NA Digest Sunday, March 5, 1989 Volume 89 : Issue 9

Today's Editor: Cleve Moler

Today's Topics:


From: David Griffiths <>
Date: Mon, 27 Feb 89 08:46:41
Subject: NA and Applied Math positions at Dundee

New Academic Appointments Scheme
Lectureships in
Applied Mathematics
Numerical Analysis

Applications are invited for the above posts from can-
didates qualified and keen to contribute to research in
Applied Mathematics, Applied Analysis and Numerical

Although the posts are to be financed under the UGC's
"New Academic Appointments Scheme", it is intended that they
should be fully established and that the appointees should
become permanent members of the academic staff. The salary
initially attached to the appointments will be at the
appropriate point on the Lecturer Grade A Scale (#9,260 -
#14,500) or Grade B Scale (#15,105 - #19,310) determined by
qualifications and experience.

Applications with curriculum vitae (6 copies or, if
posted overseas, one copy in a format suitable for photo-
copying) and the names and addresses of three referees
should be lodged with the

Personnel Officer
The University
Dundee, DDl 4HN

by 31 March 1989. Please quote EST/6/89 for the Applied
Mathematics post and EST/7/89 for the Numerical Analysis

Further details are available on e-mail by sending to .


From: Alan J. Laub <>
Date: Sat, 25 Feb 89 14:49:02 PST
Subject: Conference on Aerospace Computational Control




This conference is a sequel to the effort begun at the 1987
Workshop on Multibody Simulation, and continued at the 1988
Workshop on Computational Aspects in the Control of Flexible Sys-
tems. It is intended to bring together researchers, developers,
and users of computational tools for aerospace control systems
modeling, design, and simulation. There has been an explosion of
new computational hardware and software technologies in recent
years. These advances have the potential to improve greatly the
performance of computational control software tools. Current
computational control tools are marginal for some of today's ap-
plications and will not be able to support many future NASA and
DoD missions. As a consequence, the control and dynamics commun-
ities have become active in exploring these possibilities.

The Conference will: 1) provide a window into current computa-
tional control research and development, 2) allow researchers to
exchange ideas and experience on the correctness, efficiency, and
usability of current computational tools, and 3) identify impor-
tant unsolved problems and technology gaps which can be dealt
with by Government, Industry, and University combined efforts.

Conference organizing committee:

Dr. G.K. Man (NASA/JPL), Dr. Larry Taylor (NASA/LaRC), Mr. John
DiBattista (NASA HQ), Dr. Elbert Marsh (NSF), and Major Terry
Hinnericks (DoD)


Papers are being accepted on all aspects of computational control
tools and methods:

Tool Requirements (requirements for future tools in the areas of
spacecraft, space robotics, aircraft, and space transportation

Multibody Simulation Formulation (recursive algorithms, symbolic
manipulation, closed topology and constraints)

Articulated Multibody Component Representation (system model
reduction, component model reduction, modal synthesis, damping)

Parallel Computation (coarse-grain parallelization, fine-grain
parallelization, hardware, software)

Real-Time Simulation (hardware-in-the-loop, human-in-the-loop,
hardware/software interface)

Control Software (analysis software, synthesis software, algo-
rithms for large-order systems, numerical properties, model

User Interface (database management, visualization techniques,

Applications (spacecraft, space robotics, aircraft, space tran-
sportation systems)

Testing and Verification (reference test cases, ground experi-
ments, flight experiments)


All attendees are required to register in advance. To pre-
register, please request a registration form from:

Jet Propulsion Laboratory
Attn: Ray Miller, M/S 198-326
4800 Oak Grove
Pasadena, CA 91109
phone: (818) 354-4734

The registration fee of $150.00 will cover a set of Proceedings,
paper abstracts, refreshments, and a buffet banquet the evening
of 29 August. A $75.00 registration fee is available to full-
time students (fee does not include banquet).


Blocks of rooms have been reserved for the conference at the fol-
lowing hotels:

Radisson Suite Hotel ((805) 988-0130)
Studio Suite: $67.00 single/double
2 Story Loft Suite: $77.00 single/double

Casa Sirena ((805) 985-6311)
$65.00 single/double (approximately 5 miles from conference site)

Oxnard Hilton ((800) 445-8667)
$57.00 single/double (approximately 3 miles from conference site)

Camarillo Motel 6 ((805) 388-3467)
$28.29 single, $35.92 double (15 minute drive from conference site)

The cutoff date for reservations at these hotels is 11 June,
1989. After that date, rooms will be on an "as available" basis.
All participants are responsible for their own lodging reserva-
tions, and should contact their hotel of choice directly.


The Oxnard area is served by four airports: Los Angeles Interna-
tional (LAX), Burbank, Oxnard, and Santa Barbara. Various forms
of ground transportation are available from these airports.
Further information will be sent with registration.


Conference attendance will be limited; therefore, all "registra-
tion forms" are due no later than 15 April, 1989.

All abstracts are due no later than 15 April, 1989. Abstracts
should be between 100 and 150 words in length. Accepted papers
must be provided in vugraph and facing text format no later than
1 August, 1989. A package detailing specifics on paper prepara-
tion will be forwarded to authors whose papers are selected.


Questions concerning registration or general administrative de-
tails should be addressed to Mr. Ray Miller, (818) 354-4734.

Questions concerning technical sessions and papers should be ad-
dressed to Dr. G.K. Man, (818) 354-7142. E-mail inquiries can be
directed to Prof. Alan J. Laub ( ).


From: Scott Lamson <steinmetz!vdsvax!>
Date: 27 Feb 89 22:12:00 GMT
Subject: Bezier Curve Fit

I am looking for an algorithm for fitting a 2D curve specified by a
sequence of 2D (x,y) coordinates with a Bezier curve defined by a
minimum number of control points. The Bezier curve should fall within
some numerical tolerance of the (x,y) set of points. The number of
control points should be as low as possible, but needn't be an
absolute minimum. A reasonable test would be to evaluate an N point
Bezier curve at M points (M>>N). Then try to reconstruct the N
control points given only the M (x,y) coordinates.

Thanks in advance.

Scott| ARPA:
Lamson| UUCP: uunet!steinmetz!sierra!lamson
(518)387-5795| UUCP: uunet!steinmetz!lamson!crd


From: J. Modi <>
Date: Tue, 28 Feb 89 22:00:47 GMT
Subject: Research Associate at Manchester

University of Manchester
Department of Computer science


Title of the project: Mapping Problem Classes onto Parallel Computing Systems

In this project funded by SERC we will be developing taxonomies of computational
models, and will to answer the question: what is the best type of computational
model for a given class of problems!

An application oriented research assistant with background in
linear algebra, and an applicant with background in parallel
architectures are required.

The appointees will be encouraged to pursue their own research as well.

The appointment will be for 3 years on a RA1A scale ( $9865-$15720) starting as soon as possible.

Please send letters of application to Dr J. Modi at Computer Science Dept.,
University of Manchester,Oxford Road, Manchester M13 9PL, ENGLAND,as soon as possible.


From: Roland Sweet <rsweet%copper@boulder.Colorado.EDU>
Date: Wed, 1 Mar 89 16:19:11 mst
Subject: Information Requested on "PC Micro-Dif"
Cc: rsweet%copper@boulder.Colorado.EDU
Status: R

I just received an advertisement for a differential equations tutor software
package called "PC Micro-Dif" produced by WKM Associates. Has anyone had any
experience with this package or does anyone know of an existing review of it?

Thanks, Roland Sweet
na.rsweet... or


From: Joe Traub <>
Date: Thu, 2 Mar 1989 18:11:55 EST
Subject: Final Program-Complexity Symposium


April 3-5, 1989

SUPPORT: This Symposium is supported by a grant from the National
Science Foundation.

REGISTRATION: The symposium will be held at the Kellogg Conference
Center, on the fifteenth floor of the International Affairs Building,
Columbia University, 118th Street and Amsterdam Avenue. Registration
will start at 9:00 a.m. THERE IS NO REGISTRATION CHARGE.

PUBLICATION: Invited papers will be published in the Journal of Complexity.

FOR FURTHER INFORMATION: If you have any questions, contact Kerny
McLaughlin, Computer Science Department, Columbia University, (212)

There will be two hour-long plenary lectures and thirty-four invited
papers as well as contributed papers.


Steven Smale, "Godel, undecidability, chaos"

Leon N. Cooper, "Neural networks in real world applications"

Z. Galil, J. Tsitsiklis, H. Wozniakowski, D. Lee, G. Wahba, D. Donoho,
L. Cooper, Y. Abu-Mostafa, E. Baum, G.W. Wasilkowski, D. Saari, A. Werschulz

S. Heinrich, K. Sikorski, L. Blum, L. Levin, B. Kacewicz, J.M. Steele,
I. Johnstone, J. Kuczynski, T. Boult, Y. Yomdin, L. Plaskota

F. Gao, M. Milanese, M. Kowalski, U. Vazirani, M. Shub, E. Kalai, M. Kon,
E. Novak, P. Tiwari, A. Sukharev, J. Renegar, A. Bojanczyk


From: Brynjolfur Thorsson <mcvax!hafro!krafla!>
Date: 3 Mar 89 09:59:53 GMT
Subject: Fitting of Data Using Two Exponentials

For the last month or so I have been searching for some method that
fits data using two unrelated exponents, i.e.


I have found a method, that is supposed to work for x=x0+i*dx, but my
data is not equispaced. I know that I can make my data equispaced, using
splines, but I am afraid that I may loose accuracy. If someone knows a
solution to my problem and is willing to mail it to me I would be grateful.

Brynjolfur Thorsson
University of Iceland


From: T. R. Hopkins <>
Date: Sat, 4 Mar 89 21:24:02 GMT
Subject: NAG Users Digest Info

NAGMAG - An Electronic Mail Digest for Users of NAG products

At the recent NAGUA meeting in Brest interest was expressed in setting up an
informal electronic mail digest for the discussion of any aspects of
NAG services.

The aim of this digest is to provide a forum for anyone using any NAG
products to seek advice, discuss problems, complain, report bugs, suggest
improvements and extensions to services or anything else they feel is

The success of such a digest is totally dependent on its subscribers, we
would hope to send out issues at weekly intervals to ensure
a reasonable response to queries. Obviously the frequency could be
increased if the volume of traffic warranted it.

If you wish to subscribe to NAGMAG mail (Janet) (Arpanet)

giving your name, address and e-mail address. Failing the above routes
send the same information either snail mail to

Tim Hopkins
Computing Laboratory
University of Kent at Canterbury
Kent CT2 7NF
or via FAX to +44 227 762811
when we will do our best to sort out a path to you.

Anyone wishing to send articles for inclusion
in the digest mail nagmag via any of the above routes.

Tim Hopkins


End of NA Digest