## NA Digest Sunday, September 2, 2007 Volume 07 : Issue 36

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: Tim Davis <davis@cise.ufl.edu>
Date: Fri, 31 Aug 2007 09:53:28 -0400
Subject: Morally triangular matrices

Cleve Moler defines a matrix A as "morally triangular" if
there exists a permutation P such that P*A is lower triangular.
In contrast, John Gilbert defines a matrix A as "psychologically
triangular" if there exist permutations P and Q such that P*A*Q
is lower (or upper) triangular. Morally triangular matrices
arise in MATLAB when you do [L,U]=lu(A); in this case, the partial
pivoting is captured in L, which is morally triangular. The right
factor, U, is simply triangular. Backslash then detects whether
or not L is morally triangular when computing x=L\b, finds the
permutation, and does the forward solve.

Thus, in MATLAB:

A = sprand (100,100,0.1) ;
b = rand (100,1) ;
[L,U] = lu (A) ;
spparms ('spumoni',2) ; % turn on sparse diagnostics
x = L\b ;

you get the following output:

sp\: bandwidth = 93+1+89.
sp\: is A diagonal? no.
sp\: is band density (0.34) > bandden (0.50) to try banded solver? no.
sp\: is A triangular? no.
sp\: is A morally triangular? yes.
sp\: permute and solve.

I have a hard time keeping track of the difference between
moral triangularity and psychological triangularity, and I suspect
others who read this digest do too. Thus, in honor of Cleve and
John, I wrote a helpful tool for remembering the difference:

Tweedle Tril and Tweedle Triu, by T.D.

I'm Tweedle Tril, he's Tweedle Triu,
We're the factors of sparse LU.
My brother: right and triangle;
Me? Triangularly moral.

I like my pivots small and sweet,
but the little pivots try me.
I watch my cholesterol and yet,
my chol ist err'd all high, I bet.

My brother likes his pivots large;
to which he's absolutely partial,
they say it keeps condition low,
but to him they just give fill-in.

The pivots to which he's partial
keep my condition suitable.
My doctors math all want to show
condition number very low.

We're quite complete, as you can tell;
no preconditions, none at all.
Factor fine with no conditions,
yet we love those graph partitions.

And when at last we come to rest
and find solutions you can test,
Take norm resid and see it's best,
and note how low is our condest.

I don't think morally triangular matrices are related to the
definition of "horror matrices", but in case they do, you can
find the definition at
http://www.cise.ufl.edu/~davis/Horror_matrices.html#horror .

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

From: "Bernard Beauzamy" <Bernard.Beauzamy@scmsa.com>
Date: Tue, 28 Aug 2007 15:08:53 +0200
Subject: Robust Mathematical Modeling Program, Invitation to Institutions

Dear Colleagues,

We are presently developing a research program called "Robust Mathematical
Modeling"; it started 3 years ago and includes several institutions, both in
the US and in Europe (companies and academia). The general objective of the
program is to develop tools which can handle the uncertainties occurring in
real life problems : on the data, on the laws, on the objectives. Please see
http://www.scmsa.com/robust.htm for a general description of the program.

This program is open : any institution can freely participate. We would like
more institutions (public or private, companies or academic) to join the
program. The benefits would be :

- each institution organizes regularly a local colloquium, in which are
presented the users' needs : so, having more colloquia, we would have more
experiences to share ;

- contributing to the list of tools which are already present on the web
site ;

- criticizing the existing tools. The web site appears as some kind of open
papers.

- participating to the PhD program which is associated with the RMM program
: students should travel to several countries and get various experiences.

Interested people should send an email to

Bernard.Beauzamy@scmsa.com

Best regards,
Bernard Beauzamy
University Professor (ret.)
Chairman and CEO, Societe de Calcul Mathematique S. A.,
111 Faubourg Saint Honore
75008 Paris, France

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

From: Nick Higham <Nick.Higham@manchester.ac.uk>
Date: Thu, 30 Aug 2007 12:23:31 +0100
Subject: MIMS Workshop on Large Graphs and Networks, Sep 2007

The Manchester Institute for Mathematical Sciences (MIMS) is hosting a
"Workshop on Large Graphs and Networks: Matrix Algorithms and
Applications" on Tuesday September 11, 2007, 2-4.30pm.
This is an informal afternoon workshop covering new developments in
matrix-based algorithms for large graphs and networks.
The speakers are

Des Higham (University of Strathclyde).
"Spectral Algorithms for Biological Networks"

Ilse Ipsen (North Carolina State University)

Abstracts and further information are at
http://www.mims.manchester.ac.uk/events/workshops/graphnet07/

---Nick Higham

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

From: Kees Oosterlee <C.W.Oosterlee@math.tudelft.nl>
Date: Tue, 28 Aug 2007 13:02:42 +0200
Subject: Numerics in Finance, Nov 2007

Call for Papers and Announcement for a Financial Workshop:

Workshop "Numerics for Finance", at the Commerzbank in Frankfurt am
Main, Germany, November 5th and 6th 2007.

Invited speakers:
Mike Giles (Oxford, UK),
Peter Kloeden (Frankfurt, Germany) and
Ludger Overbeck (Giessen, Germany).

In this workshop we will present the latest advancements in numerical
methods for financial applications. Techniques include, but are not
limited to,

-- PDE methods for pricing financial derivatives;
-- Recent development in monte Carlo methods;
-- Numerical techniques in risk management.

c.w.oosterlee@cwi.nl, or wittum@uni-hd.de Acceptance will be announced
before October 1st 2007.
The workshop fee will be 150 Euro for academic researchers, 300 Euro for
industrial participants.

The Workshop Organizers:
Cornelis Oosterlee (CWI, Amsterdam), Ruediger Seydel (Cologne),
Christoph Wagner (Frankfurt) and Gabriel Wittum (Heidelberg).

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

From: Carl Riehm <criehm@fields.utoronto.ca>
Date: Thu, 30 Aug 2007 09:52:40 -0400
Subject: Fields Institute Events, 2007-2010

Regular Seminars of possible interest to NA-net subscribers:
Actuarial Science & Financial Mathematics Group Meetings
http://www.fields.utoronto.ca/programs/cim/07-08/actuarialmath/index.html

Colloquium/Seminar in Applied Mathematics
http://www.fields.utoronto.ca/programs/scientific/07-08/applied_math/index.html

Fields Industrial Optimization Seminar
http://www.fields.utoronto.ca/programs/cim/07-08/optimization_seminar/index.html

PRMIA Risk Management Seminars
http://www.fields.utoronto.ca/programs/cim/07-08/PRMIA/index.html

Seminar Series on Quantitative Finance
http://www.fields.utoronto.ca/programs/cim/financial_math/finance_seminar/07-
08/index.html

Some specific events during the next year:

May 20-23, 2008, Fifth International Workshop on Taylor Model Methods
http://www.fields.utoronto.ca/programs/scientific/07-08/taylor-model/index.html

May 21-23, 2008, Symposium on Dependent Data Structures
http://www.fields.utoronto.ca/programs/scientific/07-08/dependent/index.html

August 11-15, 2008, Fields-MITACS Industrial Problem-Solving Workshop
http://www.fields.utoronto.ca/programs/scientific/08-09/FMIPW/

Future Fields thematic programs are
2007 (Fall) Operator Algebras
2008 (Winter/Spring) New Trends in Harmonic Analysis
2008 (Fall) Arithmetic Geometry, Hyperbolic Geometry and Related Topics
2009 (Winter/Spring) O-Minimal Structures and Real Analytic Geometry
2010 (Winter/Spring) Financial Mathematics

Details of all our programs can be found on our website
http://www.fields.utoronto.ca/

To be informed of upcoming events, please subscribe to our mailing list at
http://www.fields.utoronto.ca/maillist

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

Date: Thu, 30 Aug 2007 10:40:16 -0600 (MDT)
Subject: Copper Mountain Conference on Iterative Methods, Apr 2008

ANNOUNCING:
Tenth Copper Mountain Conference on Iterative
Methods
April 6 - 11, 2008

ORGANIZED BY:
Front Range Scientific Computations, Inc.

CO-ORGANIZED BY:
The Center for Applied Scientific Computing,
Lawrence Livermore Nat'l Lab
The University of Maryland

IN COOPERATION WITH:
The Society for Industrial and Applied
Mathematics

DOE, NSF, IBM, Boeing, LANL, LLNL, Sandia

THEMES:
Stochastic PDEs and Uncertainty
Page Rank and Markov Chains
Inverse Problems and Regularization
Optimization
Systems of PDEs
Hardware Effects / Multicore Architectures

SPECIAL FEATURES:
Student Paper Competition. Travel and lodging
assistance will be awarded to students and
new PhDs judged to have submitted the best
research papers.
Workshops - Informal Topical Discussions
Preliminary Proceedings

Student Papers January 10, 2008
Author Abstracts February 1, 2008
Early Registration March 2, 2008
Guaranteed Lodging March 6, 2008

FOR FURTHER INFORMATION:
- or contact -
Annette Anthony
FRSC
(303) 554-1232 or (480) 332-2026

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

From: Fatihcan Atay <atay@member.ams.org>
Date: Sun, 2 Sep 2007 13:37:49 +0200
Subject: Workshop on biological networks, Oct 2007

Genetic and Biological Networks: Models, Dynamics, and Simulations

A satellite workshop of the ECCS'07: European Conference on
Complex Systems, 1-5 October 2007, Dresden, Germany

Biological systems show emerging functional properties originating from their
many interacting units and their inherent multi-level structure. In recent
years, biologists, physicists, mathematicians, and computer scientists have been
joining efforts in order to understand the different aspects of the collective
phenomena exhibited by aggregates of cells and by regulatory networks. The aim
of this meeting is to bring together experts in such fields to discuss and
explore our understanding of complex biological networks and their applications
in the life sciences. Specific topics include, but are not limited to,
gene-regulatory, molecular, and metabolic networks, bioinformatics, population
dynamics, neural systems, evolutionary dynamics, and mathematical biology.

Interested researchers who wish to contribute a talk are invited to send a title
and abstract to fatay@mis.mpg.de

http://personal-homepages.mis.mpg.de/fatay/eccs07

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

From: Kees Roos <c.roos@tudelft.nl>
Date: Mon, 27 Aug 2007 12:41:27 -0400
Subject: Full professor position in Optimization at TU Delft

Optimisation is one of the key professorships at Delft Institute of Applied
Mathematics (DIAM), which forms an integral part of the Faculty of
Electrical Engineering, Mathematics and Computer Science (EEMCS). The Chair
lies at the centre of the mathematical research and teaching curriculum for
Applied Mathematics . In addition to research and teaching responsibilities,
the Professor will strengthen the role of the group in the Delft Research
Centre for Computational Science and Engineering and the various BSc and MSc
programmes.

The candidate should be a university graduate with a proven track record in
Optimisation, with specific emphasis on continuous optimisation and on the
algorithmic component of the field: the efficient solution of large-scale
optimisation problems. He/she has a PhD in Mathematics or a related
discipline and has published pioneering articles in top research journals.
Those applying will be enthusiastic team players, willing to make
significant contributions to research and education, with excellent teaching
skills, an inquiring mind, and good managerial abilities. Delft University
of Technology is a bilingual organisation; a good command of English
(written and oral) is essential, while good Dutch is also preferred.

The extent of the position is a maximum of 38 hours per week (1 FTE). The
starting salary for this position is commensurate with experience and
qualifications. The maximum monthly salary is  6,841 to  8,259 per month
gross, based on a full-time appointment. Benefits and other employment
conditions are in accordance with the Collective Labour Agreement for Dutch
Universities.
TU Delft strives to increase the number of women in higher academic
positions; women are thus especially encouraged to apply.

For further details on the position you are welcome to contact Prof. A.W.
Heemink, phone: + 31 (0)15  27 85813, e-mail: A.W.Heemink@tudelft.nl.

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

From: Peter Turner <pturner@clarkson.edu>
Date: Mon, 27 Aug 2007 11:10:09 -0400
Subject: Tenure track positions in applied math at Clarkson University

The Division of Mathematics and Computer Science (www.clarkson.edu/mcs)
invites applications for two tenure-track position in applied mathematics
starting in August 2008. One position will be filled at the Associate or
Full Professor level, the other is expected to be filled at the Assistant
Professor level.

We are especially interested in candidates with expertise in computational
areas of applied mathematics, including statistics, or dynamical systems, but
all areas of applied mathematics will be considered. Responsibilities will
directing graduate students. For the assistant professor level, minimum
requirements are a Ph.D. in mathematics by the date of appointment,
demonstrated excellence in both research potential and teaching ability, and
fluency in English. In addition, the candidate should be able to interact
with other faculty in the department and the university. For the senior
position, research, including a record of funding, and teaching records
commensurate with such an appointment will be required.

Applications including vita and three reference letters should be submitted
to Prof. P.A. Turner, Department of Mathematics and Computer Science,
Clarkson University, Potsdam, NY 13699-5815. Completed applications will be
reviewed starting immediately. Women and minorities are urged to apply.
Clarkson University is an AA/EOE Employer. (Pos. # 13-07, senior, Pos. # 14-
07, junior)

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

From: AarnesJrg <Jorg.Aarnes@sintef.no>
Date: Thu, 30 Aug 2007 13:45:49 +0200
Subject: Postdoctoral researcher - reservoir modeling and simulation

SINTEF ICT

SINTEF ICT is an institute within the SINTEF group with departments in
Trondheim and Oslo. SINTEF ICT undertakes contract research and
development projects for industry and the public sector in the area of
information and communication technology. We collaborate closely with
NTNU and the University of Oslo. SINTEF ICT has approx. 260 employees
and collaborates with a large number of industries and academic

The Department of Applied Mathematics at SINTEF ICT is expanding due
to an increased project portfolio. Our activities include modelling
and simulation of petroleum reservoirs, optimisation methods for
complex planning and scheduling applications, parallel computing,
visualisation, and geometric modelling. We are currently seeking an
innovative postdoctoral researcher with excellent mathematical and
computational skills wanting to contribute to high-quality research in
a young and creative research group.

The postdoctoral researcher will take part in a research project
funded by the Research Council of Norway, NTNU, and Shell. The project
aims to establish a proof-of-concept in an industrial setting for
multiscale reservoir simulation (see
http://www.math.sintef.no/GeoScale for some of our current
research). Reaching this goal requires a collaborative effort from a
dedicated team of researchers with advanced mathematical and
computational skills.

Qualifications:

· doctoral degree in applied mathematics, scientific computing, or
related subjects.
· solid background in mathematics and numerics, strong publication
record, and extensive experience with implementation of advanced
numerical methods
· good social skills and the ability to work independently as well as
in a team
· professional ambitions and desire to help build a world-class
research team

Prior experience with modelling and simulation of flow in porous
media, reservoir simulation, or geological modelling is an advantage,
but not a requirement.

The postdoctoral researcher is expected to:
· develop, implement, and validate new numerical methods
· contribute with innovative ideas, and help the research group reach
project goals
· publish in high-quality journals and give talks at international
conferences
· initiate and maintain contact with customers and collaborating
research groups

We offer:
· a creative and social work environment
· good opportunities for scientific and career development
· competitive salary and benefits
· group pension plan and insurance
· flexible work hours

Workplace: Oslo, Norway

Bjørgan <mailto:Knut-Andreas.Lie@sintef.no> (tel. +47 22 06 74 83) or
Research Scientist Jørg Aarnes <mailto:Knut-Andreas.Lie@sintef.no>
(tel. +47 22 06 77 95).

A brief and concise application with CV can be delivered online or by
e-mail to Nora.Gibb@sintef.no. Certificates and references to be
delivered only on request.

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

From: Peter Butkovic <peter.butkovic@virgin.net>
Date: Tue, 28 Aug 2007 03:45:25 -0400
Subject: Postdoctoral Research Fellowship at Birmingham (UK)

Applications are invited for a post-doctoral research fellow in the School
of Mathematics, University of Birmingham, to work on the EPSRC funded
project Feasibility and reachability in max-linear systems under the
supervision of the Principal Investigator, Dr P Butkovic. It is a condition
of this appointment that it starts on or before 1 March 2008 and its
duration is 33 months. The successful applicant will have completed their
first degree in either mathematics or physics and a PhD, preferably in an
area of mathematics with links to numerical linear algebra and/or discrete
optimisation. Candidates in the final stages of their PhD study will also be
considered. Starting salary is £25,134 pa. Travel allowance for conferences
will be available.

Further particulars can be seen on
http://www.vacancies.bham.ac.uk/vacancies/vacancySearch.htm

The closing date for applications is 20 September 2007.
If you wish to apply for this post please use the link "Applying for a job"
in the left column at Further Particulars. Any application should contain
email addresses of referees and should be accompanied by a CV and a list of
publications.

Informal enquiries prior to formal application are encouraged; they may be
made to Dr P Butkovic (P.Butkovic@bham.ac.uk).

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

From: Ke Chen <K.Chen@liverpool.ac.uk>
Date: Wed, 29 Aug 2007 09:06:40 +0100 (BST)
Subject: Contents, Special Issue, Int J of Computer Mathematics, Vol 84 Issue 8 2007

Volume 84 Number 8 of IJCM containing a special issue of
"Fast iterative and preconditioning methods for linear and
non-linear systems"
is now available on the T&F web site:
http://www.informaworld.com/smpp/title~content=g781667528~db=all

Block preconditioning for saddle point systems with indefinite (1, 1) block
pp.1117¨C1129, Michele Benzi; Jia Liu

Efficient d-multigrid preconditioners for sparse-grid solution of
high-dimensional partial differential equations
pp.1131¨C1149, H. Bin Zubair; C. C. W. Leentvaar; C. W. Oosterlee

Greedy Tikhonov regularization for large linear ill-posed problems
pp.1151¨C1166, L. Reichel; H. Sadok; A. Shyshkov

Multigrid based preconditioners for the numerical solution of two-dimensional
heterogeneous problems in geophysics
pp.1167¨C1181, I. Duff; S. Gratton; X. Pinel; X. Vasseur

Multilevel algorithm for a Poisson noise removal model with
total-variation regularization
pp.1183-1198, Raymond H. Chan; Ke Chen

Preconditioning and convergence in the right norm
pp.1199-1209, Andy Wathen

A short survey on preconditioning techniques for large-scale dense
complex linear systems in electromagnetics
pp.1211-1223, Yin Wang; Jeonghwa Lee; Jun Zhang

Frobenius norm minimization and probing for preconditioning
pp.1225-1248, T. Huckle; A. Kallischko

A treecode for potentials of the form r-\lambda
pp.1249-1260, Kasthuri Srinivasan; Vivek Sarin

A short note on singular values of optimal and superoptimal
preconditioned matrices
pp.1261-126, Xiao-Qing Jin; Yi-Min Wei

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

From: Chi-Wang Shu <shu@dam.brown.edu>
Date: Mon, 27 Aug 2007 17:17:16 -0400
Subject: Contents, Journal of Scientific Computing

Journal of Scientific Computing
http://www.springeronline.com/journal/10915

Volume 32, Number 3, September 2007

Efficient Numerical Solution of the Density Profile Equation
in Hydrodynamics
G. Kitzhofer, O. Koch, P. Lima and E. Weinmuller, pp.411-424.

Complex Valued Spectral Hermite Approximations for the
Actively Mode-Locked Laser
Kelly Black and John B. Geddes, pp.427-448

Practical Implementation of Krylov Subspace Spectral Methods
James V. Lambers, pp.449-476.

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

From: Ron Boisvert <boisvert@nist.gov>
Date: Mon, 27 Aug 2007 11:49:54 -0400
Subject: Contents, ACM Transactions on Mathematical Software

ACM Transactions on Mathematical Software
Volume 33, Issue 4 (August 2007)

to full text, see http://www.acm.org/toms/V33.html.

Article No. 22
TestU01: A C library for empirical testing of random number generators
Pierre L'Ecuyer, Richard Simard
40 pages

Article No. 23
hpGEM---A software framework for discontinuous Galerkin finite element
methods
Lars Pesch, Alexander Bell, Henk Sollie, Vijaya R. Ambati, Onno Bokhove,
Jaap J. W. Van Der Vegt
25 pages

Article No. 24
deal.II---A general-purpose object-oriented finite element library
W. Bangerth, R. Hartmann, G. Kanschat
27 pages

Article No. 25
A parallel symmetric block-tridiagonal divide-and-conquer algorithm
Yihua Bai, Robert C. Ward
23 pages

Article No. 26
Accurate numerical derivatives in MATLAB
L. F. Shampine
17 pages

Article No. 27
Algorithm 869: ODRPACK95: A weighted orthogonal distance regression
code with bound constraints
Jason W. Zwolak, Paul T. Boggs, Layne T. Watson
12 pages

Article No. 28
Remark on Algorithm 644
Masao Kodama
12 pages

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

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