From owner-reliable_computing [at] interval [dot] louisiana.edu Sun Jun 1 23:21:25 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h524LPd03223
for reliable_computing-outgoing; Sun, 1 Jun 2003 23:21:25 -0500 (CDT)
Received: from cs.utep.edu (mail.cs.utep.edu [129.108.5.3])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h524LIA03219
for ; Sun, 1 Jun 2003 23:21:19 -0500 (CDT)
Received: from aragorn (aragorn [129.108.5.35])
by cs.utep.edu (8.11.7/8.11.3) with SMTP id h524LAb10057;
Sun, 1 Jun 2003 22:21:12 -0600 (MDT)
Message-Id: <200306020421.h524LAb10057 [at] cs [dot] utep.edu>
Date: Sun, 1 Jun 2003 22:21:10 -0600 (MDT)
From: Vladik Kreinovich
Reply-To: Vladik Kreinovich
Subject: interval talks at FUZZ-IEEE'03
To: reliable_computing [at] interval [dot] louisiana.edu
Cc: sstarks [at] utep [dot] edu
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: PvwaOd2Hg3SUWD/oKUW0gA==
X-Mailer: dtmail 1.3.0 @(#)CDE Version 1.4 SunOS 5.8 sun4u sparc
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Interval-Related Talks at the 2003 IEEE International Conference on
Fuzzy Systems, St. Louis, Missouri, May 25-28, 2003
On May 25-28, an annual international conference on fuzzy systems was
held in St. Louis, Missouri. As usual, there were quite a few
interval-related presentations.
The first day of the conference was devoted to tutorials. Out of five
tutorials, two had a strong interval-related component: Jerry Mendel
from the University of Southern California
described how to use interval-valued (and more general) fuzzy
techniques, and Oscar Castillo and Patricia Melin from the Tijuana
Institute of Technology, Mexico, described how
interval-valued fuzzy sets can be combined with fractals in industrial
applications.
The 2002 Pioneer Award presentation was given by Didier Dubois, a
co-Editor-in-Chief of the international journal Fuzzy Sets and
Systems. His talk was about possibility theory, a formalism for
describing uncertainty in which for each event E, instead of a single
number, we use two values: the degree to which E is possible and a
(smaller) degree to which E is necessary. Interval-related techniques
for processing the corresponding interval-valued degrees of
uncertainty are an important part of possibility theory. D. Dubois
especially emphasized the importance of interval-related techniques
and algorithms developed by W. Lodwick and S. Ferson.
Intervals were also explicitly mentioned in a highlighted talk that
Lotfi Zadeh, the father of fuzzy logic, have at the beginning of the
special session on fuzzy logic and the Internet. He emphasized the
importance of using intervals--as well as other techniques for
describing uncertainty, e.g., probabilistic and fuzzy--in designing
effective algorithms for intelligent web search.
Intervals were also used in several regular talks. J. M. Mendel argued
that interval-valued fuzzy properties P, in which, for each object x,
the degree of our certainty m(x) that x has the property P
is not an exact number (as in usual fuzzy
theory) but an interval, represent words more adequately, and, in
addition, are more in line with the scientific approach necessity for
experimental verification: since different experts usually produce
somewhat different degrees, we cannot claim that a single degree can
be experimentally verified, but we can claim that all (or almost all)
expert degrees are within a given interval. The possibility of defining
natural subjective probabilities on finite and infinite intervals was
discussed in a paper by H. T. Nguyen from New Mexico State University
and V. Kreinovich and L. Longpre from the University of Texas at El
Paso. O. Castillo and P. Melin illustrated their general idea
of combining fractals with interval-valued fuzzy sets on the example
of plant monitoring and diagnostics. In their other paper, P. Melin
and O. Castillo showed that the use of interval-valued fuzzy logic in
intelligent control enables us to avoid unnecessary oscillations
caused by the "random" fluctuations of degree of certainty from expert
to expert and thus, lead to a smoother control. Similarly, F. Ree and
C. Hwang from the Hanyang University, South Korea, showed that the use
of interval-valued fuzzy degrees leads to a smoother and less noisy
clustering.
Several talks used the fact that a fuzzy number can be interpreted as a
nested family of intervals--intervals consisting of all the values for
which m(x) >= a for different threshold a. As a result, layer-by-layer
interval computations can be used to process fuzzy
data. S. Auephanwiriyakul from Ching Mai University, Thailand, used
interval computations to compress the set of possible responses to a
managerial survey into several typical responses, and M. Popescu and
J. Keller from the University of Missouri-Columbia, together with
P. Gader from the University of Florida, used interval computations in
sequence recognition problems such as speech recognition, landmine
detection, gesture recognition, and bioinformatics (incidentally, Jim
Keller was a General Chair of the conference).
During the conference, a meeting of the IEEE Fuzzy Technical Committee
was held. This was the first meeting since Jerry Mendel was elected
the committee's chair. With the encouragement of J. Mendel, the
committee decided to organize an Interval Task Force with
the goal of promoting the use of interval techniques in fuzzy
applications and of increasing interaction between interval and
fuzzy communities.
The motto of the conference was "Exploring new frontiers". This motto
fits very well with the frontier spirit of the city of St. Louis, the
city that now celebrates the 100th anniversary of the Lewis and Clarke
expedition, an expedition that starts in St. Louis and mapped a large
area of the American interior. For many years, St. Louis was the place
from which trails leading West started. (One of us (S.A.S.) has a
special connection to St. Louis: his great-grandfather, when he was a
baby, was kidnapped by a family moving West and raised as their own
son.) During the slavery years, the Mississippi river that flows
through St. Louis was the border between the slave-owning and free
states, so many runaway slaves crossed the river to gain their
freedom. In the 20th century, Charles Lindberg, St. Louis' famous son,
flew from St. Louis to New York City and then to Paris to become the
first pilot to cross the Atlantic alone. We the participants
enjoyed the spirit of old St. Louis during the Jazz Festival and
during the banquet that was held on an antique boat
"Becky Thatcher" (named after Tom Sawyer's girlfriend) that was slowly
cruising the mighty river.
Scott A. Stars and Vladik Kreinovich
From owner-reliable_computing [at] interval [dot] louisiana.edu Tue Jun 3 10:25:17 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h53FPGr05175
for reliable_computing-outgoing; Tue, 3 Jun 2003 10:25:16 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h53FPBV05170
for reliable_computing [at] interval [dot] louisiana.edu; Tue, 3 Jun 2003 10:25:11 -0500 (CDT)
Received: from lcyoung.math.wisc.edu (lcyoung.math.wisc.edu [144.92.166.90])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h4UDQWh00298
for ; Fri, 30 May 2003 08:26:33 -0500 (CDT)
Received: from ultra1.math.wisc.edu (ultra1.math.wisc.edu [144.92.166.161])
by lcyoung.math.wisc.edu (8.11.6p2/8.11.6) with ESMTP id h4UDO9209196;
Fri, 30 May 2003 08:24:10 -0500 (CDT)
Date: Fri, 30 May 2003 08:24:09 -0500 (CDT)
From: Hans Schneider
To: NETS -- at-net , E-LETTER ,
Pradeep Misra ,
Shaun Fallat ,
"na.digest" , ipnet-digest [at] math [dot] msu.edu,
SIAGLA-DIGEST , wim@bell-labs.com,
hjt [at] eos [dot] ncsu.edu, SMBnet [at] smb [dot] org, vkm [at] eedsp [dot] gatech.edu,
reliable_computing [at] interval [dot] louisiana.edu
Subject: LAA contents
Message-ID:
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-MailScanner: Found to be clean
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Dear Net Organizer:
Please circulate the attached LAA contents over your net.
Thanks
hans
---
*********************************************************************
Hans Schneider
Office:
Mathematics Department
Van Vleck Hall
University of Wisconsin
480 Lincoln Drive
Madison, WI 53706-1313 USA Email: hans [at] math [dot] wisc.edu
Office Phone: 608-262-1402 WWW: http://www.math.wisc.edu/~hans
Math Dept Phone: 608-263-3054
Math Dept Fax: 608-263-8891 Home - NOTE!
521 Samuel Drive
Madison, WI 53717 USA
608-833-6567
*********************************************************************
* Linear Algebra and its Applications
Volume 368, Pages 1-388 (15 July 2003)
http://www.sciencedirect.com/science/issue/5653-2003-996319999-433305
TABLE OF CONTENTS
An SVD-like matrix decomposition and its applications, Pages 1-24
Hongguo Xu
Completions of partial P-matrices with acyclic or non-acyclic associated graph, Pages 25-51
C. Jordan, J. R. Torregrosa and A. M. Urbano
Existence and construction of nonnegative matrices with complex spectrum, Pages 53-69
Oscar Rojo and Ricardo L. Soto
Inequalities for numerical invariants of sets of matrices, Pages 71-81
Jairo Bochi
On the positive definite solutions of the matrix equations Xs+/-ATX-tA=In, Pages 83-97
Xin-Guo Liu and Hua Gao
Some determinantal inequalities for Hadamard product of matrices, Pages 99-106
Shencan Chen
The number of nonconstant invariant polynomials of matrices with several prescribed blocks, Pages 107-116
Gloria Cravo and Fernando C. Silva
Enumeration of orbits on cycles for linear and affine groups, Pages 117-127
Daniele A. Gewurz
Null spaces of correlation matrices, Pages 129-157
Wayne Barrett and Stephen Pierce
Total dilations, Pages 159-169
Jean-Christophe Bourin
The doubly graded matrix cone and Ferrers matrices, Pages 171-190
Geir Dahl
On semigroups of normal matrices, Pages 191-195
Bojana Zalar
The dynamic feedback equivalence over principal ideal domains, Pages 197-208
Jose A. Hermida-Alonso and M. T. Trobajo
The edge-isoperimetric problem on the 600-vertex regular solid, Pages 209-228
L. H. Harper and D. Dreier
Additive mappings on von Neumann algebras preserving absolute values, Pages 229-241
M. Radjabalipour
Lattices generated by orbits of subspaces under finite singular unitary group and its characteristic polynomials, Pages 243-268
You Gao
An improved upper bound for Laplacian graph eigenvalues, Pages 269-278
Kinkar ch. Das
A Schur complement approach to a general extrapolation algorithm, Pages 279-301
C. Brezinski and M. Redivo Zaglia
Positive definite Hankel matrices of minimal condition, Pages 303-314
J. M. Varah
Partitioning the edge set of a bipartite graph into chain packings: complexity of some variations, Pages 315-327
D. de Werra
Effect of linear perturbation on spectra of matrices, Pages 329-342
R. Alam and S. Bora
The continuous-time Rayleigh quotient flow on the sphere, Pages 343-357
R. Mahony and P. -A. Absil
Finite Blaschke products of contractions, Pages 359-370
Hwa-Long Gau and Pei Yuan Wu
Asymptotic similarity-preserving linear maps on , Pages 371-378
Guoxing Ji
On the Laplacian spectral radius of a tree, Pages 379-385
Ji-Ming Guo
Author index, Pages 387-388
Lists of Editors, Pages ii-iii
From owner-reliable_computing [at] interval [dot] louisiana.edu Tue Jun 3 10:25:47 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h53FPkJ05199
for reliable_computing-outgoing; Tue, 3 Jun 2003 10:25:46 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h53FPfG05194
for reliable_computing [at] interval [dot] louisiana.edu; Tue, 3 Jun 2003 10:25:42 -0500 (CDT)
Received: from smtp018.mail.yahoo.com (smtp018.mail.yahoo.com [216.136.174.115])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with SMTP id h4VFHkA00471
for ; Sat, 31 May 2003 10:17:47 -0500 (CDT)
Received: from do3t02.roma1.infn.it (HELO do3t02) (vsacksteder4 [at] 141 [dot] 108.19.100 with login)
by smtp.mail.vip.sc5.yahoo.com with SMTP; 31 May 2003 15:17:41 -0000
Reply-To:
From: "Vincent Sacksteder"
To: "'Vladik Kreinovich'" ,
Subject: RE: from NA Digest
Date: Sat, 31 May 2003 17:19:34 +0200
Message-ID: <000d01c32788$0b255f60$64136c8d@do3t02>
MIME-Version: 1.0
Content-Type: text/plain;
charset="us-ascii"
Content-Transfer-Encoding: 7bit
X-Priority: 3 (Normal)
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook CWS, Build 9.0.2416 (9.0.2910.0)
X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2800.1106
In-Reply-To: <200304140156.h3E1uGZ03342 [at] cs [dot] utep.edu>
Importance: Normal
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Dear Interval list:
This mail is in response to your mails in mid-April, which were responding
to my request for information about reliability issues in scientific
calculations. I thank Vladik Kreinovich for forwarding my request to your
group, and I thank many of you who responded with offers of resources and
counsel.
Over the last month and a half I have been writing a paper on certain risks
that the use of computing creates for physics. As part of my research for
the paper, I took the time to read through a good number of interval papers
and resources and thus educate myself about interval analysis. I just
finished a rough draft of the paper, and if any of you are interested in
reviewing it I can send you a copy. Now that I've completed this stage, I
think I can finally respond to your welcoming mails.
Firstly, I concluded that interval analysis holds promise for helping with
certain problems in physical computations. But I'm not sure exactly where
to expect that it will be helpful. Below the dashed line in this mail, I
have included a couple of pages of speculations about this issue, with the
hope that some of you can add to my speculations or, if necessary, correct
them.
Secondly, I concluded that the interval analysis will help with only a
subset of the reliability issues in physics computing. There are so many
problems with computing in general, even when floating point is not
involved. These are often discussed under the rubric of "bugs," and many
many best practices have been developed for understanding them, avoiding
their creation, and fixing them. We can expect that all those problems will
continue to apply in physics calculations, and that floating point issues
will be just one more of the many possible failure modes.
Thirdly, I notice that your journal is titled Reliable Computing,
suggesting that your community is interested in ensuring reliability in
general. I would therefore suggest addressing the larger issues that can
compromise a scientific calculation. For instance, given the result of an
interval computation, how can one be sure that the computation was correct?
When using computers, one can expect bugs in any program, and more bugs if
the program is more complex. Given that interval algorithms are often more
complex than non-interval algorithms, one can expect an increased likelihood
of bugs causing incorrect results. If your community is working toward the
creation of highly trustworthy computing libraries, it will need to not just
rely on the concepts of intervals, but also make heavy use of the full
richness of information science's resources and best practices for ensuring
reliability.
For more details about the full set of computing difficulties that can
afflict a physics calculation, and about some best practices and resources
for managing those risks, my paper might make a good starting point.
However there is also a lot of other literature around too.
Lastly, I want to note two communications that discussed certain challenges
that (they claim) need to be addressed in order to make interval results
more useful to the scientific community. These were Neumaier's paper "Grand
Challenges and Scientific Standards in Interval Analysis," in Reliable
Computing 8, pg. 313, and Marek Gutowski's mail to this list on April 14th.
Both made a lot of sense to me, though I'm a newcomer to intervals.
Thanks again for your helpfulness,
Vincent Sacksteder
---------------------------
Speculations about when Interval Analysis would be useful:
I remain a bit confused about exactly when interval analysis would be useful
for physics calculations.
When and how do you arrive at certainty that some measurable quantity is
within a specific interval?
Physically speaking, is this possible?
Much of my confusion is connected to the philosophical problem of
observation and uncertainty. Descartes raised a problem of how we can know
that anything outside of ourselves is happening, is real; he imagined, as a
thought experiment, some being which deceived him about everything he
experienced. This worry is arguably a sign of some psychological
disturbance in Descartes, and yet it is a prominent motif in modern society
at both intellectual and popular levels, and must therefore be taken
seriously.
Scientists tend to cut through the Gordian knot by simply asserting that
there is no such deception and that what we observe is real; i.e. they
assume the existence of an external and independent reality and of our
ability to observe that reality, to reproduce our observations, to record
them, and to report them to others. Nonreproducible behavior - i.e.
identical observations which obtain different results - is typically called
uncertainty, even though there is nothing uncertain about what was actually
observed: measument 1 gave me a value of 9.0 while measurement 2 gave me a
value of 10.0. In the face of irreproducible results (uncertainty), the
typical response is to go up one level of abstraction from the individual
measurement: i.e. to claim that one's measurements are governed by a
probability distribution, that the probability distribution is an observable
in the same sense as one's individual measurements, and then to aim at
measuring the probability distribution.
The approach just outlined (I will call it the "traditional" one) is a bit
simplistic in its blunt assertion that we observe that we observe, with no
qualifiers. Large realms of difficulty - those connected to Descartes'
dilemma - are simply discounted and left undiscussed. Yet even with this
simplification, the traditional approach is obviously open to a lot of
philosophical dilemmas which have been the subject of a rich and continueing
discussion.
I have reviewed this briefly in order to point out that one can expect that
more subtle approaches (including interval analysis) will return us to the
territory of Descartes, and therefore will be much more difficult to
understand and manage than the traditional approach. Very hard questions
become extremely easy to formulate: Is it meaningful to claim at the same
time both that you observed something and that you're not sure what you
observed? How can you be sure that what you observed was within a specific
interval? How can you be sure of the interval? What is observation anyway?
These questions, and many others, become very hard to resolve once we give
up the traditional idea that observation is inherently simple.
Yet physicists have been required to think carefully about observation. In
particular, quantum mechanics and statistical mechanics have required
physicists to think very carefully about observation. They began to think a
lot about quantized observables, which can have only a finite number of
values. And they developed the idea of probability distributions to a fine
art. As a result, the philosophical waters have become very muddied, and
debate continues about what observation means, what it is we are observing,
etc. This is often discussed under the rubric of "the
Einstein-Podolsky-Rosen paradox," Bell's theorem, and Schrodinger's cat.
The simplest approach is to baldly state that some things are not physically
meaningful, that other things are, and that we must therefore wear blinders
when we do science. This is exemplified to the Copenhagen approach to
quantum mechanics, which says that it is not meaningful to discuss what the
universe is doing when we are not observing. End of story. Along with this
comes an assertion that the real goal of science is to construct a useful
model of reality, and to re-evaluate physics as a model rather than as an
understanding of reality.
Perhaps this is the best way of approaching interval analysis: as simply a
possibly useful way of modelling reality, another tool in our mathematical
toolbox which may be more or less useful in different situations.
Another, complementary approach, could be to consider interval analysis as
simply another (sub?)field of mathematics, with a defining set of axioms.
Where differential geometry starts by allowing straight lines to have more t
han one intersection, interval analysis starts by taking intervals as the
object of numerical analysis. The resulting theorems can be of immediate
use to a traditional approach to science: given precisely known inputs
(without intervals) to a formula, one may cope with rounding errors by
computing an interval within which the exact result must lie.
Returning to interval analysis as a way of modelling physical reality, I now
ask when it would be useful. I don't have all the answers, and would
appreciate any input you have on this.
First, if something is known on intellectual grounds to be within an
interval, I wonder if real numbers are the correct data type to be using.
In general, if something is known to be bounded, one uses a construct which
inherently expresses that bound. In other words, one uses groups. A
variable that is both continuous and bounded is generally formulated as an
element of a Lie group. The simplest example is geometric angles, which are
members of the U(1) Lie group, which is compact and thus one can take x = (x
+ 2 pi). Another example of a continuous and bounded variable is relative
speeds, which are known to be bounded by the speed of light. It is no
accident that the theory of special relativity is founded on the assertion
that coordinate transformations are members of a Lie group, the Poincare
algebra. In contrast to U(1), the Poincare algebra is not compact, so one
can't use the U(1) trick of angle equivalence. Instead it is often
appropriate to decide to think of relative boosts between coordinate systems
rather than relative velocities. There is a simple mapping between the two,
and boosts are not bounded. This an example of a general phenomenon in
physics: if a variable has a bound, you are often using the wrong variable.
Secondly, bounded quantities are often best modeled by discrete variables.
The simplest example is a variable with only two possible values: one and
minus one. Much of modern physics, in particular statistical mechanics and
quantum physics, is best formulated in terms of discrete variables and Lie
groups. Taking the continuum limit to obtain real numbers is very tricky.
Although a lot of interesting physics is hidden in this continuum limit, it
definitely hard going, and generally impossible to do with mathematical
rigor.
(I would like to note in passing that statistical mechanics and quantum
mechanics address many of the same problems as interval analysis (but in a
more flexible way) and have been extremely well developed. In particular,
statistical mechanics can reproduce (barring NA issues) the result of any
interval analysis, but tends to provide much more detailed results and to
require a lot more resources. Where interval analysis tracks an interval
through a calculation, statistical mechanics tracks an entire probability
distribution. I suspect that there is a lot of room for collaboration
between the two fields.)
Given these two caveats, I suspect that interval analysis may nonetheless be
useful for certain physical problems. I already mentioned its ability to
compensate for rounding errors. I believe it can also be useful for
analyzing confidence limits and percentiles. Given an input A that we're
not absolutely sure of, we can make several hypothetical cases about an
output B, in the form of a table:
1. If A were within interval A1, then B would be within interval B1.
2. If A were within interval A2, then B would be within interval B2.
And so on.
Given our uncertain ideas about what value A probably has, we can arrive at
a picture of how much we know about B. As such, interval analysis would be
naturally classified as a subfield of the field of uncertainty
quantification, which is devoted to understanding the uncertainties in
scientific research. I give a number of references to this field in my
paper; or just look at the web site of the Foundations 02 web site, which
contains a review paper by Easterling et al. on uncertainty quantification.
I have not yet thought of other possible physical uses for interval
analysis, and would appreciate hearing ideas about this.
Thanks again,
Vincent Sacksteder
-----Original Message-----
From: Vladik Kreinovich [mailto:vladik [at] cs [dot] utep.edu]
Sent: Monday, April 14, 2003 3:56 AM
To: reliable_computing [at] interval [dot] louisiana.edu
Cc: vincent [at] sacksteder [dot] com
Subject: from NA Digest
Dear Vincent,
There is an additional aspect of why the results of scientific computations
are
sometimes unreliable: programmers and algorithm designers do not take into
consideration that the input values come from measurements and are therefore
known only with a certain accuracy, and that computer operations are not
precise because of rounding. I am sending a copy of your message to NA
Digest
to the interval computations mailing list, many of our folks have
accumulated
results in which people do not take into considation and get wrong results,
as
well as cases when everything is done perfectly.
Vladik
From: Vincent Sacksteder
Date: Tue, 8 Apr 2003 18:38:37 +0200
Subject: Looking for Data About the Reliability of Scientific Calculations
Dear NA community:
I am researching to what extent the numerical results published in the
scientific literature can be regarded as reliable, and am writing you to ask
for any data, experience, and opinions you have on this issue. I am
currently pursueing a Ph.D. in physics after a career in computer science
which focused on the reliability of distributed middleware used by large
enterprises. In my new shoes as a physicist I am confused by the lack of
discussion within the physics community about bugs and about ways of
ensuring the reliability of published numerical results. It seems that
while many physics articles use software to compute various results, perhaps
few authors have implemented the most basic practices for ensuring its
quality - whether planned and repeatable test suites, source code control,
or publication of their code, scripts, and configuration files. (Even when
an author uses lapack or mathematica which are themselves tested, the code,
scripts, and configuration files written by the author may not be tested,
archived, or published.) Moreover, there does not appear to be a structure
for reporting bugs, documenting them, or discussing their prevention. It's
not clear to me how much this is specific to the physics community, or
instead diffused throughout the scientific community.
Perhaps there are some mitigating factors which allow the physics
community
to do without these basic practices: perhaps it is more naturally
self-correcting, through the mutual review of many colleagues. Or perhaps
there is an alternative, informal set of practices which are passed along by
word of mouth. Et cetera.
Unfortunately, I have very little data, other than a documentable lack of
discussion of these issues within the physics literature, and some
individual conversations with my colleagues. If any of you has any
additional data, opinions, or experience to share with me, I would really
really appreciate it.
Thank you,
Vincent Sacksteder
From owner-reliable_computing [at] interval [dot] louisiana.edu Tue Jun 3 11:10:49 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h53GAnl05468
for reliable_computing-outgoing; Tue, 3 Jun 2003 11:10:49 -0500 (CDT)
Received: from zmit1.ippt.gov.pl (zmit1.ippt.gov.pl [148.81.53.8])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h53GAhA05464
for ; Tue, 3 Jun 2003 11:10:44 -0500 (CDT)
Received: (from zkulpa@localhost) by zmit1.ippt.gov.pl (8.8.5/8.7.3-zmit) id SAA05286; Tue, 3 Jun 2003 18:09:37 +0200 (MET DST)
Date: Tue, 3 Jun 2003 18:09:37 +0200 (MET DST)
From: Zenon Kulpa
Message-Id: <200306031609.SAA05286 [at] zmit1 [dot] ippt.gov.pl>
To: reliable_computing [at] interval [dot] louisiana.edu
Subject: RE: from NA Digest
Cc: vincent [at] sacksteder [dot] com
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
> From owner-reliable_computing [at] interval [dot] louisiana.edu Tue Jun 3 17:27:23 2003
> From: "Vincent Sacksteder"
>
[...]
> First, if something is known on intellectual grounds to be
> within an interval, I wonder if real numbers are the correct
> data type to be using. In general, if something is known
> to be bounded, one uses a construct which inherently expresses
> that bound. In other words, one uses groups. A variable that
> is both continuous and bounded is generally formulated as an
> element of a Lie group. The simplest example is geometric angles,
> which are members of the U(1) Lie group, which is compact
> and thus one can take x = (x + 2 pi). Another example
> of a continuous and bounded variable is relative speeds,
> which are known to be bounded by the speed of light.
> It is no accident that the theory of special relativity
> is founded on the assertion that coordinate transformations
> are members of a Lie group, the Poincare algebra. In contrast
> to U(1), the Poincare algebra is not compact, so one can't use
> the U(1) trick of angle equivalence. Instead it is often
> appropriate to decide to think of relative boosts between
> coordinate systems rather than relative velocities.
> There is a simple mapping between the two, and boosts are
> not bounded. This an example of a general phenomenon in
> physics: if a variable has a bound, you are often using
> the wrong variable.
>
> Secondly, bounded quantities are often best modeled
> by discrete variables. The simplest example is a variable
> with only two possible values: one and minus one. Much
> of modern physics, in particular statistical mechanics and
> quantum physics, is best formulated in terms of discrete
> variables and Lie groups. Taking the continuum limit
> to obtain real numbers is very tricky. Although a lot
> of interesting physics is hidden in this continuum limit,
> it definitely hard going, and generally impossible
> to do with mathematical rigor.
>
These are very interesting observations for me.
Can you elaborate and recommend some basic reading
on these subjects?
> (I would like to note in passing that statistical mechanics and quantum
> mechanics address many of the same problems as interval analysis (but in a
> more flexible way) and have been extremely well developed. In particular,
> statistical mechanics can reproduce (barring NA issues) the result of any
> interval analysis, but tends to provide much more detailed results and to
> require a lot more resources. Where interval analysis tracks an interval
> through a calculation, statistical mechanics tracks an entire probability
> distribution.
>
Exactly. Interval analysis is just much simpler (though not
trivially simple...), being at the same time quite satisfactory
for solving problems for which there were available before only
the tools based on statistical analysis.
> I suspect that there is a lot of room for collaboration
> between the two fields.)
>
There is. Some mathematical parallels are also investigated,
see e.g. R. Alt & S. Markov, "On the algebraic properties
of stochastic arithmetic: Comparison to interval arithmetic",
In: W. Kraemer, J. Wolff von Gudenberg, eds.,
Scientific Computing, Validated Numerics, Interval Methods.
Kluwer Academic/Plenum Publishers, New York 2001, 331-341.
[...]
> As such, interval analysis would be naturally classified
> as a subfield of the field of uncertainty quantification,
>
Exactly. This is what the interval analysis actually is.
-- Zenon Kulpa
From owner-reliable_computing [at] interval [dot] louisiana.edu Thu Jun 5 10:14:21 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h55FEKu07768
for reliable_computing-outgoing; Thu, 5 Jun 2003 10:14:20 -0500 (CDT)
Received: from mail.your.biglobe.net (mdsv0121.ht.necidc.net [202.225.207.150])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with SMTP id h55FECA07764
for ; Thu, 5 Jun 2003 10:14:14 -0500 (CDT)
Received: (biglobe-qmail 29691 invoked by uid 0); 6 Jun 2003 00:13:49 +0900
Received: from waseda.jp by biglobe-qmail with SMTP; 6 Jun 2003 00:13:49 +0900
Date: Fri, 6 Jun 2003 00:14:14 +0900
Subject: Announcement of Workshop related to Numerical Verification
Content-Type: text/plain; charset=US-ASCII; format=flowed
Mime-Version: 1.0 (Apple Message framework v552)
Cc: "Shin'ichi OISHI"
To: reliable_computing [at] interval [dot] louisiana.edu
From: "Shin'ichi OISHI"
Content-Transfer-Encoding: 7bit
Message-Id: <5DFF404A-9768-11D7-A6F4-000393D4DC48 [at] waseda [dot] jp>
X-Mailer: Apple Mail (2.552)
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Dear Researchers in Numerical Verification,
The following is an announcement of the workshop related to Numerical
verification which will be held at Waseda University, Tokyo Japan.
Borrowing the mailing list, I would like to inform this. If you have
an interest and have a chance to come, we are very welcome to invite
you to attend.
Sincerely yours,
Shin'ichi OISHI
Professor
Waseda University, Tokyo, Japan
------------------------Program from
here------------------------------------------------------
Waseda University 21-st Century COE Productive ICT Academia Program
"Workshop on Scalable and Accurate Numerical Computation"
Date: June 17 13:00-17:00
Place: Meeting Room 1, 62-Building, School of Science and Engineering,
Waseda University
3-4-1 Okubo Shinjuku-ku, Tokyo 169-8555, Japan
1:00-1:10 Yoichi Muraoka (Waseda university), Greeting from Program
Leader
1:10-1:40 Shin'ichi Oishi (Waseda university), Accurate Sum and Dot
Product I -- Dekker's Theorem and Related Topics (Introduction)
1:40-2:10 Takeshi Ogita (Waseda university), Accurate Sum and Dot
Product II --Algorithm and Experiments
2:20-3:20 Siegfried M. Rump (Inst. f. Computer Science III Technical
University Hamburg-Harburg), Accurate Sum and Dot Product III --
Apriori Estimates
3:20-3:40 Masao Iri, Short Comments
3:50-4:30 Kunio Tanabe (The Institute of Statistical Mathematics), A
Successive Rank One Modification Method for Solving Linear equations
4:30-5:00 Tetsuro Yamamoto (Waseda university), Second-order accuracy
of finite difference method using not necessarily uniform nodes for
semilinear ordinary differential equations subject to two-point mixed
boundary conditions
From owner-reliable_computing [at] interval [dot] louisiana.edu Fri Jun 6 09:55:48 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h56Etmb09010
for reliable_computing-outgoing; Fri, 6 Jun 2003 09:55:48 -0500 (CDT)
Received: from web14202.mail.yahoo.com (web14202.mail.yahoo.com [216.136.172.144])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with SMTP id h56EtgA09006
for ; Fri, 6 Jun 2003 09:55:42 -0500 (CDT)
Message-ID: <20030606145536.52691.qmail [at] web14202 [dot] mail.yahoo.com>
Received: from [213.172.0.6] by web14202.mail.yahoo.com via HTTP; Fri, 06 Jun 2003 07:55:36 PDT
Date: Fri, 6 Jun 2003 07:55:36 -0700 (PDT)
From: Slava Nesterov
Subject: Reliable Computing, vol.9, issue 5, 2003
To: reliable_computing [at] interval [dot] louisiana.edu
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Reliable Computing
Volume 9, issue 5, 2003
Special Issue
Proceedings of the Validated Computing 2002
conference
May 23-25, 2002, Toronto, Canada
Guest Editor: R. Baker Kearfott
Preface
315
Interval Global Optimization in Solvent Design
Luke E. K. Achenie, Manish Sinha
317-338
Extended Interval Power Function
Walter Kraemer, Juergen Wolff von Gudenberg
339-347
Are There Easy-to-Check Necessary and Sufficient
Conditions for Straightforward Interval Computations
To Be Exact?
Vladik Kreinovich, Luc Longpre, James J. Buckley
349-358
Some Computer Assisted Proofs for Solutions of
the Heat Convection Problems
Mitsuhiro T. Nakao, Yoshitaka Watanabe,
Nobito Yamamoto, Takaaki Nishida
359-372
Reliable Computation of Frequency Response Plots
for Nonrational Transfer Functions to
Prescribed Accuracy
Paluri S. V. Nataraj, Jayesh J. Barve
373-389
Accelerated Enclosure Methods for Ordinary Free
Boundary Problems
Uwe Schaefer
391-403
Information
Interval-Related Talks at the 2003 IEEE International
Conference on Fuzzy Systems
405-406
__________________________________
Do you Yahoo!?
Yahoo! Calendar - Free online calendar with sync to Outlook(TM).
http://calendar.yahoo.com
From owner-reliable_computing [at] interval [dot] louisiana.edu Fri Jun 6 12:31:16 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h56HVGr09312
for reliable_computing-outgoing; Fri, 6 Jun 2003 12:31:16 -0500 (CDT)
Received: from loretta.fzi.de (loretta.fzi.de [141.21.6.197])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h56HVBA09308
for ; Fri, 6 Jun 2003 12:31:11 -0500 (CDT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Subject: Condition of polynomials
Date: Fri, 6 Jun 2003 19:31:10 +0200
Message-ID: <878F400487275A40AC4104439D3CBD8F219108 [at] loretta [dot] fzi.de>
X-MS-Has-Attach:
X-MS-TNEF-Correlator:
Thread-Topic: from NA Digest
Thread-Index: AcMp5TcT0XrxLWYUQ1y3O/IuyEgeYIjzc0MQ
From: "Armin Bantle"
To:
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by interval.louisiana.edu id h56HVCA09309
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Dear all,
there are multiple ways to evaluate a polynomial for a given value, e.g.
Horner method, polynomial basis transformation (power basis -> taylor basis,
chebyshev basis, bernstein basis, ...), optimized forms, etc. They all
differ in their sensitivity with respect to rounding errors in intermediate
computations.
My question: Is there a method to determine an optimal form, i.e. a form
that is more stable than any other form?
Any information related to this question and the condition of polynomials in
general would be very valuable for me.
Thank you very much in advance.
Regards,
Armin Bantle
_________________________________________________________________
Dipl.-Math. Armin Bantle
Computer Science Research Center (FZI) phone +49-721-9654-316
Mobility Management and Robotics fax +49-721-9654-317
Haid-und-Neu-Str. 10-14 e-mail bantle [at] fzi [dot] de
D-76131 Karlsruhe (Germany) http://www.fzi.de
_________________________________________________________________
From owner-reliable_computing [at] interval [dot] louisiana.edu Fri Jun 6 13:44:56 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h56Iiu509473
for reliable_computing-outgoing; Fri, 6 Jun 2003 13:44:56 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h56IirZ09468
for reliable_computing [at] interval [dot] louisiana.edu; Fri, 6 Jun 2003 13:44:53 -0500 (CDT)
Received: from loretta.fzi.de (loretta.fzi.de [141.21.6.197])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h56FcqA09171
for ; Fri, 6 Jun 2003 10:38:52 -0500 (CDT)
X-MimeOLE: Produced By Microsoft Exchange V6.0.6249.0
content-class: urn:content-classes:message
MIME-Version: 1.0
Content-Type: text/plain;
charset="iso-8859-1"
Subject: Condition of polynomials
Date: Fri, 6 Jun 2003 17:38:50 +0200
Message-ID: <878F400487275A40AC4104439D3CBD8F435B2C [at] loretta [dot] fzi.de>
X-MS-Has-Attach:
X-MS-TNEF-Correlator:
Thread-Topic: from NA Digest
Thread-Index: AcMp5TcT0XrxLWYUQ1y3O/IuyEgeYIjzc0MQ
From: "Armin Bantle"
To:
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by interval.louisiana.edu id h56FcrA09172
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Dear all,
there are multiple ways to evaluate a polynomial for a given value, e.g.
Horner method, polynomial basis transformation (power basis -> taylor basis,
chebyshev basis, bernstein basis, ...), optimized forms, etc. They all
differ in their sensitivity with respect to rounding errors in intermediate
computations.
My question: Is there a method to determine an optimal form, i.e. a form
that is more stable than any other form?
Any information related to this question and the condition of polynomials in
general would be very valuable for me.
Thank you very much in advance.
Regards,
Armin Bantle
_________________________________________________________________
Dipl.-Math. Armin Bantle
Computer Science Research Center (FZI) phone +49-721-9654-316
Mobility Management and Robotics fax +49-721-9654-317
Haid-und-Neu-Str. 10-14 e-mail bantle [at] fzi [dot] de
D-76131 Karlsruhe (Germany) http://www.fzi.de
_________________________________________________________________
From owner-reliable_computing [at] interval [dot] louisiana.edu Thu Jun 12 08:43:19 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5CDhJw16797
for reliable_computing-outgoing; Thu, 12 Jun 2003 08:43:19 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5CDhE616792
for reliable_computing [at] interval [dot] louisiana.edu; Thu, 12 Jun 2003 08:43:14 -0500 (CDT)
Received: from liasun13.epfl.ch (liasun13.epfl.ch [128.178.155.37])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5CBkTA16677
for ; Thu, 12 Jun 2003 06:46:30 -0500 (CDT)
Received: from epfl.ch (lias7.epfl.ch [128.178.155.21])
by liasun13.epfl.ch (8.8.X/EPFL-8.1a) with ESMTP id NAA02772;
Thu, 12 Jun 2003 13:19:11 +0200 (MET DST)
Message-ID: <3EE8620F.4070101 [at] epfl [dot] ch>
Date: Thu, 12 Jun 2003 13:20:47 +0200
From: Christophe Jermann
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20030208 Netscape/7.02
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: comp.constraints,sci.math.num-analysis,sci.nonlinear,sci.op-research,news.announce.conferences
To: IntCP2003
Subject: [CFP] IntCP2003: 1st international workshop on Interval Analysis
and Constraint Propagation for Applications
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
(we apologize for possible multiple reception of this call)
=================================================================
CALL FOR PAPERS
IntCP 2003 workshop
Interval Analysis and Constraint Propagation for Applications
Actons Hotel, Kinsale,
County Cork, Ireland
29th September 2003
Held in conjunction with the
Ninth International Conference on Principles
and Practice of Constraint Programming (CP 2003)
=================================================================
* Important Dates:
------------------
01 Jul 2003 - Submission deadline
31 Jul 2003 - Notification of acceptance
15 Aug 2003 - Final camera-ready copies
29 Sep 2003 - Workshop day
* Description and goals:
------------------------
Many practical problems involve numerical constraints as an essential
component. While constraint propagation solvers have proven particularly
efficient in solving challenging instances of numerical problems with
nonlinear constraints, they do not yet have enough appeal in many
practical problem areas.
The goal of this workshop is to bring together researchers and
practitioners working on constraint propagation and interval analysis,
giving the opportunity to promote presentation and discussion on ongoing
work on techniques for real-world requirements.
For example, many practical problems often have a continuum of solutions
which express a spectrum of equally relevant choices, as the possible
moving areas of a mobile robot, the collision regions between objects in
mechanical assembly, or different alternatives of shapes for the
components
of a kinematic chain. These alternatives need to be identified using
complete
solving techniques. Completeness means the ability to find all solutions
if
any, or else to prove that there are no solutions to the problem.
Constraint propagation solvers, although complete, provide enclosures
that
are either prohibitively verbose or poorly informative. In contrast, a
number of interval analysis approaches have been developed, notably in
the
area of robust control, estimation, and robotics, which can
significantly
enhance the capabilities of interval-based solvers. They consist in
covering the spectrum of solutions using a reduced number of subsets
of R. Usually, these subsets are chosen with known and simple properties
(interval boxes, polytopes, ellipsoids,...).
Other questions that are often relevant in applications include, but are
by no means restricted to:
- uncertainty that can, for example, be modeled, by logical quantifiers,
- specific problem structure, for example in the case of discrete time,
continuous state systems,
- mixture of discrete and continuous problem variables,
- inequality constraints,
- problems with a huge number of discrete solutions.
We seek contributions that address such questions, and present relevant
software tools, algorithms, theoretical results, or applications of
dedicated
techniques to real-world problems.
* Workshop format:
------------------
This is a half-day workshop, with open attendance. Its aim is to provide
a forum where researchers currently working in this area can discuss
their
most recent ideas and developments and think together about the most
promising
new directions. We particularly encourage the presentation of work that
bridge
the gap between theory and practice.
* Submissions:
--------------
People wishing to give a talk should submit an extended abstract of at
least
2 pages. Submissions must be formatted using LNCS packages (see CP
formatting
instructions). The title page should include the name, address,
telephone
number and electronic mailing address for each author. Please, email all
submissions in postscript or pdf format to intcp03 [at] epfl [dot] ch by July 1st,
2003, specifying the name of the contact author in the message.
* Reviewing process:
--------------------
Submissions will be reviewed by at least one committee member, and will
be
selected on the basis of their contribution to the topic of the
workshop.
Authors will receive feedback in the form of reviewers' comments.
* Accomodation/Registration:
----------------------------
Accomodation is provided by the hosting conference CP 2003. All workshop
attendees must pay the CP 2003 regular registration fee in addition
to the workshop fee. At least one author of each accepted submission
must
attend the workshop.
* Committee:
------------
- Frédéric Goualard, Computer Science Research Institute (IRIN),
University of Nantes, France
- Luc Jaulin, Laboratoire d'Ingénierie des Systèmes Automatisés (LISA),
University of Angers, France
- Christophe Jermann, Artificial Intelligence Laboratory (LIA),
Swiss Federal Institute of Technology (EPFL), Switzerland
- Stefan Ratschan, Max-Planck-Institut für Informatik,
Saarbrücken, Germany
- Djamila Sam-Haroud, Artificial Intelligence Laboratory (LIA),
Swiss Federal Institute of Technology (EPFL), Switzerland
* Contacts:
-----------
Send questions about the conference to intcp03 [at] epfl [dot] ch
From owner-reliable_computing [at] interval [dot] louisiana.edu Mon Jun 16 17:42:16 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5GMgFn22544
for reliable_computing-outgoing; Mon, 16 Jun 2003 17:42:15 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5GMgCd22539
for reliable_computing [at] interval [dot] louisiana.edu; Mon, 16 Jun 2003 17:42:12 -0500 (CDT)
Received: from imf24aec.bellsouth.net (mail004.mail.bellsouth.net [205.152.58.24])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5GMbGA22503
for ; Mon, 16 Jun 2003 17:37:17 -0500 (CDT)
Received: from Inspiron-8200 ([65.83.167.98]) by imf24aec.bellsouth.net
(InterMail vM.5.01.05.27 201-253-122-126-127-20021220) with SMTP
id <20030616223707.GXRG16268.imf24aec.bellsouth.net@Inspiron-8200>
for ;
Mon, 16 Jun 2003 18:37:07 -0400
Message-Id: <2.2.32.20030616223653.00a0aef8 [at] pop [dot] louisiana.edu>
X-Sender: rbk5287 [at] pop [dot] louisiana.edu
X-Mailer: Windows Eudora Pro Version 2.2 (32)
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Date: Mon, 16 Jun 2003 17:36:53 -0500
To: reliable_computing [at] interval [dot] louisiana.edu
From: "R. Baker Kearfott"
Subject: Thank you: reliable_computing address verification
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Colleagues,
Thank you for your cooperation in the recent semi-annual mailing list
verification process.
As usual, number of addresses are no longer valid.
I list these below. If you know correct addresses (or other information)
for a colleague whose old address is listed below, please inform me.
I'll check and add the correct address as appropriate.
Best regards,
Baker
=============================================================================
Subscription verification started Tuesday, June 10.
Host unknown:
jcf [at] posso [dot] lip6.fr
ernst [at] eiunix [dot] tuwien.ac.at
Barth [at] eiunix [dot] tuwien.ac.at
ratschek [at] convex [dot] rz.uni-duesseldorf.de
kk [at] gcet [dot] ac.in
ielishak [at] me [dot] fau.edu
User unknown:
rnlea@ortech-engr.com
egor_722vev [at] mail [dot] ru
pvalelima [at] hotmail [dot] com
michael [at] dcs [dot] st-and.ac.uk
maw@st-andrews.ac.uk
stein@mpi-magdeburg.mpg.de
rex [at] mathematik [dot] uni-leipzig.de
User quota exceeded:
rakash [at] ee [dot] iitb.ernet.in
Possible mail loop:
mak [at] ee [dot] iitb.ac.in
Message could not be delivered in 3 (or 5) days:
rocha [at] binghamton [dot] edu
No route to host:
leclerc [at] cs [dot] cofc.edu
=============================================================================
---------------------------------------------------------------
R. Baker Kearfott, rbk [at] louisiana [dot] edu (337) 482-5346 (fax)
(337) 482-5270 (work) (337) 981-9744 (home)
URL: http://interval.louisiana.edu/kearfott.html
Department of Mathematics, University of Louisiana at Lafayette
Box 4-1010, Lafayette, LA 70504-1010, USA
---------------------------------------------------------------
From owner-reliable_computing [at] interval [dot] louisiana.edu Fri Jun 20 08:11:05 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5KDB4r27670
for reliable_computing-outgoing; Fri, 20 Jun 2003 08:11:04 -0500 (CDT)
Received: (from rbk5287@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5KDB0K27665
for reliable_computing [at] interval [dot] louisiana.edu; Fri, 20 Jun 2003 08:11:00 -0500 (CDT)
Received: from lcyoung.math.wisc.edu (lcyoung.math.wisc.edu [144.92.166.90])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5JJrvA26757
for ; Thu, 19 Jun 2003 14:53:58 -0500 (CDT)
Received: from ultra7.math.wisc.edu (ultra7.math.wisc.edu [144.92.166.188])
by lcyoung.math.wisc.edu (8.11.6p2/8.11.6) with ESMTP id h5JJq9224688;
Thu, 19 Jun 2003 14:52:09 -0500 (CDT)
Date: Thu, 19 Jun 2003 14:52:08 -0500 (CDT)
From: Hans Schneider
To: NETS -- at-net , E-LETTER ,
Pradeep Misra ,
Shaun Fallat ,
"na.digest" , ipnet-digest [at] math [dot] msu.edu,
SIAGLA-DIGEST , wim@bell-labs.com,
hjt [at] eos [dot] ncsu.edu, SMBnet [at] smb [dot] org, vkm [at] eedsp [dot] gatech.edu,
reliable_computing [at] interval [dot] louisiana.edu
Subject: LAA contents
Message-ID:
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
X-MailScanner: Found to be clean
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
* Linear Algebra and its Applications
Volume 369, Pages 1-352 (1 August 2003)
http://www.sciencedirect.com/science/issue/5653-2003-996309999-435593
TABLE OF CONTENTS
Perturbation of null spaces with application to the eigenvalue problem and generalized inverses, Pages 1-25
Konstantin E. Avrachenkov and Moshe Haviv
A complementary result of Kantorovich type order preserving inequalities by Mii-Peari-Seo, Pages 27-40
Takayuki Furuta and Mariko Giga
Analysis of preconditioning strategies for collocation linear systems, Pages 41-75
Stefano Serra Capizzano and Cristina Tablino Possio
Wielandt and Ky-Fan theorem for matrix pairs, Pages 77-93
Ivica Naki and Kreimir Veseli
Identities of bilinear mappings and graded polynomial identities of matrices, Pages 95-112
Yu. A. Bahturin and V. Drensky
The Schur algorithm for generalized Schur functions III: J-unitary matrix polynomials on the circle, Pages 113-144
Daniel Alpay, Tomas Azizov, Aad Dijksma and Heinz Langer
Linear systems with nilpotent leading term, Pages 145-152
Werner Balser
Error analysis of signal zeros: a projected companion matrix approach, Pages 153-167
F. S. V. Bazan
Existence and construction of nonnegative matrices with prescribed spectrum, Pages 169-184
Ricardo L. Soto
On the solvability of the commutative power-associative nilalgebras of dimension 6, Pages 185-192
Ivan Correa, Irvin Roy Hentzel and Luiz Antonio Peresi
Determinants and multiplicative functionals on quaternion matrices, Pages 193-201
Jiangnan Fan
Stabilizing a class of time delay systems using the Hermite-Biehler theorem, Pages 203-216
Vilma A. Oliveira, Marcelo C. M. Teixeira and Lucia Cossi
Weak majorization inequalities and convex functions, Pages 217-233
Jaspal Singh Aujla and Fernando C. Silva
Lebesgue perturbation of a quasi-definite Hermitian functional. The positive definite case, Pages 235-250
A. Cachafeiro, F. Marcellan and C. Perez
On the critical group of the n-cube, Pages 251-261
Hua Bai
Rank-1 preserving linear maps on nest algebras, Pages 263-277
Jinchuan Hou and Jianlian Cui
Quasi-real normal matrices and eigenvalue pairings, Pages 279-294
Geoffrey R. Goodson, Roger A. Horn and Dennis I. Merino
Multilinear functional inequalities involving permanents, determinants, and other multilinear functions of nonnegative matrices and M-matrices, Pages 295-310
Assaf Goldberger and Michael Neumann
On determinant preserver problems, Pages 311-317
Victor Tan and Fei Wang
The general trapezoidal algorithm for strongly regular max-min matrices, Pages 319-338
Martin Gavalec
Some complete Lie superalgebras, Pages 339-349
Li Yun Wang and Dao Ji Meng
Author index, Pages 351-352
Editorial board, Pages ii-iii
From owner-reliable_computing [at] interval [dot] louisiana.edu Tue Jun 24 11:46:25 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5OGkPC00626
for reliable_computing-outgoing; Tue, 24 Jun 2003 11:46:25 -0500 (CDT)
Received: from liasun13.epfl.ch (liasun13.epfl.ch [128.178.155.37])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5OGkI600622
for ; Tue, 24 Jun 2003 11:46:19 -0500 (CDT)
Received: from epfl.ch (lias7.epfl.ch [128.178.155.21])
by liasun13.epfl.ch (8.8.X/EPFL-8.1a) with ESMTP id OAA29156;
Tue, 24 Jun 2003 14:54:22 +0200 (MET DST)
Message-ID: <3EF87D9F.8000505 [at] epfl [dot] ch>
Date: Tue, 24 Jun 2003 18:34:39 +0200
From: Christophe Jermann
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.0.2) Gecko/20030208 Netscape/7.02
X-Accept-Language: en-us, en
MIME-Version: 1.0
Newsgroups: comp.constraints,sci.math.num-analysis,sci.nonlinear,sci.op-research,news.announce.conferences
To: IntCP2003
Subject: [CFP] IntCP2003: 1st international workshop on Interval Analysis
and Constraint Propagation for Applications
Content-Type: text/plain; charset=iso-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
(we apologize for possible multiple reception of this call)
=================================================================
CALL FOR PAPERS
IntCP 2003 workshop
Interval Analysis and Constraint Propagation for Applications
Actons Hotel, Kinsale,
County Cork, Ireland
29th September 2003
Held in conjunction with the
Ninth International Conference on Principles
and Practice of Constraint Programming (CP 2003)
=================================================================
* Important Dates:
------------------
01 Jul 2003 - Submission deadline
31 Jul 2003 - Notification of acceptance
15 Aug 2003 - Final camera-ready copies
29 Sep 2003 - Workshop day
* Description and goals:
------------------------
Many practical problems involve numerical constraints as an essential
component. While constraint propagation solvers have proven particularly
efficient in solving challenging instances of numerical problems with
nonlinear constraints, they do not yet have enough appeal in many
practical problem areas.
The goal of this workshop is to bring together researchers and
practitioners working on constraint propagation and interval analysis,
giving the opportunity to promote presentation and discussion on ongoing
work on techniques for real-world requirements.
For example, many practical problems often have a continuum of solutions
which express a spectrum of equally relevant choices, as the possible
moving areas of a mobile robot, the collision regions between objects in
mechanical assembly, or different alternatives of shapes for the
components
of a kinematic chain. These alternatives need to be identified using
complete
solving techniques. Completeness means the ability to find all solutions
if
any, or else to prove that there are no solutions to the problem.
Constraint propagation solvers, although complete, provide enclosures
that
are either prohibitively verbose or poorly informative. In contrast, a
number of interval analysis approaches have been developed, notably in
the
area of robust control, estimation, and robotics, which can
significantly
enhance the capabilities of interval-based solvers. They consist in
covering the spectrum of solutions using a reduced number of subsets
of R. Usually, these subsets are chosen with known and simple properties
(interval boxes, polytopes, ellipsoids,...).
Other questions that are often relevant in applications include, but are
by no means restricted to:
- uncertainty that can, for example, be modeled, by logical quantifiers,
- specific problem structure, for example in the case of discrete time,
continuous state systems,
- mixture of discrete and continuous problem variables,
- inequality constraints,
- problems with a huge number of discrete solutions.
We seek contributions that address such questions, and present relevant
software tools, algorithms, theoretical results, or applications of
dedicated
techniques to real-world problems.
* Workshop format:
------------------
This is a half-day workshop, with open attendance. Its aim is to provide
a forum where researchers currently working in this area can discuss
their
most recent ideas and developments and think together about the most
promising
new directions. We particularly encourage the presentation of work that
bridge
the gap between theory and practice.
* Submissions:
--------------
People wishing to give a talk should submit an extended abstract of at
least
2 pages. Submissions must be formatted using LNCS packages (see CP
formatting
instructions). The title page should include the name, address,
telephone
number and electronic mailing address for each author. Please, email all
submissions in postscript or pdf format to intcp03 [at] epfl [dot] ch by July 1st,
2003, specifying the name of the contact author in the message.
* Reviewing process:
--------------------
Submissions will be reviewed by at least one committee member, and will
be
selected on the basis of their contribution to the topic of the
workshop.
Authors will receive feedback in the form of reviewers' comments.
* Accomodation/Registration:
----------------------------
Accomodation is provided by the hosting conference CP 2003. All workshop
attendees must pay the CP 2003 regular registration fee in addition
to the workshop fee. At least one author of each accepted submission
must
attend the workshop.
* Committee:
------------
- Frédéric Goualard, Computer Science Research Institute (IRIN),
University of Nantes, France
- Luc Jaulin, Laboratoire d'Ingénierie des Systèmes Automatisés (LISA),
University of Angers, France
- Christophe Jermann, Artificial Intelligence Laboratory (LIA),
Swiss Federal Institute of Technology (EPFL), Switzerland
- Stefan Ratschan, Max-Planck-Institut für Informatik,
Saarbrücken, Germany
- Djamila Sam-Haroud, Artificial Intelligence Laboratory (LIA),
Swiss Federal Institute of Technology (EPFL), Switzerland
* Contacts:
-----------
Send questions about the conference to intcp03 [at] epfl [dot] ch
From owner-reliable_computing [at] interval [dot] louisiana.edu Thu Jun 26 17:34:53 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5QMYqs06114
for reliable_computing-outgoing; Thu, 26 Jun 2003 17:34:52 -0500 (CDT)
Received: from nwkea-mail-2.sun.com (nwkea-mail-2.sun.com [192.18.42.14])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5QMYk606110
for ; Thu, 26 Jun 2003 17:34:47 -0500 (CDT)
Received: from heliopolis.eng.sun.com ([152.70.1.39])
by nwkea-mail-2.sun.com (8.12.9/8.12.9) with ESMTP id h5QMXwfU006449;
Thu, 26 Jun 2003 15:33:59 -0700 (PDT)
Received: from sun.com (vpn-129-156-96-11.EMEA.Sun.COM [129.156.96.11])
by heliopolis.eng.sun.com (8.11.6+Sun/8.11.6/ENSMAIL,v2.1p1) with ESMTP id h5QMXsg21738;
Thu, 26 Jun 2003 15:33:54 -0700 (PDT)
Message-ID: <3EFB71FE.83C30D79 [at] sun [dot] com>
Date: Thu, 26 Jun 2003 15:21:50 -0700
From: Bill Walster
X-Mailer: Mozilla 4.79 [en] (Win98; U)
X-Accept-Language: en,ru
MIME-Version: 1.0
To: "reliable_computing [at] interval [dot] louisiana.edu"
,
"interval [at] cs [dot] utep.edu"
Subject: [Fwd: Jonathan Schwartz Feedback Panel: Looking for unhappy or
non-customers(Fwd)]
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
Friends,
Here is an opportunity to provide candid
feedback to Sun executives that might make
a difference.
Best regards,
Bill
-------- Original Message --------
Subject: Jonathan Schwartz Feedback Panel: Looking for unhappy or
non-customers(Fwd)
Date: Thu, 26 Jun 2003 09:33:59 -0500 (Central Daylight Time)
From: Conrad Geiger - Sun Academic Region HPC
Technologist
Reply-To: Conrad Geiger - Sun Academic Region HPC
Technologist
To: edu.usa.all [at] sun [dot] com, hpc-aces [at] sun [dot] com, amb.ts [at] sun [dot] com
If you have developer customers or partners unhappy with Sun software
technology who are willing to articulate their reasons for not choosing
Sun software technologies or products, please read...
This is a call for nominations for a feedback panel gathered for
Sun V.P. Jonathan Schwartz that will ultimately speak to Sun software
executives at a Software Staff meeting in August. For example, Sun is
looking for developers who have chosen competing products and technologies
such as .Net, Visual Studio, C#, Visual Basic, Eclipse, Websphere,
Windows Mobile, Apache, Linux etc. The nomination form is attached.
Conrad
>----------------Begin Forwarded Message----------------<
Date: Wed, 25 Jun 2003 15:01:33 -0700
From: Joe Kuriakose
Subject: Request for developer panelist nominations - due July 8
To: sunone-interest [at] sun [dot] com
Hi,
As part of the Voice to the Developer group's ongoing developer
conversations and panels, we are putting together a developer panel
in July titled 'Barriers to Sun adoption'. The panel members from
outside Sun will comprise individual developers who have specifically
not chosen Sun software products or technologies. We expect a
lively discussion on specific developer requirements that are
not being met by Sun products. We need your input on possible
panelists and discussion topics.
Specifically, I would like to receive nominations of individual
developer candidates for this panel. We are looking for individuals
who have chosen competing products and technologies such as
.Net, Visual Studio, C#, Visual Basic, Eclipse, Websphere, Windows
Mobile, Apache, Linux etc.
One of our goals is also to use the July panel as a screening process
to recruit articulate developer panelists who may be invited to a
developer panel planned for Sun software executives at a Software
Staff meeting in August.
** Please Save the Date! **
Developer Council - Barriers to Sun Adoption
Tuesday, July 22 from 11:30 - 1:30
Location: Sun Menlo Park Bldg 18, Pebble Beach
We will share our findings from this panel on our Vodev website.
We will be able to accommodate a small number of in-person Sun
attendees, but a call-in number will also be available. Please
RSVP to dawn.giusti [at] sun [dot] com to get on the notification list,
indicating how you prefer to participate (i.e call-in or in-person).
Please email directly to me the details of the panel candidates you
want to nominate, using the template text below - please fill in the
contact information and as much other information as you know.
The deadline for nominations is Tuesday, July 8.
Thanks,
Joe Kuriakose
DEVELOPER PANEL CANDIDATE NOMINATION
Barriers to Sun Adoption - July 2003
(a) Name:
(b) Title [Architect | VP Engineering | Development Manager |
Group Manager | Senior Developer | Staff Engineer]
(c) Self-described Role: [what do they do on a daily basis;
what are they engaged in now]
(d) Organization:
(e) Phone, Email [and address if available]
(f) Expected segment(s) represented {please circle}:
Corporate [create targeted applications; tools-conscious]
Enterprise [develop multi-tier, corporate-wide apps that "run
the business"; very technical]
Wireless [various platforms, application areas]
Content [web-accessible app development using scripting
languages and authoring tools]
(g) Technologies in use {please circle}: [Solaris | Linux | Websphere |
| Windows .Net | Windows Mobile | Eclipse | VS.net]
{please circle}: [J2EE | J2SE | J2ME | C++/C | C# |
| JSP | ASP | VB | Flash | HTML | PERL | CGI | Other]
(h) Competitor chosen over Sun: [IBM | Microsoft | HP | Dell |
| BEA | Oracle | Qualcomm | Other ]
(i) Known through what connection (e.g., focus group participation,
user group evangelist, partner, customer, former-colleague)
(j) How well known
(k) Sense of presentation skills, articulateness, plausibility
as role-model for the segment, ability to stay on topic,
etc. (scale of 1 to 10, 10 highest)
-----------------------------
Joe Kuriakose
Sun Microsystems
Software Developer Marketing
email: joe.kuriakose [at] sun [dot] com
phone: 650.786.2791
cell: 510.517.0725
>----------------End Forwarded Message----------------<
From owner-reliable_computing [at] interval [dot] louisiana.edu Sat Jun 28 10:58:05 2003
Received: (from daemon@localhost)
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) id h5SFw4p08249
for reliable_computing-outgoing; Sat, 28 Jun 2003 10:58:04 -0500 (CDT)
Received: from cs.utep.edu (mail.cs.utep.edu [129.108.5.3])
by interval.louisiana.edu (8.11.3/8.11.3/ull-interval-math-majordomo-1.3) with ESMTP id h5SFvw608245
for ; Sat, 28 Jun 2003 10:57:58 -0500 (CDT)
Received: from aragorn (aragorn [129.108.5.35])
by cs.utep.edu (8.11.7/8.11.3) with SMTP id h5SFvlg25125;
Sat, 28 Jun 2003 09:57:47 -0600 (MDT)
Message-Id: <200306281557.h5SFvlg25125 [at] cs [dot] utep.edu>
Date: Sat, 28 Jun 2003 09:57:47 -0600 (MDT)
From: Vladik Kreinovich
Reply-To: Vladik Kreinovich
Subject: interval meeting in Novosibirsk, Russia (in Russian)
To: interval [at] cs [dot] utep.edu, reliable_computing [at] interval [dot] louisiana.edu
MIME-Version: 1.0
Content-Type: TEXT/plain; charset=us-ascii
Content-MD5: 2HH/hJFn4I8fmia/0FeH7g==
X-Mailer: dtmail 1.3.0 @(#)CDE Version 1.4 SunOS 5.8 sun4u sparc
Sender: owner-reliable_computing [at] interval [dot] louisiana.edu
Precedence: bulk
FYI: a Russian-language Workshop "Interval Mathematics and Constraint
Propagation" will be held on July 8-9 in Novosibirsk; for details, see
http://www.iis.nsk.su/psi03/meeting/index_r.shtml