NA Digest Monday, July 5, 2010 Volume 10 : Issue 27

Today's Editor:
Tamara G. Kolda
Sandia National Labs

Submissions for NA Digest:

Mail to

Information via email about NA-NET:

Mail to


From: Joseph Grcar <>
Date: Sun, 4 Jul 2010 15:50:33 -0700
Subject: George Forsythe photo

Dear NA Digest,

I am looking for the source of a photograph of George Forsythe wearing
a business suit and leaning against a tape drive in an old fashioned
computer center with an asbestos tile ceiling. Does anyone know the

Joseph Grcar


From: Joseph Grcar <>
Date: Sat, 3 Jul 2010 16:36:16 -0700
Subject: Report on condition number definitions

Dear NA Digest,

I want to thank everyone who answered my question about condition

Some people sent definitions, while others cited the following 12
sources: Trefethen and Bau (5 citations), Wikipedia (3), Dahlquist
and Björck (2), Demmel (2), Deuflhard and Hohmann (2), Atkinson (1),
Chaitin-Chatelin and Fraysse (1), Opfer (1), Quarteroni, Sacco and
Saleri (1), Stoer and Bullisrsch (1), and papers by Rice (1) and
Wozniakowski (1). I personally looked at 8 sources: T&B, W, D&B, D&H,
CC&F, QS&S, S&B, and R.

I was surprised to find two definitions. Both deal with a vector-
vector function f at a point x. The condition number is:

1) the lim sup at x of a ratio: the norm of changes to the function
divided by the norm of changes to the argument.

2) the minimum of coefficients that appear in perturbation bounds of
the form: norm of change to the function at x bounded by a coefficient
times norm of change to x plus an error term of size Landau's little o.

I will not mention the sometimes complicated terminology in the
literature other than to say I have ignored scalings that might be
used to make the condition numbers "normwise relative." It seems to
me that the norms can include the scalings so the statements above
encompass the relative case.

The first definition is usually attributed to Rice (1966) who has an
expression that is equivalent to the definition of the lim sup. The
second definition might as well be attributed to Wilkinson (1973) who
said he would refer to the coefficient in an error bound as a
condition number; it is of course natural to ask for the smallest

I could imagine an epsilon-and-delta proof that these two definitions
are the same except the second definition makes me uneasy by stating
"min" rather than "inf". The textbooks I examined that use either
definition further say the condition number of a continuously
differentiable function is the induced norm of the Jacobian matrix.

These definitions raise the following questions:

a) are 1 and 2 equivalent perhaps with some minor modification?

b) what is the purpose of this level of abstraction? That is, for
what non-differentiable functions are condition numbers need in
scientific or engineering problems?

There are of course other cases to be considered such as componentwise
condition numbers and problems with indeterminate solutions.

Joseph Grcar


From: Garry Tee <>
Date: Mon, 28 Jun 2010 18:14:13 +1200
Subject: Response to " What is the definition of "condition number"?"

Joseph Grcar asked "What is the general definition of the
"condition number" of a numerical analysis problem?" (NA Digest Vol.
10: Issue 26).
One of the best discussions of condition numbers for
numerical analysis problems is that given by J. H. Wilkinson, in
"Rounding Errors in Algebraic Processes", National Physical
Laboratory Notes on Applied Science No.32, HMSO, London, 1963, on
pages 29, 33, 91-94 and 135-136. He explained that "We have avoided
framing a precise definition of condition numbers so that we may use
the term freely" (p.29). Also, Wilkinson discussed the condition of a
polynomial with respect to zeros (p.38), and the condition of a
matrix with respect to computing eigenvalues (pp.136-138). He
remarked that "The term condition number seems first to have been
used by Turing in his paper on rounding errors in matrix processes
[in 1948], though the term 'ill-condition' had been in common use
among numerical analysts for some considerable time before this" (p.
33). Wilkinson gave many detailed examples in his major treatise on
"The Algebraic Eigenvalue Problem", Oxford University Press, 1965.
A noteworthy early treatment of conditioning was given by Isaac
Newton's young friend Roger Cotes (1682-1716) in his short tract
"Aestimatio Errorum in Mixta Mathesi", published in 1722. Cotes gave
complete perturbation analyses (first-order) for plane triangles and
for spherical triangles, and he explained how to conduct the
measurement of astronomical angles in such a way as to minimize the
uncertainty in the computed result. [Ronald Gowing, "Roger Cotes -
Natural Philosopher", Cambridge University Press, 1983, pp.91-109.
Reviewed in Math. Rev. 87b:01033.]
Garry J. Tee,
Department of Mathematics, University of


From: Edwin Vollebregt <>
Date: Mon, 28 Jun 2010 14:00:43 +0200
Subject: Data-assimilation toolbox OpenDA 1.0 available

On May 10, the first official version of the free data-assimilation toolbox
OpenDA has been announced. A complete version for both Windows and Linux,
including the source code, is now available through the web site
and will soon be available on

OpenDA is a collection of building blocks and tools that allow rapid
implementation of data-assimilation for arbitrary (large scale) numerical
models. It includes various approximate Kalman filters and parameter
methods. Software components are available to couple the methods to your
numerical model. Language bindings exist for Fortran, C/C++ and Java.

Tools are available for using models with data-assimilation. These
include a
workbench that allows you to select any of the available data-assimilation
methods and configure its parameters.

The OpenDA software supports high performance computing. The
data-assimilation methods have been parallellized using MPI. The design of
OpenDA allows it to handle models that have themselves been parallelized
with MPI or with OpenMP.

OpenDA has been developed in a joint effort by Delft University of
the water management research institute Deltares and the scientific
engineering company VORtech. It has been applied in various practical
applications in fields like water management and atmospheric chemistry. A
coupling has been made to the open source wave model SWAN, the flood early
warning system FEWS, the open source atmospheric chemistry model
Chimere, as
well as several closed source applications.

More information can be found at or directly through the
address The OpenDA association is interested in
projects that
will help to spread the use of OpenDA and to develop it further. Please
communicate any initiatives to

Mark Roest, Arnold Heemink, Martin Verlaan and other contributors.


From: Joseph Traub <>
Date: Mon, 28 Jun 2010 09:30:19 -0400 (EDT)
Subject: 2009 Best Paper Award--Journal of Complexity

Frank Aurzada, Steffen Dereich, Michael Scheutzow, and Christian Vormoor

Win the 2009 Best Paper Award

The Award Committee -- Stephan Dahlke, Universitaet Marburg, Germany and
Dick, University of New South Wales, Australia - determined that the
paper exhibited exceptional merit and therefore awarded the prize to:

Frank Aurzada, Steffen Dereich, Michael Scheutzow, and Christian Vormoor
their paper "High resolution quantization and entropy coding of jump
processes", which appeared in April, 2009. Vol. 25, pp 163-188.

The $3000 prize will be divided between the authors. Each author will also
receive a plaque. The award will be presented at a suitable location.


From: Bob Palais <>
Date: Mon, 5 Jul 2010 16:27:07 -0400
Subject: New book, ODE, Mechanics, and Computation

Differential Equations, Mechanics, and Computation, by Richard S. Palais
and Robert A. Palais. published by the AMS, Student Mathematical Library
2009; 313 pp; softcover Volume: 51 ISBN-10: 0-8218-2138-5 ISBN-13:
978-0-8218-2138-1 List Price: US$51 Member Price: US$40.80

More information about the book is available in its Web Companion:

that includes excerpts of each section.

In modern applied mathematics, theory and rigorously analyzed
computation go
hand-in-hand. This calls for a text that discusses both in detail, and
we have
undertaken to provide one. The discussion of numerical methods strives to be
visual and comparative, based on carefully chosen examples of prototypical
methods and model problems.

A more detailed description of the approach to numerical methods, with a
to the excerpt of the first 38 pages of the chapter on design and
analysis of
computational methods, as well as several sample figures may be found at:

This book also provides a conceptual introduction to the theory of ordinary
differential equations, concentrating on the initial value problem for
equations of evolution and with applications to the calculus of
variations and
classical mechanics, along with a discussion of chaos theory and ecological
models. While the book would be suitable as a textbook for an
undergraduate or
elementary graduate course in ordinary differential equations, the authors
have designed the text also to be useful for motivated students wishing to
learn the material on their own or desiring to supplement an ODE textbook
being used in a course they are taking with a text offering a more
approach to the subject.

Readers of NA Digest may also be interested in a page linked from the web
companion that discusses an inconsistency in the literature in the
of the region of absolute stability of a numerical method . Examples from
eight well-known books, half of which give definitions requiring decay of
solutions, and half of which require only boundedness of solutions may be
found here:

For more information, please visit


From: Larry Nazareth <>
Date: Wed, 30 Jun 2010 11:59:06 -0700 (PDT)
Subject: New book, E-Book on Algorithmic Science and Engineering

I've recently published the following research monograph in
the form of an e-book:

J.L. Nazareth (2010), Algorithmic Science and Engineering within
Scientific Computing: Foundations and Organization, 128 pgs.,

which is available free of charge from the above web address. An
abbreviated table of contents is as follows:

Preface............................................... v
Part I: Foundational Models and Paradigms
Chapter 1: Symbol-Based Computing......................1
Chapter 2: Magnitude-Based Computing..................21
Chapter 3: Complex Behavior of Simple Programs........41
Chapter 4: Essentialism versus Population-Thinking....51
Part II: Context and Organization
Chapter 5: A Visual Icon for Sci-En-Math..............69
Chapter 6: The Emerging AS&E Discipline...............75
Chapter 7: Illustrations of AS&E......................87

Comments/corrections/feedback on this material sent by e-mail
to or would
be welcome.


From: "Chen Sagiv" <>
Date: Mon, 28 Jun 2010 08:09:06 +0200
Subject: Image Processing with CUDA Course, San Francisco area, Sep 2010

SagivTech plans to offer a 3-days course that deals with Image Processing
with CUDA in the USA this September.

This is an advanced course that is intended for experienced CUDA developers
who look for optimization methods for image processing applications
implemented on Nvidia$B!l(Bs GPUs.

The course will take place in the San Francisco area, between September 27
and September 29, just following the GTC. Course hours are 9-17.

The course focuses on the practical issues involved in the implementation of
image processing algorithms on the GPU and provides several hands on
sessions. Among the topics dealt with in the course are: Bottlenecks
assessment, GPU Memory Access Patterns, Asynchronously and stream
processing, Multiple GPUs, Linear and Non-Linear Filtering, Image
Registration, Mathematical Morphology, Gray level Co-Occurrence Matrices and
Optical Flow. For a detailed syllabus see:

Registration is now open, and will be closed on August 15th. Number of
participants is limited. The final confirmation on the course will be given
on September 1st.

For registration email:


From: Peter Arbenz <>
Date: Mon, 5 Jul 2010 13:38:34 +0200
Subject: Workshop on High-Performance Computing at ETH Zurich, Sept 2010

39th SPEEDUP Workshop on High Performance Computing,
September 6/7, 2010, at ETH Zurich

The intention of this workshop is to present and discuss the
state-of-the-art in high-performance and parallel scientific computing.
Presentations will focus on algorithms, applications, and software
issues related to high-performance parallel computing. The focus of the
workshop on Monday Sept 6 will be on software environments for large
scale simulations and on the issues of fault tolerance in massively
parallel systems.

The scientific program of September 6 consists of seven 45-minute talks
and a poster session. Invited Speakers are Omar Gattas (University of
Texas, Austin), Jedediah Brown (ETH Zurich), Michael Heroux (Sandia
National Laboratories, Albuquerque), Rolf Krause (USI Lugano), Romain
Teyssier (University of Zurich), Christian Engelmann (Oak Ridge National
Laboratory), Georg Hager (University of Erlangen-Nuernberg).

On September 7 we organize a tutorial on hybrid MPI/OpenMP computing.
It will be taught by Timothy Stitt and Neil Stringfellow (both from the
National Supercomputing Centre, CSCS).

Details can be found at

Organizing Committee: Andreas Adelmann (PSI Villigen), Peter Arbenz
(ETH Zurich), Olaf Schenk (University of Basel), Vittoria Rezzonico
(EPF Lausanne), Ales Janka (University of Fribourg).


From: Margreet Nool <>
Date: Tue, 29 Jun 2010 16:37:31 +0200
Subject: Woudschoten Conf on Scientific Computing, the Netherlands, Oct

On 6-8 October 2010 the thirty-fifth Woudschoten Conference will be held
at the Woudschoten Conference Centre, Zeist, The Netherlands.
The web form for electronic registration has been made available.

Themes of the conference are:
- Parallel numerical linear algebra
Invited talks by:
o Patrick Amestoy, Université de Toulouse, Toulouse, France
o Peter Arbenz, ETH Zürich, Switzerland
o Alex Pothen, Purdue University, West Lafayette, USA
- Immersed boundary methods and Cartesian grids
Invited talks by:
o Gianluca Iaccarino, Stanford University, California, USA
o Petros Koumoutsakos, ETH Zürich, Switzerland
o Rajat Mittal, Johns Hopkins University, Baltimore, USA

Organising committee 2010: Prof. Arthur Veldman (chair), Prof. Rob
Bisseling, Prof. Kees Vuik, Prof. Barry Koren, Drs. Margreet Nool

Information: (English) (Dutch)
Registration: (English) (Dutch)

All participants are invited to present a poster about their current
See the website and registration form for details.


From: Tatjana Stykel <>
Date: Sat, 3 Jul 2010 18:33:25 -0400
Subject: Workshop on Model Reduction, TU Berlin, Dec 2010

Workshop on Model Reduction for Complex Dynamical Systems
December 2-3, 2010
TU Berlin, Germany

This workshop of the BMBF research network SyreNe (
aims to bring together researchers and users of model order reduction
techniques with special emphasis on applications in micro- and
nanoelectronics. Contributions from other areas such as computational
electromagnetics, mechanical systems, computational fluid dynamics and
related disciplines are welcome.

Invited speakers:
* Michel S. Nakhla (Carleton University, Ottawa, Canada)
* Joel R. Phillips (Cadence Research Laboratories, Berkeley, USA)
* Timo Reis (TU Hamburg-Harburg, Germany)

Important dates:
Abstract submission: July 15, 2010 (extended)
Please submit an abstract of your talk (up to 1 page) using
the LaTeX template.
Acceptance notification: September 1, 2010
Registration: November 1, 2010

For further information, please visit


From: Giovanni Samaey <>
Date: Mon, 28 Jun 2010 15:47:36 -0400
Subject: Workshop multiscale simulation for materials, Belgium, Jan 2011

A workshop on 'Multiscale simulation of heterogeneous materials and
coupling of thermodynamic models' will be held in Leuven, Belgium, on
January 12-14, 2011.


Since the macroscopic properties of a large class of materials depend
on the heterogeneities on micro- and mesoscopic scales, appropriate
mathematical models are needed to adequately describe the evolution of
the spatial structure and composition variations at each of these
scales. This has lead to a number of modeling approaches that describe
a material’s behavior on different scales ranging from the (sub)atomic
to the continuum level.

In this workshop, the focus will be on two closely connected themes:
computational multiscale methodology and the coupling of thermodynamic

We refer to the website for a list of invited speakers. Contributed
presentations of 20 minutes are welcomed on topics related to the
scope of the workshop. We also encourage each participant to bring a
poster on his/her work, which will be displayed for the whole duration
of the workshop. Please submit your contribution before August 31,
2010 via the website.

This workshop is sponsored by the Scientific Research Networks
'Advanced numerical methods for mathematical models', 'Surface
modification of materials', 'Computational modeling of materials', and
'Structural and chemical characterization of materials at the micro-
and nanoscale', the European COST MP0602 action, as well as by Res
Metallica (consisting of OCAS, Bekaert and Umicore).


From: "Pinar, Ali" <>
Date: Fri, 2 Jul 2010 15:36:16 -0600
Subject: SIAM Workshop, Combinatorial Scientific Computing, Germany, May

Fifth SIAM Workshop on Combinatorial Scientific Computing
May 19-21, 2011, Darmstadt, Germany

With the continuing advances in high-performance computing
(HPC) the role of computational science and engineering (CSE)
has gained significant importance over the last decades.
At the same time scientific simulation faces a number of
challenges. Many of those are combinatorial in nature and
unified by a common set of abstractions, data structures, and
algorithms based on combinatorics, graphs, and hypergraphs.
CSC11 provides a forum for researchers interested in the
interaction of combinatorial mathematics and algorithms
with CSE. The workshop will follow and in part overlap with
the 2011 SIAM Conferenceon Optimization (OP11,

We invite 2-page extended abstracts for 25-minute oral
as well as for poster presentations to be submitted via

Confirmed invited presentations:
Thomas F. Coleman (University of Waterloo)
(joint plenary presentation with OP11)
Burkhard Monien (Paderborn University)
Trond Steihaug (Bergen University)


From: Petter Bjorstad <>
Date: Sat, 3 Jul 2010 07:26:18 +0200
Subject: 1 PostDoc and 2 PhD positions at Univ Bergen, Norway

A postdoctoral position (2-4 years) and 2 PhD positions
(4 years) are available for immediate considerations.

The Postdoc gross salary is approx. 450.000 NOK/year
The PhD gross salary is approx. 350.000 NOK/year

Postdoc candidates should apply via:

PhD candidtaes should apply via:

The research shall be carried out at the

University of Bergen, Norway ( )
Department of Informatics

For further details contact or


From: Ke Chen <>
Date: Mon, 28 Jun 2010 04:23:49 -0400
Subject: 3 PhD positions from University of Liverpool, UK

The Department of Mathematical Sciences, The University of Liverpool, UK,
3 PhD positions. Please forward to any suitable

(1) Graduate TA Posts 1 and 2:
These are funded by the University of Liverpool in partnership with
Xi'an Jiaotong-Liverpool
University (XJTLU). Candidates can choose any available projects within
Mathematical Sciences.
Each post is for 4 years, starting from 1 Sep 2010, at the research
council level of £13,590 pa.
There are no residence requirements. The GTAs’ deadline is: 6 July 2010.
(2) Post 3: EPSRC INDUSTRIAL CASE studentship
This is funded by the EPSRC following a successful competition for the
general pool, for the project
“Blind Deblurring Techniques for Retinal Imaging”,
in collaboration with the St Pauls Eye Unit of the Royal Liverpool and
Broadgreen University Hospital,
a leading eye hospital that treats around 90000 patients per year from
all over UK and further afield.
Essentially involving variational imaging modelling and advanced
numerical methods, the project will be
supervised by Prof Ke Chen (Math Sciences) and Prof Simon Harding (St
Paul’s unit).
The usual UK EPSRC rules and conditions apply.


From: Chi-Wang Shu <>
Date: Mon, 5 Jul 2010 02:34:39 -0400 (EDT)
Subject: Contents, Journal of Scientific Computing 44(2) Aug 2010

Journal of Scientific Computing

Volume 44, Number 2, August 2010

Boussinesq Systems of Bona-Smith Type on Plane
Domains: Theory and Numerical Analysis
V.A. Dougalis, D.E. Mitsotakis and J.-C. Saut,

On the Quadrature and Weak Form Choices in
Collocation Type Discontinuous Galerkin Spectral
Element Methods
David A. Kopriva and Gregor Gassner, pp.136-155.

A Unified Analysis of Several Mixed Methods for
Elasticity with Weak Stress Symmetry
J. Guzman, pp.156-169.

Modeling the Early Ionization of Dielectrics by
Ultrashort Laser Pulses
Antoine Bourgeade, Candice Mezel and Olivier
Saut, pp.170-190.

Two-step Runge-Kutta Methods with Quadratic
Stability Functions
D. Conte, R. D'Ambrosio and Z. Jackiewicz,


From: Emma Avery <>
Date: Fri, 2 Jul 2010 16:48:51 +0100
Subject: Contents, Inverse Problems, volume 26, issue 7, July 2010


Volume 26, Issue 7, July 2010

Article numbers: 070201--071016

Individual articles are free for 30 days following their publication on the
web. This issue is available at:


Introduction to the special issue on electromagnetic inverse problems:
emerging methods and novel applications
O Dorn and D Lesselier (Guest Editors)


Resolution analysis for imaging with noise
Josselin Garnier and George Papanicolaou

Conformal mapping and impedance tomography
Houssem Haddar and Rainer Kress

Electromagnetic passive localization and tracking of moving targets in a
WSN-infrastructured environment
F Viani, P Rocca, M Benedetti, G Oliveri and A Massa

The inverse electromagnetic scattering problem for anisotropic media
Fioralba Cakoni, David Colton, Peter Monk and Jiguang Sun

Increasing stability of the continuation for the Maxwell system
Deepak Aralumallige Subbarayappa and Victor Isakov

Adjoint-based sampling methods for electromagnetic scattering
H Egger, M Hanke, C Schneider, J Sch\"oberl and S Zaglmayr

Subspace-based optimization method for inverse scattering problems with an
inhomogeneous background medium
Xudong Chen

On the imaging of thin dielectric inclusions buried within a half-space
Won-Kwang Park

Contrast-enhanced microwave imaging of breast tumors: a computational study
using 3D realistic numerical phantoms
J D Shea, P Kosmas, B D Van Veen and S C Hagness

Reduced-order estimation of nonstationary flows with electrical impedance
A Lipponen, A Sepp\"anen and J P Kaipio

Acceleration techniques for regularized Newton methods applied to
electromagnetic inverse medium scattering problems
Thorsten Hohage and Stefan Langer

Kriging-based generation of optimal databases as forward and inverse
surrogate models
S Bilicz, M Lambert and Sz Gyim\'othy

Application of a two-and-a-half dimensional model-based algorithm to
crosswell electromagnetic data inversion
Maokun Li, Aria Abubakar and Tarek M Habashy

Numerical solution of an inverse medium scattering problem with a
stochastic source
Gang Bao, Shui-Nee Chow, Peijun Li and Haomin Zhou

A study on orthogonality sampling
Roland Potthast

Recent developments of a monotonicity imaging method for magnetic induction
tomography in the small skin-depth regime
A Tamburrino, S Ventre and G Rubinacci

End of NA Digest