NA Digest Thursday, October 15, 1987 Volume 87 : Issue 73

This weeks Editor: Cleve Moler

Today's Topics:


Date: Thu 8 Oct 87 08:43:03-MDT
Subject: Faculty Positions at University of Utah

The University of Utah Math Department has several positions available.
Most of these will be filled in the areas of Applied Mathematics,
particularly Scientific Computing and Numerical Analysis. We may be
able to hire a group of people who already work together.

I'll be glad to provide any additional information. Send
inquiries to


on the arpanet. Here is the official announcement:

University of Utah
Department of Mathematics

invites applications for the following positions:

1. Three full time tenure track appointments are available. Two of these
are to be at the Assistant Professor or Associate Professor rank and one is
to be at the Associate Professor or Professor rank. Selection will be based
on research expertise and teaching ability. Applications will be accepted
until January 31, 1988 or until all three positions are filled.

2. One tenure track position of one-third to full time at the rank of
Associate Professor or Professor. Selection as above. Applications will
be accepted until November 30, 1987 or until the position is filled.

3. Two or more nonrenewable three-year instructorships. Persons of any age
receiving Ph.D. degrees in 1987 or 1988 are eligible. Applicant will be
selected on the basis of ability and potential in teaching and research.
Starting salary this academic year is $27,000 and cost of living increases
are contingent on action by the State Legislature. Duties consist of
teaching two courses through the academic year. Applications will be
accepted until February 28. 1988, or until the positions are filled.

4. One or more visiting positions of one year or less. Selection criteria
are teaching ability and potential contribution to our research
environment. Applications will be accepted until February 29, 1987, or
until the positions are filled.

Applications must include curriculum vita, bibliography and three letters
of reference. (Instructorship applications must also include an abstract of
thesis and a list of graduate courses completed or transcripts.

Please send your application to

Committee on Staffing
Department of Mathematics
University of Utah
Salt Lake City, Utah 84112

The University of Utah is an equal opportunity-affirmative action employer.


Date: Fri, 9 Oct 87 12:20:24 CDT
From: Brian Smith <>
Subject: Addendum to the Fortran 8x Announcement

Addendum to the Announcement of the Public Review
of the Draft Fortran Standard.

The printing of the draft Fortran standard Fortran 8x by Global
Engineering Documents has been delayed. Therefore, the document
is not available from Global until October 19, 1987. To request the
document from them, refer to it as X3.9 Draft (Fortran), not Fortran 8x,
or draft proposed standard for Fortran.

The date of the availability of the book by Metcalf and Reid, titled
Fortran 8x Explained, was incorrect. It is available from Oxford University
Press in November, 1987 (not 1988 as stated in the previous announcement).
Orders for the book are currently being taken at the following address:

Oxford University Press
200 Madison Ave.
New York, NY 10016

--Brian Smith


Date: 12 October 87 18:22-MST
From: Hans Mittelmann <aihdm@asuacad.bitnet> <na.mittelmann>
Subject: Software for Generalized Eigenproblem

What is the best available code (pref. FORTRAN) to find the smallest
positive eigenvalue of the problem (only the eigenvalue is needed)

(A - lambda*B)x = 0, A and B symmetric but indefinite

where A and B have the sparsity typical for finite element discretiz-
ation of PDE's. A public domain code or a separately available routine
from a package would be preferable.


Date: Tue, 13 Oct 87 11:31:04 EDT
From: William Hager <>
Subject: Alternative to EISPACK

In the October 8, 1987 issue of NA Digest, Cleve Moler and
Eric Grosse reported a problem concerning the calculation
of small eigenvalues using the EISPACK routine TQLRAT.
[Ed. note: More precisely, the difficulties were associate
with underflow, not small eigenvalues, per se.]
Please note that Kahan with Pal and Walker developed a
root free QR algorithm called the PWK method by Parlett
(see Parlett's book, The Symmetric Eigenvalue Problem, pp. 164-169)
which addresses the problem associated with the
calculation of small eigenvalues using TQLRAT. This
algorithm is implemented in the software package NAPACK
which is available from Netlib as well as Psulib. For the 9x9
symmetric matrix mentioned by Cleve and Eric, the relevant routine
is SVALS (mail netlib@research.uucp ... send svals from napack
or mail ... send svals from napack).

Bill Hager (


Date: Wed, 14 Oct 87 09:13:58 cdt
From: Peter Tang <tang@anl-mcs.ARPA>
Subject: Augment that runs on UNIX

Does anybody have a version of the precompiler augment
that runs on UNIX? I would like to run one of my programs
in multiple precision using Brent's MP package.

Any related information will be appreciated.


Date: 15-OCT-1987 10:33:54
From: John Carroll <>
Subject: The EUROMATH Project

FROM: Dr. John Carroll
School of Mathematical Sciences
Dublin 9
Tel: +353 1 37 00 77

I write in the hope that your readers can assist me in compiling
a state-of-the-art report on the central issues of the Euromath
project (a brief summary of which is attached for your
information) and / or contribute their views on a likely
structure for the eventual Euromath system.

The areas of interest include technical word processing, networks
and communication protocols, electronic mail and
tele-conferencing as well as the crucial issues surrounding the
provision of a (distributed or centralised) database of
mathematical literature. Also relevant is the problem of
"portability" of mathematical text, how it can be transmitted
electronically and viewed in or retrieved from a suitably
constructed database.

Primarily I am interested in learning about what is currently
available in any or all of the above areas or information
(reports, reviews or other forms) about any existing products
similar to the components of the proposed Euromath system. Any
information on products or systems which will soon become
available is also welcome.

It is hoped that one of the consequences of the Euromath project
will be a clear recommendation for a standard approach to
technical word processing and the communication of mathematical
text. If successful, it will provide the basis for other
scientific disciplines to follow suit.

I look forward to hearing from you,

John Carroll



The objective of EUROMATH is to improve the research environment
for European mathematicians with the aid of modern information
technology. By establishing an integrated information retrieval
and communication system as well as a technical word processing
standard, it will strive to stimulate the research potential
within mathematics in Europe, increase the availability of
mathematical research and create an environment which will
encourage mathematicians towards increased collaboration through
effective communication.

The first phase of the EUROMATH project will produce guidelines
for the provision of the following inter-related facilities:

Information Retrieval: Access to various directories,
to (reviews of) published literature, to other
databases as well as drafts and notes of individual

Inter-personal Communication: Provision of suitable
electronic mail and electronic conferencing facilities.

Document Preparation and Delivery: The establishment of
a European standard for mathematical communication
embracing the main activities of entering, editing,
transmitting, receiving and printing mathematical

Traditionally, mathematicians have relied on computers mainly for
such specific tasks as scientific computation and symbolic
manipulation. A goal of EUROMATH is to expand computer usage by
enabling easy access to modern communication facilities. Its
basic concept is however equally applicable to other scientific
disciplines. The success of EUROMATH could inspire others to see
the benefits of a modern, full-scale solution to the
communication needs of a scientific community.

The CEC project EUROMATH (The Integrated Database and
Communications System for European Mathematicians), is a
collaborative effort linking CWI (Amsterdam), DDC (Copenhagen)
and NIHE (Dublin) in a technical partnership under the management
team of CRC (Dublin) and EMT (European Mathematical Trust).


Date: Wed, 14 Oct 87 20:48:13 pdt
From: Masaaki Shimasaki <>
Subject: Kyoto Supercomputer Symposium

Conference Announcement
Title: 4th Annual Kyoto University Symposium on
Supercomputing by Vectorcomputers
Date : March 14-15, 1988
Place: Data Processing Center
Kyoto University
Kyoto, Japan
Contact: Dr. M. Shimasaki
Associate Professor
Kyoto University
Kyoto, 606 JAPAN
E-Mail Address:


End of NA Digest