, 2002
Organizing Committee: Guillermo Sapiro, Chair (UMN), Richard Kiehl
(UMN), and Michael Vannier (Univ of Iowa)

Workshop IV Modeling and Simulation for Materials: November 18-22, 2002
Organizing Committee: Emily Carter (UCLA), Mark F. Gyure (HRL), Mitchell
Luskin (Minn), Tamar Schlick (Courant), and Weinan E (Princeton)

Full details of the program and its workshops can be found at Program specific
questions can be addressed to .=20


From: David F Griffiths <>
Date: Fri, 6 Sep 2002 17:24:50 +0100
Subject: Dundee NA Conference 2003

Tuesday 24 June - Friday 27 June, 2003


The special invited lecture in honour of A. R. Mitchell will be
presented by

Professor Stan Osher, University of California

The other Principal Speakers will include

Mario Arioli, Rutherford Appleton Laboratory, UK
Martin Buhmann, Justus-Liebig-Universitat Giessen, Germany
Bob Fourer, Northwestern University, USA
Gene Golub, Stanford University, USA
Jacek Gondzio, The University of Edinburgh, UK
David Gottlieb, Brown University, USA
Max Gunzburger, Iowa State University, USA
Des Higham, University of Strathclyde, UK
Martin Stynes, National University of Ireland
Philippe Toint, University of Namur, Belgium

Some further information is available at the conference web site:

Details of registration/accommodation fees as well as arrangements for
contributed talks will be posted on this page when they become
available in a few weeks time.

Conference Secretaries:
David Griffiths
Alistair Watson

Contact (email preferred)

Dr David F. Griffiths
Numerical Analysis Conference
Department of Mathematics
The University of Dundee
Dundee DD1 4HN
Scotland, UK

Telephone: +44(1382)344467/344471
FAX : +44(1382)345516


From: Bo Kagstrom <>
Date: Sat, 7 Sep 2002 13:33:51 +0200 (MEST)
Subject: Conference Celebrating Axel Ruhe's 60th birthday

Conference celebrating Axel Ruhe's 60th birthday, Oct 14, 2002

In celebration of Axel Ruhe's sixtieth birthday and his many contri-
butions to numerical analysis, there will be a one-day conference
at Nada/KTH, Stockholm, October 14, 2002. The theme of the conference
is numerical analysis with the emphasis on numerical linear algebra.

Axel Ruhe is a professor in numerical analysis at NADA/KTH. He took
his Ph D in Lund 1970 and he worked in Umea 1970-1983 and at Chalmers
in Goteborg 1983-2001. Axel's main research field is numerical linear
algebra, and in particular eigenvalue problems. He has also worked in

So far the following speakers, representing international and Swedish
colleagues and friends of Axel, have accepted the invitation to give a

Zhaojun Bai, University of California at Davis
Ake Bjorck, University of Linkoping
Katarina Blom, Chalmers
Kenneth Holmstrom, Malardalen University
Bo Kagstrom, Umea University
Joakim Moller, KTH
Lina von Sydow, Uppsala University
Henk A. van der Vorst, University of Utrecht

Organizing committee

Thomas Ericsson, Chalmers, G