NA Digest Sunday, March 18, 1990 Volume 90 : Issue 11

Today's Editor: Cleve Moler

Today's Topics:


From: Gene Golub <>
Date: Sun, 11 Mar 1990 11:10:33 PST
Subject: More on Pronunciation of Lanczos

At any meeting where I attempt to pronounce Lanczos, there is always a
Hungarian who corrects my pronunciation. It might be easiest to call Lanczos
by his original name: Loewy. He changed his name because of the repressive
situation in Hungary in the '20's.

By the way, Lanczos would have been 100 years old in 1993. Isn't it
appropriate that the numerical analysts hold a meeting in his honor?
I'm sure other scientists will hold a meeting but that usually means our
interests are pushed aside as was evidenced by the recent meetings dedicated
to Von Neumann and Gauss.



From: George Ostrouchov <ost@ostsun.EPM.ORNL.GOV>
Date: Fri, 16 Mar 90 12:10:31 EST
Subject: More on Pronunciation of Cholesky

It is not clear which of three possibilities is the correct
pronunciation of Cholesky. If the name is of Russian origin, there is
a possibility for three pronunciations: 'T-sholesky', 'Sholesky', and
'Kholesky'. Each of these reflects a possible first letter of the name
in the Cyrillic alphabet. (Unfortunately two of these possible
Cyrillic letters do not look like any Latin letters so I cannot include
them here.) All three of these would likely be transliterated into the
Latin alphabet in French as Cholesky. If the name is of Polish origin
(this is less likely because of the 'y' ending), it would not need to
be transliterated because Polish uses the Latin alphabet (with a few
small exceptions). The correct pronunciation would then be 'Kholesky',
as Cholesky is pronounced in Polish. Until we find out what Major
Andre-Louis Cholesky preferred or the exact origin of his name, all
three current pronunciations seem acceptable.

George Ostrouchov, Oak Ridge National Laboratory
(pronounced 'Ost-rou-khov' -- Octpoyxob )


From: Jim Demmel <>
Date: Sun, 11 Mar 90 14:17:06 EST
Subject: Access to IEEE Exceptions

This is in response to the query of David Bernholdt (NA Digest, v. 90, i. 10)
about the availability of IEEE floating point exception queries. There is
unfortunately no generally agreed upon way to do these queries. This has
had the negative effect of discouraging their use, because it is difficult
to transport them. More significant, it has discouraged the production of
expert software which uses the exception flags in clever ways, because
such software would not be easy to transport. There is an effort underway
to remedy this in the context of C. A committee called NCEG is meeting in
order to set standards for this issue. To get on their mailing list send
e-mail to David Hough at SUN Microsystems,

Jim Demmel


From: Paul Calamai <>
Date: Mon, 12 Mar 90 09:34:05 EST
Subject: Jaccard Metric

One of my graduate students has been trying to find references
to problems employing the Jaccard metric or the Jaccard
coefficient (without much success).
The Jaccard coefficient is used to measure the similarity
(or dissimilarity) of binary vectors where each vector
component indicates the presence or absence of some
characteristic or feature. This coefficient can be
interpreted as a metric (sometimes called the Tanimoto
metric) on the set of binary vectors.
The one article we've managed to locate is a 1983 article
in SIAM J. Sci. Stat. Comput., 4, "An algorithm for the
single facility location problem using the Jaccard metric"
by G.A. Watson.

We are trying to gauge the usefulness of this
metric/coefficient, find any applications and locate any
relevant references.

If you can help us in any way, please contact me by
email or snail mail. My email address is:
and my snail mail address is:
Paul H. Calamai
Dept. of Systems Design Eng.
Univerity of Waterloo,
Waterloo, Ont.



From: Joel Saltz <>
Date: Tue, 13 Mar 90 17:50:34 EST
Subject: Positions at ICASE

The Institute for Computer Applications in Science and
Engineering (ICASE) is seeking a researcher
in the area of parallel numerical methods.
We have a particular interest in
recruiting a researcher with interests in
in adaptive methods for solving partial differential equations or
in sparse linear algebra, we would
alternately be interested in recruiting
a researcher involved in the
development and analysis of new parallel algorithms.

We are accepting applications for
two year post-doctoral appointments with possible renewal,
along with applications of researchers on sabbatical
or unpaid leave who might be interested
a visiting appointment.

ICASE has an iPSC/2 (to be upgraded in a few weeks to
an 80860 based iPSC/RX), along with local access to
a Cray-2 and minisupercomputers.
We also have the usual assortment of Suns and graphics workstations.

Applications should respond by e-mail to
or should send resumes and descriptions of proposed
research to:

Dr. Robert G. Voigt
Director, ICASE MS 132C,
NASA Langley Research Center
Hampton VA

Joel Saltz
Lead Computer Scientist, ICASE


From: Paul Nelson <>
Date: Tue, 13 Mar 90 16:22:44 CST
Subject: Positions at Texas A&M

I am communicating to request your assistance in two matters:

1. The Computer Science Department at Texas A&M is seeking to fill
an endowed chair in the area(s) of Software Engineering and Parallel
Computing. Do you have any suggestions for possible candidates?

2. We also are seeking to fill a junior position in computational
matrix theory. Same question. (There follows the text for an ad that is
to appear in the March SIAM News. Usual terms for junior tenure-track
faculty include $30 k start-up funds, and a teaching load of one course
per term for an initial two years.)

If you would like further information about either of these, please
send me mail or call at 409/846-7917.

Paul Nelson

Texas A&M University
Department of Computer Science

Applications are invited for a position for a specialist in
computational matrix theory. Requirements include a PhD and evidence of
potential for excellence in both teaching and research. The position is
nominally at the level of Assistant Professor (tenure-track), but
extraordinarily well-qualified applicants for more senior appointments
will be considered. Strong preference will be given to those with the
doctorate in Computer Science, but applicants with doctorates in closely
related fields will be considered.

The Department of Computer Science presently consists of 30 faculty and
approximately 200 graduate students. It awards graduate degrees through the
Ph.D. Both the department and the College of Engineering have a strong
commitment to develop the Computational Science and Engineering Area.
Texas A&M has approximately 40,000 students, and is located in College Station,
approximately 90 miles northwest of Houston. The university has recently
installed a CRAY Y-MP, and the department/college anticipate purchase of a
state-of-the-art parallel processing system.

Applicants should send a resume, and three letters of reference to
Dr. Mac Lively, Search Committee Chairman, Department PN,
Texas A\&M University, College Station, TX 77843-3112.

Texas A&M University is an equal opportunity/affirmative action employer.


End of NA Digest