NA Digest Friday, July 13, 2007 Volume 07 : Issue 29

Today's Editor:
Tamara G. Kolda
Sandia National Labs
tgkolda@sandia.gov

Submissions for NA Digest:

Mail to na.digest@na-net.ornl.gov

Information via email about NA-NET:

Mail to na.help@na-net.ornl.gov

-------------------------------------------------------

From: "Dianne O'Leary" <oleary@cs.umd.edu>
Date: Mon, 9 Jul 2007 11:56:17 -0400 (EDT)
Subject: Petaflop challenge

Do you want to win a FREE trip to SC07?

Computing in Science and Engineering magazine (http://computer.org/cise)
is sponsoring a contest based on a challenge from Francis Sullivan
(IDA Center for Computing Sciences):

"A lot of my working life has been spent arguing for more compute power
and explaining the wonderful things that will happen when a more powerful
machine is delivered. So far, nobody has called my bluff by saying, 'Okay,
you can have all the computing you want. Now, what are you going to do
with it?'

What would YOU compute?

Let's fantasize. Suppose you had available, at your desk, a petaflop
machine. Assume that power and cooling were somehow taken care of and that
you could get the full power of the machine by writing programs in a
flexible and expressive language, no heroics necessary."

Email your answer in the form of a 1,000-word essay to Jenny Stout
(jstout@computer.org), with the subject line "petaflop challenge," by 10
August 2007. The winner will receive a year's subscription to Computing in
Science & Engineering magazine (http://computer.org/cise) and will see
their entry published in the magazine's November/December 2007 issue.

SC07 will also cover the winner's travel costs to attend its conference in
Reno, Nevada, USA. SC07 is the international conference for
high-performance computing, networking, storage and analysis.

Visit http://sc07.supercomputing.org for conference details.

-------------------------------------------------------

From: "Rafael Camilo Nunez" <rafael.nunez@accelogic.com>
Date: Thu, 12 Jul 2007 12:05:00 -0400
Subject: Looking for large-scale banded matrices to test fast FPGA solver

I know matrix collections is a recurrent topic in the Digest, so I'm
writing hoping that someone can point us in the right direction.

We are looking for one or several large-scale banded matrices used in
practice (preferably from CFD or aerospace, but matrices from other fields
would be useful too) to test Accelogic's latest narrowband linear equation
solver. Bands from 3 to 64, and matrix sizes from a few thousands to
hundreds of millions.

Any help would be greatly appreciated.
We could provide benchmark data (both speed and solution values) for your
matrices in return. When running on a single FPGA chip, this solver
recently demonstrated the performance of ScaLAPACK on a 240-node CPU
cluster (the baseline is a cluster of 240 Woodcrest CPUs). The technology
is the result of a multi-agency effort to leverage FPGA performance for
supercomputing applications, via innovative reconfigurable computing
algorithms. It is led by Accelogic, with sponsorship from NASA, DOE, and
the Air Force, among others. I'd be more than glad to provide additional
information to those interested.

Thanks,

Rafael Nunez
Chief Hardware Architect
Accelogic LLC

-------------------------------------------------------

From: "Randy LeVeque" <rjleveque@gmail.com>
Date: Fri, 6 Jul 2007 17:13:21 -0700
Subject: New book on Finite Difference Methods

New book from SIAM:
Finite Difference Methods for Ordinary and Partial Differential
Equations: Steady-State and Time-Dependent Problems
by Randall J. LeVeque
This book is suitable as a text for a first year graduate level course.
For more information please see
http://www.ec-securehost.com/SIAM/OT98.html

Copies will be available for inspection or purchase at the SIAM booth
at ICIAM in Zurich next week.

July 2007 / xiv + 339 pages / Softcover / ISBN 978-0-898716-29-0
List Price $63.00 / SIAM Member Price $44.10

-------------------------------------------------------

From: "Craig Lucas" <Craig.Lucas@manchester.ac.uk>
Date: Tue, 10 Jul 2007 15:18:31 +0100
Subject: HPC Workshop, Univ. Manchester, Aug 2007

The University of Manchester is holding a One Day Workshop on August 17th.
The speakers include:

* Jack Dongarra, University of Tennessee
An Overview of High Performance Computing and Challenges for the Future

* Stephen Jarvis, University of Warwick
The Production of Predictive Performance Models for Future Computing
Requirements

* John Levesque, Cray Inc.
Scaling to a Petaflop - It doesn't seem too difficult

* Marc Mendez, Bull
Losing Performance Due to Poor Coupling in the Execution Environment

For further details and registration please see the link below.
This event is free.

http://www.mc.manchester.ac.uk/services/courses/hpcworkshop

The event is part of our Parallel Programming Summer School.

-------------------------------------------------------

From: "Nicole C. Jorlett" <Jorlett@siam.org>
Date: Tue, 10 Jul 2007 12:49:43 -0400
Subject: CFP, SIAM International Conference on Data Mining (SDM08), Apr 2008

SIAM International Conference on Data Mining
Hyatt Regency Hotel, Atlanta, Georgia
April 24-26, 2008

The Call for Presentations for this conference is available at:
http://www.siam.org/meetings/sdm08

**Deadlines**

Abstract due: October 5, 2007

Manuscripts due: October 12, 2007

For additional information, contact SIAM Conference Department at
meetings@siam.org.

-------------------------------------------------------

From: "Nicole C. Jorlett" <Jorlett@siam.org>
Date: Tue, 10 Jul 2007 15:25:15 -0400
Subject: CFP, SIAM Conference on Optimization (OP08) , May 2008

SIAM Conference on Optimization
Boston Park Plaza Hotel and Towers, Boston, Massachusetts
May 10-13, 2008

Invited Plenary Speakers:
Etienne de Klerk, Tilburg University, Netherlands
Matthias Heinkenschloss, Rice University
Jan Modersitzki, University of Lubeck, Germany
Annick Sartenaer, Universite Notre Dame de la Paix, Belgium
Stefan Scholtes, Cambridge University, United Kingdom
Pascal Van Hentenryck, Brown University
Andreas Wächter, IBM Research
Robert Weismantel, University of Magdeburg, Germany

The Call for Presentations for this conference is available at:
http://www.siam.org/meetings/op08/

**Deadlines**

Minisymposium proposals: October 9, 2007 EDT

Abstracts for all contributed and minisymposium presentations:
November 8, 2007 EST

For additional information, contact SIAM Conference Department at
meetings@siam.org.

-------------------------------------------------------

From: "Kirsten Wilden" <Wilden@siam.org>
Date: Tue, 10 Jul 2007 10:26:40 -0400
Subject: Registration open, SIAM Conf. on Mathematics for Industry, Oct 2007

Subject: SIAM Conference on Mathematics for Industry: Challenges and
Frontiers (MI07) - Registration and Program Now Available!

Conference Name: SIAM Conference on Mathematics for Industry: Challenges
and Frontiers (MI07)

Location: Hyatt Regency Philadelphia, Philadelphia, Pennsylvania

Dates: October 9-11, 2007

Registration is Now Available!

Pre-Registration Deadline: September 11, 2007
Hotel Reservation Deadline: September 11, 2007

Registration and the preliminary program for this conference are
available at: http://www.siam.org/meetings/mi07/

For additional information, contact the SIAM Conference Department at
meetings@siam.org.

-------------------------------------------------------

From: Robert EYMARD <Robert.Eymard@univ-mlv.fr>
Date: Thu, 12 Jul 2007 11:56:22 -0400
Subject: CFP, 5th Inl. Sym. on Finite Volumes for Complex Apps. (FVCA5), Jun 2008

We are pleased to announce
the 5th International Symposium on Finite Volumes
for Complex Applications, which will be held
in Aussois, France, June 08-13, 2008, see

http://www.latp.univ-mrs.fr/fvca5/

The following speakers have accepted to give a keynote lecture:

Ivar AAVATSMARK, Bergen,
Jerome DRONIOU, Montpellier,
Edwige GODLEWSKI, Paris,
Christophe LE POTIER, Saclay,
Gianmarco MANZINI, Pavia,
Roland MASSON, Rueil-Malmaison,
Christian ROHDE, Bielefeld
Mary F. WHEELER, Austin

Contributed talks or posters are welcome: projects of papers related to
these talks or posters should be electronically submitted before the end
of november, 2007, and will be reviewed by two referees, before
publication in the proceedings.

A benchmark on finite volume schemes for anisotropic
diffusion problems on different types of grids will be organized,
and contributions to this benchmark will be greatly encouraged.
A special session will be dedicated to the presentation of the
results.

Precisions on the benchmark will be given on the conference
website by mid-september.

Would you be so kind as to forward this announcement
to anybody you might think interested ? We apologize for
possible multiple posting.

-------------------------------------------------------

From: Ronald Cools <Ronald.Cools@cs.kuleuven.be>
Date: Mon, 9 Jul 2007 11:16:11 +0200
Subject: Symposium on the birth of numerical analysis, Oct 2007

Symposium on "The birth of numerical analysis"
Leuven (Belgium), October 29-30, 2007

Numerical analysis has known an enormous evolution since the introduction
of digital computers in the years 1950-1960. This symposium wants to
bring together lecturers who were at the origin of this evolution to
testify about their experiences, and attendants who are the current
users of numerical analysis or who are just interested in the roots of
this discipline.

The following lecturers accepted our invitation:
Gerard Alberts, Kendall Atkinson, Claude Brezinski, Philip Davis, Jack
Dongarra, Brian Ford, Rolf Jeltsch, James Lyness, Robert J. Plemmons,
Mike Powell, Gerhard Wanner, Alistair Watson.

For more information and registration, see
http://www.cs.kuleuven.be/~ade/WWW/WOG/history/

This symposium is an activity of the Scientific Research Network
"Advanced Numerical Methods for Mathematical Modelling". It receives
additional funding from "Fonds voor Wetenschappelijk Onderzoek
Vlaanderen FWO" and "Fonds de la Recherche Scientifique FNRS".

-------------------------------------------------------

From: Peter Jimack <pkj@comp.leeds.ac.uk>
Date: Wed, 11 Jul 2007 10:34:58 -0400
Subject: Research Fellow in Parallel Scientific Computing (Leeds, UK)

This project builds upon existing collaboration between the School of
Computing and the Institute of Materials Research at Leeds in order to develop
a parallel 3-d implementation of phase-field models of the solidification of
dilute binary alloys. The primary challenges posed by these problems are their
severe nonlinearity combined with a high degree of stiffness, resulting from
substantial differences in thermal and solutal diffusion rates. Our prior work
in two dimensions has allowed us to develop an implicit adaptive solver that
uses a nonlinear multigrid scheme at each time step, and the aim of this
project is to develop a 3-d parallel version of this scheme.

The Research Fellow will work within the Scientific Computing group in the
School of Computing, and jointly with the Institute of Materials Research,
starting on October 1st 2007. Candidates should posess a PhD in Scientific
Computing, Numerical Analysis, Computational Materials Science, Computational
Physics, or a related subject. Candidates should also have existing expertise
in at least some of the following topics: parallel computing, phase-field
methods, multigrid methods, mesh adaptivity (and be able to demonstrate the
potential to quickly acquire expertise in the remainder).

The post is for a fixed period of eighteen months and is offered within
Research Grade 7 (£27,466 - £30,913 p.a.)

Further particulars are available from
http://www.comp.leeds.ac.uk/pkj/PF-fp.html

Closing date 27th July 2007.

-------------------------------------------------------

From: a <asherbaf@im.bas-net.by>
Date: Mon, 9 Jul 2007 15:43:49 +0300
Subject: Contents, Computational methods in applied matematics (CMAM)

Contents Comp. Meth. Appl. Math., Vol. 7 (2007), No. 2

COMPUTATIONAL METHODS IN APPLIED MATHEMATICS
All papers are available at our web-site http://www.cmam.info/issues/ .

Contents:

Simulation of integrating discontinious solutions on dynamically
adaptive grids
P.V.Breslavsky and V.Mazhukin (Russia)

On parallel numerical algorithms for simulating industrial filtration
problems
R. Ciegis (Lithuania), O.Iliev, and Z.Lakdawala (Germany)

Supraconvergence and supercloseness of a discretisation for elliptic
third-kind boundary-value problems on polygonal domains
E.Emmrich (Germany)

Mathematical modeling of geofiltration and geomigration processes in
multilayer systems
G.Gromyko, M.Chuiko, A.Smychnik, A.Hrechka, and A.Zlebava (Belarus)

Exact difference schemes for multidimensional heat conduction equations
M.Lapinska-Chrzczonowicz (Poland)

------------------------------
End of NA Digest

**************************
-------