NA Digest Sunday, November 19, 1989 Volume 89 : Issue 45

Today's Editor: Cleve Moler

Today's Topics:

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


From: Alan Cline <cline@cs.utexas.edu>
Date: Wed, 15 Nov 89 14:53:03 CST
Subject: Unsymmetric Eigenvalue Problem

Unsymmetric Eigenvalue Problem

A problem that is important in dynamical systems is the
computation of an eigenvalue of maximum real part given a large
(perhaps of order in the thousands), sparse, real, unsymmetric matrix.
(Since the matrix is real, there may be a complex conjugate pair
having the maximum real part.) The matrix is parameterized and the
region of interest is exactly that point where any of the eigenvalues
enter the right half-plane, so the maximum real part may be quite
small. Are there some suggestions for dealing with this problem?
Please direct responses to cline@cs.utexas.edu.

Alan Cline
The University of Texas at Austin


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

From: Norm Schryer <nls@research.att.com>
Date: Thu, 16 Nov 89 17:17:35 EST
Subject: Fortran 77 to C Converter Available via Netlib

The Unix command

(echo filter f2c; cat file) | mail netlib@research.att.com

converts the Fortran 77 in "file" to C and returns it via email.

The converted C requires a "header" file which can be obtained by:

echo "send f2c.h from f2c" | mail netlib@research.att.com

The command

echo "send index from f2c" | mail netlib@research.att.com

returns an index of useful f2c information available from Netlib
including a "man page", source for libraries, etc.

F2c "just" prints out the C parse tree internally produced by f77,
the reliable old Fortran 77 compiler of Stu Feldman.
The programming was done by D. M. Gay of Bell Labs and
Mark Maimone of Carnegie-Mellon University.

The usual caveat applies:
The service is free and you get what you pay for.

An additional caveat is that a daemon saves the Fortran submitted
to Netlib so we can do regression testing on f2c.

We intend to put f2c in the public domain and want as much
experience with it as possible before foisting it on the world.

Comments, questions and complaints to

Norm Schryer, nls@research.att.com, 201-582-2912.


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

From: Giovanni Monegato <MONEGATO%ITOPOLI.BITNET@Forsythe.Stanford.EDU>
Date: Mon, 13 Nov 89 09:03 GMT+1
Subject: International Conference on Numerical Methods

International Conference
on
NUMERICAL METHODS IN APPLIED SCIENCE AND INDUSTRY

Torino, Italy, June 18-21,1990

Organized by: Seminario Matematico dell'Universita e del Politecnico
di Torino

Scientific Committee: F.Brezzi, C.Canuto, L.Gatteschi, G.Monegato

Invited Speakers: F.Andreuzzi, O.Axelsson, I.Babuska, K.J.Bathe,
C.A.Brebbia, J.Douglas, W.Hackbusch, T.J.R.Hughes,
G.C.Hsiao, J.Lyness, G.Maier, P.P.Strona, W.Wendland

There will be no short communication sessions

Registration Fee: 200.000 Italian Lira; this amount includes the Proceeding
of the Conference, four working lunches, gala dinner.

For information write to:
Numerical Methods Conference
Dipartimento di Matematica
Politecnico di Torino
Corso Duca degli Abruzzi, 24
I-10129 Torino, Italy

or to MAT1@ITOPOLI.BITNET


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

From: Deepankar Medhi <dmedhi@vax2.cstp.umkc.edu>
Date: 14 Nov 89 08:46:00 CDT
Subject: Change of Address for Deepankar Medhi

I have left AT&T Bell Laboratories to join University of Missouri-Kansas
City starting this fall. Please note my new address:

Deepankar (Deep) Medhi
Computer Science Telecommunications Program
University of Missouri-Kansas City
5100 Rockhill Road
Kansas City, MO 64110-2499

Phone: (816) 276-2006

FAX: (816) 276-5159

E-mail: dmedhi@vax2.cstp.umkc.edu
or, dmedhi@umkcvax2.bitnet
or, na.medhi...


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

From: Jerzy Wasniewski <mfci!wasniews@uunet.UU.NET>
Date: Tue, 14 Nov 89 08:17:10 EST
Subject: Kaj Madsen Visiting Multiflow Computer

KAJ MADSEN TO SPEAK AT MULTIFLOW COMPUTER
December 15th, 1989, 11:30 a.m.
Multiflow Computer, Inc.
31 Business Park Drive
Branford, CT 06405
(203) 488-6090
No Fee
Attendance by reservation only
Please contact Mike Bernhardt to register for this lecture.

A METHOD FOR LINEAR L1-APPROXIMATION
Kaj Madsen, Institute for Numerical Analysis
The Technical University of Denmark
This talk concerns the problem of minimizing the sum of the
absolute values of m linear functions.
This problem is similar (equivalent) to the linear programming
problem, and it has previously been solved using active set
strategies, similar to the Simplex Method for this problem.
The new method presented is based on smoothing the
(non-differentiable) L1 problem. A Newton type method is applied
to the smooth problems. In the final stages of the iteration an
active set strategy is applied. Numerical experiments with the new
method indicate superiority to the simplex type methods by a factor
of at least 2.


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

From: Ian Gladwell <smu!gladwell@uunet.UU.NET>
Date: Fri, 17 Nov 89 09:12:38 CST
Subject: Positions at Southern Methodist

Southern Methodist University
Department of Mathematics
Junior Tenure-Track Positions

The Department of Mathematics at Southern Methodist University seeks
applications for junior tenure-track positions with employment beginning
Fall 1990. An outstanding candidate may be considered for a more senior
position. Candidates should be active researchers in applied or numerical
mathematics and should have a strong commitment to teaching. The
department has an active doctoral program. Applicants must be able to
teach graduate level courses in applied mathematics, numerical analysis or
scientific computation. The teaching load for each position is two courses
(six hours) per semester. There may also be visiting positions available
in academic year 1990-91.

Southern Methodist University has about 8000 students. The Department of
Mathematics has a strong and ongoing commitment to the development of
classical and modern applied mathematics. Thirteen of the sixteen
full-time faculty are applied or numerical mathematicians. Current areas
of research include mathematical modeling of physical and biological
phenomena, nonlinear waves, perturbation methods, fluid dynamics,
numerical bifurcation, mathematical software, numerical solution of
differential equations, and parallel computation. Candidates should be
active in one of these areas or a related one. Senior faculty and their
interests include D.H. Anderson (mathematical biology), W.E. Ferguson
(numerical partial differential equations), I. Gladwell (mathematical
software), R. Haberman (nonlinear waves), G.W. Reddien (bifurcation
theory), D.A. Reinelt (fluid dynamics) and L.F. Shampine (numerical
ordinary differential equations). Among its computers, the university has
a Sequent Symmetry for research use.

Applications must be received by January 10, 1990. Please send a vita and
three letters of recommendation to: Professor I. Gladwell, Chair,
Department of Mathematics, Southern Methodist University, Dallas, Texas
75275 (Tel: (214) 692-2506, FAX: (214) 692-4099) who may also be contacted
with any questions concerning the positions.

I. Gladwell's email addresses are:
na.gladwell@na-net.stanford.edu
and
h5nr1001 at smuvm1 (on bitnet)

SMU is an equal opportunity/affirmative action/Title IX employer.


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

From: Andrew White <abw@scrump.c3.lanl.gov>
Date: Fri, 17 Nov 89 15:06:58 MST
Subject: Group Leader Position at Los Alamos

Attached is a job ad for a Group Leader position at Los Alamos. I hope that
this is self-explanatory, but if not plese send me e-mail at abw@lanl.gov.

Andy White


The Computing and Communications Division at the Los Alamos National Laboratory
is seeking a candidate to provide management and technical leadership for the
Computer Research Group.

Current activities include general computational applied mathematics, signal
processing, mathematical algorithms and libraries, experimental computing
technology, computer performance measurement, expert systems, neural networks,
speech recognition, advanced computing architectures, user tools, languages,
pattern recognition, and image processing.

AREAS OF RESPONSIBILITY

o Providing management and leadership in computer research applicable to
problems of interest to the Laboratory
o Collaborating with universities, other Laboratories, and industrial
partners
o Recruiting and evaluating personnel
o Salary administration, financial reporting, and affirmative action
planning
o Communicating and interacting with management and staff throughout the
Laboratory, DOE, other government agencies, and academia

The successful candidate should have extensive experience in the following
areas: leading technical research efforts; managing and supervising technical
staff; demonstrated ability in interacting with funding agencies, other
research efforts, potential users of research results, and management; and
a proven research record in some area of computer research, application of
computers to problem solving, and development of computer software.

A Master's, Ph.D., or equivalent experience in computer science, physics,
mathematics, or related scientific or engineering field, is necessary.

The Los Alamos National Laboratory, which is located in the scenic mountains of
northern New Mexico, offers a competitive salary and benefits package.

FOR INITIAL CONSIDERATION, submit a resume no later than 12/1/89 to Andy
White, Los Alamos National Laboratory, MS B260, #90878RA, Los Alamos, New
Mexico 87545. For additional information, call Andy at (505) 667-6164.
TO FORMALLY APPLY, submit a resume to Ray Archuleta, Los Alamos National
Laboratory, Personnel Services Division, #90878RA, Los Alamos, New Mexico 87545.

Los Alamos is an Affirmative Action/Equal Opportunity Employer. Security
clearance required.


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

End of NA Digest

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