NA Digest Sunday, March 13, 1988 Volume 88 : Issue 11

Today's Editor: Cleve Moler

Today's Topics:

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


From: Hans Munthe-Kaas <munthe_kaas%vax.runit.unit.uninett@TOR.nta.no>
Date: 7 Mar 88 15:36 +0100
Subject: Visiting Positions at NTH, Norway

VISITING POSITIONS AT THE SUPERCOMPUTING CENTRE OF THE NTH IN NORWAY

At The Norwegian Institute of Technology (NTH) and the research institute
SINTEF a Cray X-MP/28 was installed in January -87. At the same time a group
called Section for Industrial Mathematics (SIMa) was founded. This group
is a link between the mathematical sciences department at the NTH and the users
of the supercomputer. NTH and SINTEF are situated in Trondheim which is the
centre for technological R&D in Norway. Combined they have more than 3500
people employed in research, and about 6000 students at NTH.

The SIMa group is conducting research in both algorithm development and in
applications of supercomputing. Our main areas of interest are currently:

Algorithms: Image processing, Numerical ODE, Linear Algebra, Inverse Problems.

Applications: Marine Hydrodynamics, Oil reservoir simulations, Image processing
of NMR / tomographic pictures, Civil engineering, Electrical engineering,
Mechanical engineering, Chemical engineering, Computer animation.

THIS NOTE IS TO GET IN CONTACT WITH RESEARCHERS INTERESTED IN SUPERCOMPUTING,
BOTH FROM AN APPLICATIONS AND FROM AN ALGORITMIC POINT OF VIEW.

We can offer (depending on the length of stay):
Travel expences, Local expences, Office space, Cpu-time on CRAY,
Salary up to 3 months.

Besides this, Trondheim is a nice town in the middle of Norway with
its roots back in the viking ages. A visit to Trondheim can well be combined
with holiday in Norway.

Interested people should contact:
Prof. B. Pettersen, arpa: b-pettersen%vax.runit.unit.uninett@nta-vax
bitnet: BJORNAR at NORUNIT
or Prof. S.P. Norsett, nanet: na.norsett@score.stanford.edu
arpa: norsett%vax.runit.unit.uninett@nta-vax
bitnet: NORSETT at NORUNIT

Mail: SIMa/RUNIT
N-7034 Trondheim
Norway

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

From: Per Hansen <hansen@msr.EPM.ORNL.GOV>
Date: Tue, 8 Mar 88 15:21:05 EST
Subject: Hankel and Toeplitz Matrices

Matrix-vector-multiplication with Hankel and Toeplitz matrices.

Does anyone have algorithms or, preferably, Fortran programs for fast multi-
plication of a vector by a matrix with Toeplitz- or Hankel-form?
Please reply via e-mail to: na.hansen@na-net.stanford.edu.

Per Christian Hansen
Copenhagen University Observatory
Oster Voldgade 3, DK-1350 Copenhagen K, Denmark


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

From: Gene Golub <golub%ifi.ethz.ch@relay.cs.net>
Date: 8 Mar 88 23:16 +0100
Subject: NATO summer workshops

As indicated in a previous message, there will be a NATO Advanced Study
Institute (ASI) on NUMERICAL LINEAR ALGEBRA, DIGITAL SIGNAL PROCESSING
and PARALLEL ALGORITHMS in Leuven, Belgium during the period August 1-12.
Applications must be sent to
Danny Sorensen (na.sorensen@na-net.stanford.edu) or
Paul Van Dooren (na.vandooren@na-net.stanford.edu)
by March 20, 1988. More detailed information and an application
form have been previously distributed via NA-NET, and a copy of the
announcement can be obtained via netlib.


There will also be MINI-WORKSHOPS in association with the ASI during
the period of July 18 thru Aug 26. All are invited to submit
proposals, even if they do not plan to attend the ASI .The
MINI-WORKSHOPS of, say, 2 days each should be organized around
certain themes. We can guarantee some support (housing/meals).

The meetings scheduled so far are :
Adaptive filtering (Bellanger) Aug. 16-17
Identification (De Moor-Vandewalle) Aug. 18-19
VLSI architectures for real time sign. proc. (Cattoor-Deman) July 25-29

Send proposals to Van Dooren or Sorensen.

Hope to see you in Leuven,
Gene Golub


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

From: Bob Russell <russell%lccr.sfu.cdn%ean.ubc.ca@relay.cs.net>
Date: 8 Mar 88 21:54 -0800
Subject: New Book on Boundary Value Problems for ODEs

Uri Ascher, Bob Mattheij, and I (Bob Russell) have recently completed
our book Numerical Solution of Boundary Value Problems for ODEs
(ISBN 0-13-627266-5). We have attempted to make it a comprehensive
treatment of the topic, suitable for both researchers and graduate
students. The publisher is Prentice Hall - their series for Computa-
tional Mathematics (in which some are unfortunately out of print).

Chapter titles: 1. Introduction
2. Review of Numerical Analysis/Math. Background
3. Theory of ODEs
4. Initial Value Methods
5. Finite Difference Methods
6. Decoupling
7. Solving Linear Equations
8. Solving Nonlinear Equations
9. Mesh Selection
10. Singular Perturbations
11. Special Topics
Appendices with Codes

Ordering Information:
US: 201-767-5937 Prentice-Hall, College Operations, Englewood Cliffs, NJ 07632
Canada: Carl Henderson, Prentice-Hall Canada,1870 Birchmount Road,
Scarborough, Ontario M1P 2J7
International: Simon & Schuster International Customer Service Group,
200 Old Tappan Road, Old Tappan, NJ 07675, USA


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

From: Douglass Turner <thorin!unc!turner@mcnc.org>
Date: 11 Mar 88 15:05:11 GMT
Subject: Solution of Quartic Equations


Hello, does anyone know of a good, robust, method to find all roots of a
quartic equation?. I am a computer graphics person and am writing
code to intersect a ray with a quartic (as part of a ray tracing program).
The problem boils done to finding valid roots of a quartic. I have heard
that direct solution is prone to numerical instability, so an iterative
technique is called for. I do have a way of getting good inital guesses
to start a rather slow iterative method like regula falsi, but I'm looking
for something that converges as quickly as possible. Perhaps a hybrid
method of some kind.

Any code, pseudo-code, or description of algorithm would be much
appreciated. I intend to post a synopsis to the comp.graphics group
(and this one if someone wishes).

Thanks.
Douglass Turner

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

End of NA Digest

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