From owner-reliable_computing Mon Feb 5 14:07:38 1996 Received: by interval.usl.edu id AA07850 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Mon, 5 Feb 1996 22:07:45 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA07844 (5.65c/IDA-1.4.4 for ); Mon, 5 Feb 1996 22:07:42 -0600 Received: from earth.cs.utep.edu ([129.108.5.14]) by cs.utep.edu (4.1/SMI-4.1) id AA19825; Mon, 5 Feb 96 21:07:38 MST Date: Mon, 5 Feb 96 21:07:38 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602060407.AA19825 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: faculty position in El Paso Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 2588 X-Lines: 48 The University of Texas at El Paso seeks a tenure-track Assistant or Associate Professor of Computer Science. The successful can- didate must demonstrate a strong commitment to undergraduate edu- cation and research experiences. Additionally, the successful candidate is expected to develop a successful research program in a technical area of Computer Science. Although applications are welcome from anyone who holds a Ph.D. in Computer Science, Com- puter Engineering, or Computer Systems Engineering, the depart- ment has particular interest in candidates who specialize in sys- tems or concurrency. UTEP's Computer Science Department is part of the College of En- gineering. It currently consists of seven full time and two part-time faculty members and offers M.S. and B.S. degrees in computer science. Computer Science and Electrical Engineering jointly offer a Ph.D. in computer engineering. Despite its small size, the department maintains significant research programs in Artificial Intelligence and Logic Programming, Cryptography and Theory of Randomness, Interval Computation, and Software En- gineering. More information about the city, the university and the department can be found in the WWW pages (http://www.utep.edu). The University of Texas at El Paso has recently been selected as a Model Institution for Excellence (MIE) by the National Science Foundation. The university's MIE focus is on undergraduate edu- cation and undergraduate involvement in meaningful research ex- periences. Substantial funding is available for the development of new courses and the redesign of appropriate existing courses. Course development is to place emphasis on student participation. The successful applicant is expected to participate in the University's MIE initiative and play an active role in the University's commitment to providing the environment to encourage and enable student success. Due to the MIE involvement, it is expected that the faculty member will successfully employ innova- tive teaching methods and involve students in substantive research activities. Applicants should submit a detailed resume and the names of at least four references to Daniel Cooke, CS Dept., University of Texas at El Paso, El Paso Tx. 79968. (e-mail: dcooke [at] cs [dot] utep.edu). The University of Texas at El Paso does not discriminate on the basis of race, color, national origin, sex, religion, age, or disbility in employment or the provision of services. From owner-reliable_computing Tue Feb 6 09:17:55 1996 Received: by interval.usl.edu id AA10186 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Tue, 6 Feb 1996 17:18:02 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA10180 (5.65c/IDA-1.4.4 for ); Tue, 6 Feb 1996 17:17:57 -0600 Received: from earth.cs.utep.edu ([129.108.5.14]) by cs.utep.edu (4.1/SMI-4.1) id AA01718; Tue, 6 Feb 96 16:17:55 MST Date: Tue, 6 Feb 96 16:17:55 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602062317.AA01718 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: ACM classification system update Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 956 X-Lines: 29 Dear Colleagues, Thanks a lot for your suggestions on the interval-related revision of the American Mathematical Society (AMS) classification scheme. George Corliss, Baker Kearfott, and I have sent the resulting proposal to AMS. Meanwhile, another classification scheme is about to be updated: the ACM Computing Classification Scheme (CCS) used by Computing Reviews and many other publications in the computing field. It will be revised in January 1997, and ACM is asking for suggestions. The latest (1991) version of the CCS is available on the Web at http://www.acm.org/class/1991/ Additional information about this project can be obtained from Carol Wierzbicki, the Managing Editor of Computing Reviews, at wierzbicki [at] acm [dot] org. The current CCS does not mention any interval computations at all, only error analysis as part of Section G.1 Numerical Analysis. Please send all suggestions for revising this scheme to me. Thanks. Vladik From owner-reliable_computing Tue Feb 6 12:15:06 1996 Received: by interval.usl.edu id AA10508 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Tue, 6 Feb 1996 20:44:31 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA10502 (5.65c/IDA-1.4.4 for ); Tue, 6 Feb 1996 20:44:27 -0600 Received: from earth.cs.utep.edu ([129.108.5.14]) by cs.utep.edu (4.1/SMI-4.1) id AA02497; Tue, 6 Feb 96 19:15:06 MST Date: Tue, 6 Feb 96 19:15:06 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602070215.AA02497 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: a new book on applications of interval computations Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1285 X-Lines: 43 Bounding Approaches to System Identification Edited by: Mario Milanese Politecnico di Torino, Italy, John Norton University of Birmingham, England, Helene Piet-Lahanier Office National d'Etudes et de Recherches Aeronautique, Chatillon, France, Eric Walter, CNRS -- Ecole Superieure d'Electricite, Gif-sur-Yvette, France. Plenum Press ISBN: 0-306-45021-6. 557 pp. 1996 Domestic price: $135.00 (US and Canada) Foreign price: $162.00 (elsewhere) Text adoption price on orders of six or more copies: $69.50 each Description: In response to the growing interest in bounding error approaches, the editors of this volume offer the first collection of papers describing advances in techniques and applications of bounding of the parameters or state variables of uncertain dynamical systems. Contributors explore the application of the bounding approach as an alternative to the probabilistic analysis of such systems, relating its importance to robust control-system design. Additional topics include: * modeling and control; * linear and nonlinear systems; * theory, algorithms, and applications; and * Western and Eastern approaches. Subjects: * electrical systems science and control; * communication systems; * applications of mathematics. From owner-reliable_computing Tue Feb 6 14:17:58 1996 Received: by interval.usl.edu id AA10827 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Tue, 6 Feb 1996 22:18:03 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA10821 (5.65c/IDA-1.4.4 for ); Tue, 6 Feb 1996 22:18:00 -0600 Received: from earth.cs.utep.edu ([129.108.5.14]) by cs.utep.edu (4.1/SMI-4.1) id AA03068; Tue, 6 Feb 96 21:17:58 MST Date: Tue, 6 Feb 96 21:17:58 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602070417.AA03068 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: An interval conference in China? Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1966 X-Lines: 52 An interval conference in China? Dear Colleagues, I have just got the following proposal from Dr. Chenyi Hu (Houston, TX): > From: hu [at] happy [dot] dt.uh.edu > > Dear Vladik, > > Just got a fax from China ... > > If the interval community is interested in holding a conference in China, > I can surely make very nice arrangements with reasonable price. I'll be very > happy to organize such meeting in China in 1997 if there is a need. > > Best regards! > > Chenyi China is a great place to go, and having a conference there will definitely boost the local interest in interval computations. R. Baker Kearfott suggested from his organizational experience that it may be somewhat late to organize a big International conference for 1997, especially since some potential participants may have already made their plans. To make a decision, it would be nice to get some idea of who would come, and if there are any conflicting conferences people know about. The only thing that I know is the workshop on computational complexity of interval computations and related topics that I planned to organize in El Paso in May 1997 as a satellite to ACM STOC (Symposium on Theory of Computing, the largest ACM conference) that will be held in El Paso in May. This workshop idea got several positive responses, both from interval community and from people from theory of Computing who are interested in broadening their contacts, but the topic is reasonably narrow and therefore, this workshop, to my mind, is not an obstacle to having a more large-scale interval conference elsewhere. Are SCAN or any other regular major interval-related conferences scheduled for 1997? If yes, for when? Please send the information about the planned 1997 conferences and about your desire to participate to Chenyi Hu and to me (I will place this information on the Interval Computations web page). Maybe, if 1997 is out of question, we can still aim for 1998? From owner-reliable_computing Wed Feb 7 14:28:26 1996 Received: by interval.usl.edu id AA14214 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 7 Feb 1996 06:30:44 -0600 Received: from wrzx01.rz.uni-wuerzburg.de by interval.usl.edu with SMTP id AA14208 (5.65c/IDA-1.4.4 for ); Wed, 7 Feb 1996 06:29:34 -0600 Received: from wi2x27.informatik.uni-wuerzburg.de by wrzx01.rz.uni-wuerzburg.de (4.1/uniwue-M-3.2) id AA26196; Wed, 7 Feb 96 13:28:27 +0100 Received: by wi2x27.informatik.uni-wuerzburg.de (5.65v3.0/uniwue-C-3.2) id AA22512; Wed, 7 Feb 1996 13:28:26 +0100 Date: Wed, 7 Feb 1996 13:28:26 +0100 Message-Id: <9602071228.AA22512 [at] wi2x27 [dot] informatik.uni-wuerzburg.de> From: "Prof.Dr.J.Wolff v. Gudenberg" To: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Cc: reliable_computing [at] interval [dot] usl.edu Subject: An interval conference in China? In-Reply-To: <9602070417.AA03068 [at] cs [dot] utep.edu> References: <9602070417.AA03068 [at] cs [dot] utep.edu> Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1050 X-Lines: 31 Vladik Kreinovich writes: > An interval conference in China? > > Dear Colleagues, > > I have just got the following proposal from > Dr. Chenyi Hu (Houston, TX): > > > From: hu [at] happy [dot] dt.uh.edu > > > > Dear Vladik, > > > > Just got a fax from China ... > > > > If the interval community is interested in holding a conference in China, > > I can surely make very nice arrangements with reasonable price. I'll be very > > happy to organize such meeting in China in 1997 if there is a need. > > Please send the information about the planned 1997 conferences > and about your desire to participate to Chenyi Hu and to me > (I will place this information on the Interval Computations web page). > > Maybe, if 1997 is out of question, we can still aim for 1998? > Since we have INTERVAL 96 in Wuerzburg, Germany 9/30 - 10/2/96 and SCAN 97 in Lyon, France, with related topics I would prefer and propose to organize INTERVAL 98 in China. Juergen Wolff v Gudenberg (chairman of INTERVAL 96 From owner-reliable_computing Thu Feb 8 01:52:25 1996 Received: by interval.usl.edu id AA15164 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 7 Feb 1996 13:57:52 -0600 Received: from rcom.spb.su by interval.usl.edu with SMTP id AA15158 (5.65c/IDA-1.4.4 for ); Wed, 7 Feb 1996 13:57:22 -0600 Received: from nit by rcom.spb.su with UUCP id AA15588 (5.65.kiae-1 for reliable_computing [at] interval [dot] usl.edu); Wed, 7 Feb 1996 22:45:12 +0300 Received: by nit.spb.su (UUPC/@ v5.00, 25Nov92); Wed, 7 Feb 1996 22:52:25 +0300 To: reliable_computing [at] interval [dot] usl.edu Message-Id: Organization: Institute of New Technologies (St. Petersburg) From: nest [at] nit [dot] spb.su (Slava Nesterov) Date: Wed, 7 Feb 96 22:52:25 +0300 X-Mailer: BML [MS/DOS Beauty Mail v.1.36] Lines: 11 Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 288 X-Lines: 11 Dear Colleagues, -- Since we have INTERVAL 96 in Wuerzburg, Germany 9/30 - 10/2/96 -- and SCAN 97 in Lyon, France, with related topics -- I would prefer and propose to organize INTERVAL 98 in China. I think it is a good idea. V.Nesterov co-chairman of INTERVAL 94,96 From owner-reliable_computing Wed Feb 7 07:43:43 1996 Received: by interval.usl.edu id AA15417 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 7 Feb 1996 15:44:03 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA15411 (5.65c/IDA-1.4.4 for ); Wed, 7 Feb 1996 15:43:53 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA08708; Wed, 7 Feb 96 14:43:43 MST Date: Wed, 7 Feb 96 14:43:43 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602072143.AA08708 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: Interval conference in China Cc: postmath [at] netra [dot] nju.edu.cn Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 2544 X-Lines: 66 Dear Colleagues, Thanks a lot for prompt replies. It looks like a consensus is to make it in China in 1998. I put together a short proposal. This proposal is also placed on the Web, as part of the Forthcoming conferences section of the Interval Computations homepage http://cs.utep.edu/interval-comp/main.html The Web version has links to beautiful and informative homepages of Nanjing and Nanjing University. I am sending a copy of this proposal to Dr. Zuhe Shen. Any further comments or suggestions will be greatly appreciated. Vladik INTERVAL'98, Nanjing, China (a proposal) Following the suggestion of Dr. Chenyi Hu (Houston, TX) to organize an international interval conference in China, Prof. Dr. Juergen Wolff von Gudenberg, the organizer of the International INTERVAL'96 conference, proposed to organize the 1998 International Interval Conference INTERVAL'98 in Nanjing, China. This idea has been supported: * by Slava Nesterov, co-Chair of INTERVAL'94 and INTERVAL'96, and the Editor-in-Chief of the International Journal Reliable Computing; * by R. Baker Kearfott, co-Chair of INTERVAL'96 and the Representative of Reliable Computing in the Westren Hemisphere, and by several other researchers. NANJING AND NANJING UNIVERSITY Nanjing University is where Prof. Dr. Zuhe Shen, the leading Chinese researcher in interval computations, works. It is among the top four best universities in China. Nanjing University often hosts big international conferences. (The latest one is International Congress of Chemists.) The city of Nanjing was the capital of 6 dynasties in the histroy of China. the very name Nanjing means Southern Capital. Nanjing is one of the best cities in China for tour purpose. The best time to visit Nanjing is April and May. ORGANIZATION Two American interval researchers will be visiting China in 1996 and 1997 as guests of Academia Sinica, and they volunteered to help with the organization: * Chenyi Hu, University of Houston-Downtown, Houston, TX, email hu [at] happy [dot] dt.uh.edu * Jiu Ding, Department of Mathematics, University of Southwestern Missisipi, Hattiesburg, Mississippi jiu_ding [at] bull [dot] cc.usm.edu RESEARCHERS WHO EXPRESSED INTEREST IN ATTENDING (so far) * Jiu Ding (Hattiesburg, Missisipi) * Andreas Frommer (Wuppertal, Germany) * Chenyi Hu (Houston, Texas) * R. Baker Kearfott (Lafayette, Louisiana) * Vladik Kreinovich (El Paso, TX) * Slava Nesterov (St. Petersburg, Russia) * Juergen Wolff von Gudenberg (Wuerzberg, Germany) From owner-reliable_computing Thu Feb 8 06:20:53 1996 Received: by interval.usl.edu id AA15689 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 7 Feb 1996 18:35:23 -0600 Received: from sovcom.kiae.su by interval.usl.edu with SMTP id AA15683 (5.65c/IDA-1.4.4 for ); Wed, 7 Feb 1996 18:34:49 -0600 Received: by sovcom.kiae.su id AA09359 (5.65.kiae-1 for reliable_computing [at] interval [dot] usl.edu); Thu, 8 Feb 1996 03:22:10 +0300 Message-Id: <199602080022.AA09359 [at] sovcom [dot] kiae.su> Received: by sovcom.KIAE.su (UUMAIL/2.0); Thu, 8 Feb 96 03:22:09 +0300 Received: by globlab.msk.su (UUPC/@ v5.09gamma, 14Mar93); Thu, 8 Feb 1996 03:20:55 +0300 To: reliable_computing [at] interval [dot] usl.edu From: Alexander G. Yakovlev Date: Thu, 8 Feb 1996 03:20:53 +0300 Subject: Please subscribe (or resubscribe)! Lines: 334 Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 14111 X-Lines: 334 International Journal ***** ***** RELIABLE COMPUTING (formerly INTERVAL COMPUTATIONS) ***** ***** A refereed international journal, RELIABLE COMPUTING is the only periodical in the world devoted specifically to various aspects of reliable mathematical computations based on finite representations of mathematical objects, and giving a guaranteed accuracy of computed results. It is managed by an international editorial board from Bulgaria, Germany, Japan, Russia, and the United States, and printed in Russia. The journal is distributed in approximately 30 countries throughout the world. It includes various items in the fields of theoretical research, computer tools, applications, interdisciplinary research, and other relevant areas in the field of mathematical computations with the use of a computer. The following subjects are within the focus of the journal: numerical, symbolic and combined algorithms for solving computational problems, the interval approach, computer architectures for mathematical computations, contemporary languages for mathematical processes description, methods of localization of solutions, teaching of reliable computational methods at universites, etc. "Reliable Computing" also publishes proceedings of international conferences, such as INTERVAL'XX, SCAN'XX, etc. The editorial policy is based on the fruitful experience of 4 years of publishing the journal "Interval Computations," and continues its best traditions, but now it is not restricted to the interval approach. Additionaly, many enchancements has been made in the journal's typesetting and layout, and its size is now larger. We hope the widened scope of our journal will attract a wider circle of readers. Consistent with the policy of "Interval Computations," the languages of publication are English and Russian, with at least 95% of the items in English. All articles are provided with abstracts in both languages. Papers published in "Reliable Computing" are listed in well-known abstracting and bibliographic journals, i.e. "Mathematical Reviews", "Referativnyi Zhurnal VINITI" and others. The journal includes: o original papers o surveys and tutorials o reports on new computer tools o bibliographies o reviews of new books o letters to the editor o information about scientific meetings o advertisements of software and hardware products All published issues of "Interval Computations" and "Reliable Computing", excluding No 1/1991, are available for purchase (see the section below). Additionally, Supplementum 1, consisting of a ``Bibliography of Works on Interval Computations Published in Russian'', with a subject index, is now available. It includes almost all sensible references in the field (547 items in the current version) in English translation. Upon demand, the Editorial Board can provide translation to English of the texts referenced in the bibliography (papers, reports, parts of books, etc.). In addition, the Editorial Board has published the collection of abstracts of the International Conference on Interval and Computer- Algebraic Methods in Science and Engineering (Interval'94), which was held in St.Petersburg, Russia, March 7-10, 1994. Abstracts are in English, two pages average, including postal and e-mail addresses of the authors (total 226 p.). The other Supplementum contains peer-refereed and edited abstracts of APIC'95 (International Workshop on Applications of Interval Computations), which was held in El Paso, Texas, February 23-35, 1995. Abstracts are in English, including authors' addresses (total 244 p.) EDITORIAL BOARD The Editorial Board of "Reliable Computing" includes: G. ALEFELD Karlsruhe, Germany B. S. DOBRONETS Krasnoyarsk, Russia R. B. KEARFOTT Lafayette, LA, USA - official representative in the Western Hemisphere V. KREINOVICH El Paso, TX, USA S. M. MARKOV Sofia, Bulgaria E. A. MUSAEV St. Petersburg, Russia - Executive Editor M. T. NAKAO Fukuoka, Japan V. M. NESTEROV St. Petersburg, Russia - Editor-in-Chief H. RATSCHEK Dusseldorf, Germany S. M. RUMP Hamburg, Germany A. L. SEMENOV Moscow, Russia S. P. SHARY Krasnoyarsk, Russia J. WOLFF VON GUDENBERG Wuerzburg, Germany - official representative in Europe A. G. YAKOVLEV Moscow, Russia - Managing Editor V. S. ZYUZIN Saratov, Russia The address of the Editorial Board: Box 52 St.Petersburg 195256, Russia Phone: +7-812-534-4120 Fax: +7-812-234-4852 E-mail: nest [at] nit [dot] spb.su (preferable), intcom [at] glasnet [dot] ru INFORMATION FOR AUTHORS Materials submitted for publication in "Reliable Computing" may represent original articles, reviews, presentations of new hardware and software tools, book reviews, information on scientific meetings on relevant topics which are scheduled or have been recently held, etc. Responses, amendments, and additions to published materials can be submitted as letters to the Editor. Manuscripts are submitted in English (preferably) or in Russian in three hard copies. An article should have a title, name(s) of the author(s), and an abstract (11 lines or shorter). If the title is 60 characters or longer, author(s) should also provide a short title to be placed on top of every page. The bibliography must be placed at the end of the paper. After the bibliography, the author(s) must place their mailing and e-mail addresses. (Electronic mail addresses facilitate and speed up communications and are highly desirable.) We encourage submitting computer files of the articles, preferably in LaTeX or any other TeX variety format. However, it is acceptable to submit files in other computer formats or even typewritten hard copies. Files are sent on diskettes or via electronic mail. IN ANY CASE, SUBMITTING THREE HARD COPIES IS OBLIGATORY. If typewritten hard copies only are submitted, they should be printed with double spacing. Formulae should be handwritten and (in the first copy) marked as is usual in publishing practice: Superscripts and subscripts should be marked by arcs, Greek letters should be underlined in red, Gothic letters, in blue, and vectors, in black. All letters in formulae are normally printed in italics; if it is necessary to set them in the normal typeface, they should be marked by a square bracket below. To distinguish capital letters from corresponding lower case letters, capitals should be marked by double underlining, and small letters, by double overlining. Figures can be either placed in the body of the text or supplied on separate sheets. In the latter case, the author should mark in the margin of the manuscript where the figures are to be inserted. All tables and figures should be distinctly and consequently enumerated. Only figures in black and white (without halftones) are accepted. For figures submitted as computer files, PostScript (preferably) and PCX formats are allowed. While there is no limit on the volume of articles, an author should obtain permission from an Editor prior to sending papers more than 30 pages long. The Editors reserve the right to divide long publications into parts to be published in several issues. All papers submitted for publication are subject to review, and may be admitted, rejected, or sent to the author for revision. The Editorial Board reserves the right to translate submitted articles from Russian into English. To partially cover the expenses of the journal production, we encourage the authors or their institutions to pay the page charges. These charges are not mandatory, but they are vitally important for the journal. We want to thank all who managed to page these charges and thus financially support this publication. Materials for publication should be submitted either to the Editor-in-Chief or to any other member of the Editorial Board. If the cover letter does not designate the corresponding author, all the correspondence will be sent to the first author. The authors of every published article will receive 50 free reprints of their article. These reprints will be also sent either to the first author, or to the corresponding author. SUBSCRIPTION INFORMATION Subscriptions to "Reliable Computing" are available, starting with any of the issues, from the first still available onward. The subscription price (including postage) is for 4 issues, but you can subscribe to more issues at a proportional price. Also, you may order Supplementum 1 and the collections of INTERVAL'94 and APIC'95 abstracts. ATTENTION: FOR INSTITUTIONS (UNIVERSITIES, LIBRARIES, COMPANIES, ETC.) ALL INDICATED PRICES SHOULD BE DOUBLED. Western Hemisphere Europe, Asia and Africa Price for 4 issues US$ 60.00 DM 84.00 (postage included) Price for Supplementum 1 US$ 10.00 DM 14.00 (postage included) Price for the collection US$ 15.00 DM 21.00 of Interval'94 abstracts (postage included) Price for the collection US$ 15.00 DM 21.00 of APIC'95 abstracts (postage included) If you wish to subscribe to "Reliable Computing" and/or receive the Supplements, please send this order form, along with a check for the appropriate amount (or use a bank transfer), to one of our official representatives: In the Western Hemisphere In Europe, Asia and Africa Dr. R. Baker Kearfott Prof. Dr. J. Wolff von Gudenberg Department of Mathematics Lehrstuhl f. Informatik II Univ. of Southwestern Louisiana Universitaet Wuerzburg U.S.L. Box 4-1010 Lafayette Am Hubland LA 70504-1010 Wuerzburg D-97074 USA Germany Office: (318) 231-5270 Phone: +49-931-888-5517 Home: (318) 981-9744 Fax: +49-931-888-4602 E-mail: rbk [at] usl [dot] edu E-mail: wolff [at] informatik [dot] uni- wuerzburg.de Please make a check to Instructions for bank transfer: the "University of holder: J. Wolff v. Gudenberg Southwestern Louisiana" number: 560219826 bank: Kreissparkasse Wuerzburg bank code: 790 501 30 ATTENTION: all subscribers paid for "Interval Computations" before its renaming will received the corresponding issues of "Reliable Computing" for old prices! ---------------------------------------------------------------------- "RELIABLE COMPUTING" ORDER FORM (for individuals only) Name: ________________________________________________________________ Organization: ________________________________________________________ Address: _____________________________________________________________ ______________________________________________________________________ State: _______________________________ Zip: __________________________ Tel.: ___________________________ (optional) Fax: ____________________________ (optional) E-mail: __________________________________________ (highly desirable!) Please check: [ ] I wish to subscribe to the journal "Reliable Computing", [ ] sum of $60/84 DM paid for 4 issues [ ] sum of _________ paid for ___ issues (cheque/money order enclosed). First issue to send: No___, 199__. [ ] Send me additionally Supplementum 1 (Bibliography of Works on Interval Computations Published In Russian). Sum of $10/14 DM paid (cheque/money order enclosed). [ ] Send me additionally the collection of INTERVAL'94 abstracts. Sum of $15/21 DM paid (cheque/money order enclosed). [ ] Send me additionally the collection of APIC'95 abstracts. Sum of $15/21 DM paid (cheque/money order enclosed). Date: ______________________ Signature: ______________________ ---------------------------------------------------------------------- "RELIABLE COMPUTING" ORDER FORM (for institutions only) Name (for contact): __________________________________________________ Organization: ________________________________________________________ Address: _____________________________________________________________ ______________________________________________________________________ State: _______________________________ Zip: __________________________ Tel.: ___________________________ (optional) Fax: ____________________________ (optional) E-mail: __________________________________________ (highly desirable!) Please check: [ ] We wish to subscribe to the journal ``Reliable Computing'', [ ] sum of $120/168 DM paid for 4 issues [ ] sum of __________ paid for ___ issues (delivery via surface mail, cheque/money order enclosed). First issue to send: No___, 199__. [ ] Send us additionally Supplementum 1 (Bibliography of Works on Interval Computations Published in Russian). Sum of $20/28 DM paid (cheque/money order enclosed). [ ] Send us additionally the collection of INTERVAL'94 abstracts. Sum of $30/42 DM paid (cheque/money order enclosed). [ ] Send me additionally the collection of APIC'95 abstracts. Sum of $30/42 DM paid (cheque/money order enclosed). Date: _______________________ Signature: _____________________ From owner-reliable_computing Fri Feb 9 04:19:22 1996 Received: by interval.usl.edu id AA16020 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 00:19:39 -0600 Received: from arch.su.EDU.AU by interval.usl.edu with SMTP id AA16014 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 00:19:34 -0600 Received: from localhost.arch.su.EDU.AU by arch.su.EDU.AU (4.1/SMI-4.1) id AA02904; Thu, 8 Feb 96 17:19:23 EST Message-Id: <9602080619.AA02904 [at] arch [dot] su.EDU.AU> Cc: reliable_computing [at] interval [dot] usl.edu Date: Thu, 08 Feb 96 17:19:22 +1100 From: The Balkan Brain Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 388 X-Lines: 16 > Dear Colleagues, > -- Since we have INTERVAL 96 in Wuerzburg, Germany 9/30 - 10/2/96 > -- and SCAN 97 in Lyon, France, with related topics > -- I would prefer and propose to organize INTERVAL 98 in China. > I think it is a good idea. > V.Nesterov > co-chairman of INTERVAL 94,96 Should be great Simeon J. Simoff neither chairman nor co-chairman of INTERVAL 94, 96 From owner-reliable_computing Thu Feb 8 11:09:36 1996 Received: by interval.usl.edu id AA16464 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 05:18:05 -0600 Received: from tamdhu.dcs.st-andrews.ac.uk (tamdhu.dcs.st-and.ac.uk) by interval.usl.edu with SMTP id AA16451 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 05:10:10 -0600 Received: from kestrel.dcs.st-and.ac.uk (chaser.dcs.st-and.ac.uk) by tamdhu.dcs.st-andrews.ac.uk (4.1/SMI-4.1) id AA02586; Thu, 8 Feb 96 11:08:33 GMT Received: from [138.251.192.164] (moore) by kestrel.dcs.st-and.ac.uk (4.1/SMI-4.1) id AA22502; Thu, 8 Feb 96 11:08:31 GMT X-Sender: michael [at] tartan [dot] dcs.st-and.ac.uk Message-Id: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Date: Thu, 8 Feb 1996 11:09:36 +0000 To: reliable_computing [at] interval [dot] usl.edu From: michael [at] dcs [dot] st-and.ac.uk (Michael Wolfe) Subject: Reliable Computing Re-subscription Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1050 X-Lines: 27 I have just received Prof. Yakovlev's message regarding subscription or re-subscription to the journal Reliable Computing. Being a regular subscriber to the journal, I have always received, with the latest number, a request to renew my subscription when due. Does Prof. Yakovlev's message mean that in future requests for re-subscription will not be sent as they have been until now? I mention this in case others might be as confused as I. ******************************************************************************* Michael Wolfe, University of St Andrews, School of Mathematical and Computational Sciences, Mathematical Institute, North Haugh, St Andrews, Fife KY16 9SS, SC0TLAND. ------------------------------------------------------------------------------- Tel: +44 01334 476161 Ext. 3752 +44 01334 463752 (direct line) Fax: +44 01334 463748 Email: michael [at] dcs [dot] st-and.ac.uk URL: http://www-solar.dcs.st-and.ac.uk/~michael/intervals.html ******************************************************************************* From owner-reliable_computing Thu Feb 8 01:40:04 1996 Received: by interval.usl.edu id AA16768 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 09:40:10 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA16762 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 09:40:07 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA13336; Thu, 8 Feb 96 08:40:04 MST Date: Thu, 8 Feb 96 08:40:04 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602081540.AA13336 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: subscription renewals Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1061 X-Lines: 21 In his inquiry to the mailing list, Professor Michael Wolfe asked: > I have just received Prof. Yakovlev's message regarding subscription or > re-subscription to the journal Reliable Computing. Being a regular > subscriber to the journal, I have always received, with the latest number, > a request to renew my subscription when due. Does Prof. Yakovlev's message > mean that in future requests for re-subscription will not be sent as they > have been until now? I have asked Sasha Yakovlev whether his message meant any changes in the journal's policy and he assured me that it is not the case. His appeal (if I understood him correctly), was caused by the fact that only a minority people on the mailing list are currently subscribed to the journal, and although some new people subscribed, some subscribers did not renew their subscripiton. Requests for renewal will be sent as before, there are no changes planned. Changes will occur in 1997 when Kluwer takes over the journal; then, Kluwer will be also handling subscriptions, renewal notices, etc. From owner-reliable_computing Thu Feb 8 07:40:13 1996 Received: by interval.usl.edu id AA17086 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 15:40:44 -0600 Received: from mailhost.lanl.gov by interval.usl.edu with SMTP id AA17080 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 15:40:41 -0600 Received: from xdiv.lanl.gov by mailhost.lanl.gov (8.6.12/1.2) id OAA23560; Thu, 8 Feb 1996 14:40:33 -0700 Received: from xdiv.lanl.gov.xdiv (angus.lanl.gov [128.165.123.30]) by xdiv.lanl.gov (8.6.12/8.6.12) with SMTP id OAA22375 for ; Thu, 8 Feb 1996 14:40:13 -0700 Date: Thu, 8 Feb 1996 14:40:13 -0700 Message-Id: <199602082140.OAA22375 [at] xdiv [dot] lanl.gov> Received: by xdiv.lanl.gov.xdiv (4.1/SMI-4.1) id AA24012; Thu, 8 Feb 96 14:40:13 MST From: "Rajendra B. Patil" To: reliable_computing [at] interval [dot] usl.edu Subject: Midpoint test...comparisons Reply-To: Raj Patil Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1028 X-Lines: 26 Hi: Regarding some of the empirical comparison (Jansson/Knueppel, Ratz/Csnedes, Hansen's book, etc)..... For example, it is well know that having a good value of upperbound early in interval branch an bound is useful in removing boxes. In many of the test functions where the global optima is at the center of the test domain the first midpoint test itself leads to the best value for the upperbound, which if we shift the domain would not be the case and the results could be totally different. As in the worst case it could be the last box in the branch and bound process whose midpoint will be this best upperbound which because of the symmetric domain was obtained in the first midpoint test itself. For example on problems like Schwefel 3.7, Matyas, Rastrigin, Griewank, Treccani, Three hump camel back, Powell, etc. when tested over symmetric domain will lead to this situation as all of these functions have global opt. exactly at the center (0,...0) of the test domain used in these empirical studies. Raj Patil From owner-reliable_computing Thu Feb 8 07:16:42 1996 Received: by interval.usl.edu id AA17367 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 17:16:03 -0600 Received: from mercury.Sun.COM by interval.usl.edu with SMTP id AA17361 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 17:15:58 -0600 Received: from Eng.Sun.COM by mercury.Sun.COM (Sun.COM) id PAA13732; Thu, 8 Feb 1996 15:15:55 -0800 From: Bill.Walster [at] eng [dot] sun.com Received: from gww.eng.sun.com by Eng.Sun.COM (5.x/SMI-5.3) id AA15708; Thu, 8 Feb 1996 15:15:53 -0800 Received: by gww.eng.sun.com (5.x/SMI-SVR4) id AA05885; Thu, 8 Feb 1996 15:16:42 -0800 Date: Thu, 8 Feb 1996 15:16:42 -0800 Message-Id: <9602082316.AA05885 [at] gww [dot] eng.sun.com> To: reliable_computing [at] interval [dot] usl.edu, rbp1 [at] lanl [dot] gov Subject: Re: Midpoint test...comparisons X-Sun-Charset: US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 2071 X-Lines: 51 In doing all the empirical investigations reported in Hansen's book, we discovered that even providing the exact minimum of the objective function from the outset, did not have a dramatic effect on overall run time. However, this result depends on using the lease lower bound on the objective function to select the next box to process. In addition, care must be used when deciding which dimensions will be profitable to split. The heuristic we used is to split on the dimension with the most violent fluctuation in the objective function. We explicitly offset initial boxes so initial splits did not land on the global solution. Bill Walster >From owner-reliable_computing [at] interval [dot] usl.edu Thu Feb 8 14:12 PST 1996 >Date: Thu, 8 Feb 1996 14:40:13 -0700 >From: "Rajendra B. Patil" >To: reliable_computing [at] interval [dot] usl.edu >Subject: Midpoint test...comparisons >Precedence: bulk >Content-Type: text >X-Lines: 26 > >Hi: > >Regarding some of the empirical comparison (Jansson/Knueppel, Ratz/Csnedes, >Hansen's book, etc)..... > >For example, it is well know that having a good value of upperbound >early in interval branch an bound is useful in removing boxes. >In many of the test functions where the global optima is at the >center of the test domain the first midpoint test itself leads to the >best value for the upperbound, which if we shift the domain would >not be the case and the results could be totally different. As in the >worst case it could be the last box in the branch and bound process >whose midpoint will be this best upperbound which because of the >symmetric domain was obtained in the first midpoint test itself. > >For example on problems like Schwefel 3.7, Matyas, Rastrigin, >Griewank, Treccani, Three hump camel back, Powell, etc. when tested >over symmetric domain will lead to this situation as all of these >functions have global opt. exactly at the center (0,...0) of the test >domain used in these empirical studies. > >Raj Patil > > > > > From owner-reliable_computing Thu Feb 8 14:22:36 1996 Received: by interval.usl.edu id AA17695 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 8 Feb 1996 22:22:43 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA17689 (5.65c/IDA-1.4.4 for ); Thu, 8 Feb 1996 22:22:40 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA17835; Thu, 8 Feb 96 21:22:36 MST Date: Thu, 8 Feb 96 21:22:36 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602090422.AA17835 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: INTERVAL'98 in China Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 2376 X-Lines: 61 Dear Friends, Thanks for the overwhelming response. It looks like there is practically a consensus that Interval'98 in China is a good idea. I have contacted Dr. Zuhe Shen from Nanjing University, the leading Chinese interval researcher. The University is currently on vacation related to Chinese New Year (until February 26), so, probably, we will hear from him in March. I was advised that the city of Nanjing is not the only possible place where Nanjing University can sponsor conferences, there are also picturesque places around Nanjing that served as sites of international meetings. With Misha (my co-maintainer) we added some info about these places and links to the pictures and homepage to China conference proposal homepage on Interval URL http://cs.utep.edu/interval-comp/main.html The choice of the exact location will be probably made by the local organizers. We have also updated the list of researchers who are interested in participation. I am very thankful to everyone who responded. Please send further suggestions (and criticisms) re homepage (for one, I only have a link to a pre-1991 map of China, which still has Soviet Union as a neigbor; if anyone knows a link to a better map, please let me know). Vladik P.S. In response to reasonable doubts expressed by several respondents: my experience of a former citizen of the USSR has shown me that international scientific conferences greatly benefit local researchers (especially students and young researchers, for whom it may be the only chance to attend a major conference), and that these conference do not in any way benefit political entities. I personally strongly believe, therefore, that the question of whether to have this conference in China is up to Chinese interval researchers: if interval researchers from China will agree to host this scientific conference, we should do our best to organize it. I appreciate the fact that the concerns were raised in prinate emails to me, not to the list; It is not my intention to turn this list into an arena of political discussions, and I apologize for bringing this issue if there is a general belief that it was inappropriate. I just wanted to make my attitude clear: I am aware of the possible controversy, and I am sure that this conference will benefit Chinese researchers and not political entities. Vladik From owner-reliable_computing Fri Feb 9 11:42:46 1996 Received: by interval.usl.edu id AA18017 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 9 Feb 1996 01:46:32 -0600 Received: from helle.uthes.ariadne-t.gr by interval.usl.edu with SMTP id AA18011 (5.65c/IDA-1.4.4 for ); Fri, 9 Feb 1996 01:46:13 -0600 Message-Id: <199602090746.AA18011 [at] interval [dot] usl.edu> Received: by helle.uthes.ariadne-t.gr (1.38.193.4/16.2) id AA04573; Fri, 9 Feb 1996 09:42:46 +0200 Date: Fri, 9 Feb 1996 09:42:46 +0200 From: A. Akritas Apparently-To: reliable_computing [at] interval [dot] usl.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 612 X-Lines: 22 Hi, I recently had a paper published in RC (together with Malaschonok) and a letter with page charges was sent to my University address. However, I am on Sabbatical leave for the whole year and the letter got lost. Please send me another letter requesting reimbursement for the page charges. The University has agreed to pay. -------------------------------- Please send the letter to: Prof. Alkis Akritas c/o Prof. Z. BAVEL Department of Computer Science University of Kansas 415 Snow Hall Lawrence, KS 66045-2192 Thank you for your understanding. Sincerely, Alkis Akritas From owner-reliable_computing Thu Feb 8 20:50:45 1996 Received: by interval.usl.edu id AA18400 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 9 Feb 1996 04:51:10 -0600 Received: from mailhost.lanl.gov by interval.usl.edu with SMTP id AA18394 (5.65c/IDA-1.4.4 for ); Fri, 9 Feb 1996 04:51:06 -0600 Received: from xdiv.lanl.gov by mailhost.lanl.gov (8.6.12/1.2) id DAA15870; Fri, 9 Feb 1996 03:51:04 -0700 Received: from xdiv.lanl.gov.xdiv (angus.lanl.gov [128.165.123.30]) by xdiv.lanl.gov (8.6.12/8.6.12) with SMTP id DAA27179; Fri, 9 Feb 1996 03:50:45 -0700 Date: Fri, 9 Feb 1996 03:50:45 -0700 From: "Rajendra B. Patil" Message-Id: <199602091050.DAA27179 [at] xdiv [dot] lanl.gov> To: Bill.Walster [at] eng [dot] sun.com, reliable_computing [at] interval [dot] usl.edu Subject: Re: Midpoint test...comparisons Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1496 X-Lines: 37 Dear Dr. Walster: I agree that in some cases providing the exact value may not have dramatic effect on the speed of the algorithm. But, one of the only few alternatives to speed things up in such cases, especially if no derivative information is available, is to provide a good value of the upperbound. Consider a simple example of Rastrigin function in 2-D. f(x1, x2) = x1^2 + x2^2 - cos(18x1) - cos(18x2) With midpoint test at the center over the domain of [-1,1]x[-1,1] gives you the best upperbound of -2.0 at (0,0) and it takes a simple interval BB algorithm about 250 evaluations (eps = 10^-2 termination, Max(Diam) list ordering, etc) and creates 4 boxes at termination. If you do not evaluate the function at the FIRST midpoint, the same algorithm takes 66354 evaluations and creates 4620 boxes and takes orders of magnitude longer to solve the same problem with same termination criterion. The reason for this is that the split takes place exactly at the optimum and take a long time to isolate it. This is a dramatic difference and is just a 2-D example. Though this can be taken care by detecting such situation, the main concern I had was the validity of some of the results I have seen in comparing interval branch and bound algorithms. For example Jansson and Knueppel (report 92.1) report 250 evaluations (pp. 46) of this same function and using the same algorithm with simply removing the first midpoint test leads to the above results. Regards, Raj. From owner-reliable_computing Fri Feb 9 00:48:08 1996 Received: by interval.usl.edu id AA18800 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 9 Feb 1996 10:48:11 -0600 Received: from mercury.Sun.COM by interval.usl.edu with SMTP id AA18793 (5.65c/IDA-1.4.4 for ); Fri, 9 Feb 1996 10:48:01 -0600 Received: from Eng.Sun.COM by mercury.Sun.COM (Sun.COM) id IAA27739; Fri, 9 Feb 1996 08:47:21 -0800 From: Bill.Walster [at] eng [dot] sun.com Received: from gww.eng.sun.com by Eng.Sun.COM (5.x/SMI-5.3) id AA20065; Fri, 9 Feb 1996 08:47:19 -0800 Received: by gww.eng.sun.com (5.x/SMI-SVR4) id AA07086; Fri, 9 Feb 1996 08:48:08 -0800 Date: Fri, 9 Feb 1996 08:48:08 -0800 Message-Id: <9602091648.AA07086 [at] gww [dot] eng.sun.com> To: rbp1 [at] xdiv [dot] lanl.gov Subject: Re: Midpoint test...comparisons Cc: reliable_computing [at] interval [dot] usl.edu X-Sun-Charset: US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 3027 X-Lines: 76 Dear Dr. Patil, In all our numerical work, Eldon and I assumed twice continuously differentiable objective functions. We exploited this information as much as possible. With no derivative information available, your point is obvious. Using derivative information, our experience is different. The reason we duplicated all our testing while providing the exact value of the objective function at the global mininum, was to check on the benefit of our point-search algorithm to quickly obtain a good f-bar (least upper bound on the objective function). Once we took the steps I mentioned in my previous email, we were surprised to find that while some help, a perfect f-bar from the outset was not all that big a help. We were able to get a good f-bar quite fast. Had we shut off these parts of the algorithm, no doubt we would have seen results like those you report. For good performance, it is important for all the components of the overall algorithm to be permitted to complement each other. I have not run your test case on our algorithm. Indeed, at present I don't have the ability to do so. Regards, Bill >From rbp1 [at] xdiv [dot] lanl.gov Fri Feb 9 02:52 PST 1996 >Date: Fri, 9 Feb 1996 03:50:45 -0700 >From: "Rajendra B. Patil" >To: Bill.Walster@Eng, reliable_computing [at] interval [dot] usl.edu >Subject: Re: Midpoint test...comparisons >Content-Type: text >X-Lines: 37 > > >Dear Dr. Walster: > >I agree that in some cases providing the exact value may not have dramatic >effect on the speed of the algorithm. But, one of the only few >alternatives to speed things up in such cases, especially if no >derivative information is available, is to provide a good value of >the upperbound. > >Consider a simple example of Rastrigin function in 2-D. > >f(x1, x2) = x1^2 + x2^2 - cos(18x1) - cos(18x2) > >With midpoint test at the center over the domain of [-1,1]x[-1,1] >gives you the best upperbound of -2.0 at (0,0) and it takes a >simple interval BB algorithm about 250 evaluations (eps = 10^-2 >termination, Max(Diam) list ordering, etc) and creates 4 boxes >at termination. > >If you do not evaluate the function at the FIRST midpoint, >the same algorithm takes 66354 evaluations and creates 4620 >boxes and takes orders of magnitude longer to solve the same problem >with same termination criterion. The reason for this is that >the split takes place exactly at the optimum and take a long >time to isolate it. > >This is a dramatic difference and is just a 2-D example. Though >this can be taken care by detecting such situation, >the main concern I had was the validity of some of the results I have seen >in comparing interval branch and bound algorithms. For example Jansson and >Knueppel (report 92.1) report 250 evaluations (pp. 46) of this same >function and using the same algorithm with simply removing the >first midpoint test leads to the above results. > >Regards, > >Raj. > From owner-reliable_computing Fri Feb 9 03:24:58 1996 Received: by interval.usl.edu id AA18910 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 9 Feb 1996 11:47:22 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA18904 (5.65c/IDA-1.4.4 for ); Fri, 9 Feb 1996 11:47:19 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA20157; Fri, 9 Feb 96 10:24:59 MST Date: Fri, 9 Feb 96 10:24:58 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602091724.AA20157 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: Interval 98: clarification Cc: vladik [at] cs [dot] utep.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 653 X-Lines: 21 Dear Colleagues, It turns out that my Web access instructions were not clear enough. I apologize for that. Here are the detailed instructions. To get to the web site on the proposed Interval'98 in China, go to the main Interval Computations homepage http://cs.utep.edu/interval-comp/main.html when there, click on Forthcoming conferences, and when in Forthcoming Conferences page, click on Interval'98 (China). You can also access the Interval'98 page directly as http://cs.utep.edu/interval-comp/china.html If you do not access to the Web, please mail me, and I will send you the latest text (ASCII) version without pictures by email. Vladik From owner-reliable_computing Fri Feb 9 04:18:28 1996 Received: by interval.usl.edu id AA19333 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 9 Feb 1996 12:18:35 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA19327 (5.65c/IDA-1.4.4 for ); Fri, 9 Feb 1996 12:18:31 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA20502; Fri, 9 Feb 96 11:18:28 MST Date: Fri, 9 Feb 96 11:18:28 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602091818.AA20502 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: BibTex interval bibliography Cc: pkahl [at] cs [dot] utep.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 842 X-Lines: 21 Dr. Nelson Beebe is currently working on the BibTeX interval bibliography. We have included a link to the current version (it is 1 Meg, via ftp) to the new Bibliographies pages of the Interval Computations website. To get to this bibliography, go to the main page http://cs.utep.edu/interval-comp/main.html click on Bibliographies, and then click on the corresponding bibliography. You can also get to this bibliography directly, it is at ftp://ftp.math.utah.edu/pub/tex/bib/intarith.bib Please feel free to download the bibliography and/or to send your existing BibTex bibliographies to Dr. Beebe for inclusion. We are most thankful to Dr. Beebe for his efforts that will help everyone who write interval papers in LaTeX and to Arnold Neumaier for informing us about the link. Vladik and Misha, maintainers of the interval website. From owner-reliable_computing Wed Feb 21 04:26:59 1996 Received: by interval.usl.edu id AA25950 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 21 Feb 1996 10:27:06 -0600 Received: by interval.usl.edu id AA25940 (5.65c/IDA-1.4.4 for reliable_computing); Wed, 21 Feb 1996 10:26:59 -0600 Date: Wed, 21 Feb 1996 10:26:59 -0600 From: "Kearfott R. Baker" Message-Id: <199602211626.AA25940 [at] interval [dot] usl.edu> To: reliable_computing Subject: Re: Midpoint test...comparisons Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 5202 X-Lines: 116 This concerns a thread begun two weeks ago (using exact optimizers from the outset in global optimization). I believe the important aspect of Bill Walster's commentary is that a local optimizer be used, in addition to the raw midpoint test, to obtain a good upper bound to the optimum. I also wholeheartedly agree that various parts of the algorithm will interact, and care should be taken to make this interaction beneficial. In addition, several researchers (including me) have used epsilon inflation around the approximate optimum to REMOVE as large a region as possible, subject to uniqueness of the critical point. My experiments have found this effective. The reason: portions of the region where the objective function is near its optimum value are removed from consideration, and it is easier to reject the remaining portions. Madsen and Caprani were perhaps the first to use this technique. Ron van Iwaarden recently completed a Ph.D. dissertation on the subject -- Ron has some carefully designed multidimensional experiments, plus a graphical interface to illustrate the process for the 2-dimensional problems. Ron has C++ software that he intends to make publicly available in the near future. I have Fortran 90 software that implements the process. I intend to make this software publicly available later this year or early next year, roughly concurrently with publication of my book on the subject. (The book will introduce interval techniques, automatic differentiation, optimization, the tesselation process, and the software. It will also include consideration of the non-smooth case, among other things.) --------------------------------------------------------------- R. Baker Kearfott, rbk [at] usl [dot] edu (318) 482-5346 (fax) (318) 482-5270 (work) (318) 981-9744 (home) URL: ftp://interval.usl.edu/pub/interval_math/www/kearfott.html Department of Mathematics, University of Southwestern Louisiana --------------------------------------------------------------- > Dear Dr. Patil, > > In all our numerical work, Eldon and I assumed twice continuously > differentiable objective functions. We exploited this information > as much as possible. With no derivative information available, your > point is obvious. Using derivative information, our experience > is different. > > The reason we duplicated all our testing while providing the exact > value of the objective function at the global mininum, was to check > on the benefit of our point-search algorithm to quickly obtain a good > f-bar (least upper bound on the objective function). Once we took > the steps I mentioned in my previous email, we were surprised to find > that while some help, a perfect f-bar from the outset was not all > that big a help. We were able to get a good f-bar quite fast. Had > we shut off these parts of the algorithm, no doubt we would have seen > results like those you report. > > For good performance, it is important for all the components of > the overall algorithm to be permitted to complement each other. > > I have not run your test case on our algorithm. Indeed, at present > I don't have the ability to do so. > > Regards, > > Bill > > > >From rbp1 [at] xdiv [dot] lanl.gov Fri Feb 9 02:52 PST 1996 > >Date: Fri, 9 Feb 1996 03:50:45 -0700 > >From: "Rajendra B. Patil" > >To: Bill.Walster@Eng, reliable_computing [at] interval [dot] usl.edu > >Subject: Re: Midpoint test...comparisons > >Content-Type: text > >X-Lines: 37 > > > > > >Dear Dr. Walster: > > > >I agree that in some cases providing the exact value may not have dramatic > >effect on the speed of the algorithm. But, one of the only few > >alternatives to speed things up in such cases, especially if no > >derivative information is available, is to provide a good value of > >the upperbound. > > > >Consider a simple example of Rastrigin function in 2-D. > > > >f(x1, x2) = x1^2 + x2^2 - cos(18x1) - cos(18x2) > > > >With midpoint test at the center over the domain of [-1,1]x[-1,1] > >gives you the best upperbound of -2.0 at (0,0) and it takes a > >simple interval BB algorithm about 250 evaluations (eps = 10^-2 > >termination, Max(Diam) list ordering, etc) and creates 4 boxes > >at termination. > > > >If you do not evaluate the function at the FIRST midpoint, > >the same algorithm takes 66354 evaluations and creates 4620 > >boxes and takes orders of magnitude longer to solve the same problem > >with same termination criterion. The reason for this is that > >the split takes place exactly at the optimum and take a long > >time to isolate it. > > > >This is a dramatic difference and is just a 2-D example. Though > >this can be taken care by detecting such situation, > >the main concern I had was the validity of some of the results I have seen > >in comparing interval branch and bound algorithms. For example Jansson and > >Knueppel (report 92.1) report 250 evaluations (pp. 46) of this same > >function and using the same algorithm with simply removing the > >first midpoint test leads to the above results. > > > >Regards, > > > >Raj. > > > From owner-reliable_computing Wed Feb 21 01:42:15 1996 Received: by interval.usl.edu id AA26280 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 21 Feb 1996 11:41:38 -0600 Received: from mercury.Sun.COM by interval.usl.edu with SMTP id AA26272 (5.65c/IDA-1.4.4 for ); Wed, 21 Feb 1996 11:41:26 -0600 Received: from Eng.Sun.COM by mercury.Sun.COM (Sun.COM) id JAA22893; Wed, 21 Feb 1996 09:41:24 -0800 From: Bill.Walster [at] eng [dot] sun.com Received: from gww.eng.sun.com by Eng.Sun.COM (5.x/SMI-5.3) id AA20309; Wed, 21 Feb 1996 09:41:21 -0800 Received: by gww.eng.sun.com (5.x/SMI-SVR4) id AA03305; Wed, 21 Feb 1996 09:42:15 -0800 Date: Wed, 21 Feb 1996 09:42:15 -0800 Message-Id: <9602211742.AA03305 [at] gww [dot] eng.sun.com> To: reliable_computing [at] interval [dot] usl.edu, rbk5287 [at] interval [dot] usl.edu Subject: Public Distribution of Interval Software X-Sun-Charset: US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 851 X-Lines: 22 Baker Kearfott writes in part: > >I have Fortran 90 software that implements the process. I intend to >make this software publicly available later this year or early next >year, roughly concurrently with publication of my book on the subject. >(The book will introduce interval techniques, automatic >differentiation, optimization, the tesselation process, and the >software. It will also include consideration of the non-smooth case, >among other things.) > I encourage everybody to make public as much of their software as possible. This is an important step in creating a market for interval methods. In this connection, SunSoft distributes freely available software on a separate CD. Baker has distributed his f90 module for interval arithmetic this way. It is a good way to get wide distribution. Regards to all, Bill From owner-reliable_computing Wed Feb 21 05:52:01 1996 Received: by interval.usl.edu id AA27045 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 21 Feb 1996 13:52:26 -0600 Received: from mailhost.lanl.gov by interval.usl.edu with SMTP id AA27037 (5.65c/IDA-1.4.4 for ); Wed, 21 Feb 1996 13:52:20 -0600 Received: from xdiv.lanl.gov by mailhost.lanl.gov (8.6.12/1.2) id MAA01127; Wed, 21 Feb 1996 12:52:01 -0700 Received: from xdiv.lanl.gov.xdiv (angus.lanl.gov [128.165.123.30]) by xdiv.lanl.gov (8.6.12/8.6.12) with SMTP id MAA21514; Wed, 21 Feb 1996 12:52:01 -0700 Date: Wed, 21 Feb 1996 12:52:01 -0700 From: "Rajendra B. Patil" Message-Id: <199602211952.MAA21514 [at] xdiv [dot] lanl.gov> To: rbk5287 [at] interval [dot] usl.edu Subject: Re: Midpoint test...comparisons Cc: reliable_computing [at] interval [dot] usl.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 483 X-Lines: 19 Regarding the Midpoint Test thread.... My original concern was the the validity of some empirical results published (number of function evaluations). Some of these results, under a simple coordinate rotation and/or shift of the domain are not valid for some of the test functions. This is due to the symmetry of the domain and global minima being at the center, which leads to a bias in using the first midpoint test in the standard interval BB algorithm. Raj Patil LANL From owner-reliable_computing Wed Feb 21 12:02:44 1996 Received: by interval.usl.edu id AA27690 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 21 Feb 1996 20:02:56 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA27684 (5.65c/IDA-1.4.4 for ); Wed, 21 Feb 1996 20:02:52 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA01356; Wed, 21 Feb 96 19:02:44 MST Date: Wed, 21 Feb 96 19:02:44 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602220202.AA01356 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: interval workshop in Australia Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 3934 X-Lines: 107 >From simeon [at] arch [dot] su.edu.au WORKSHOP IASTED International Conference Modeling, Simulation and Optimisation Gold Coast, Australia Title: Interval Techniques in Computing, Modelling, Simulation and Optimization (Dedicated to the 30th anniversary of the Interval Analysis) Convenor: Dr Simeon Simoff University of Sydney, Australia Co-Convenor: Fay Sudweeks University of Sydney, Australia International Advisory Panel: (to come) Duration: Half day Date: Afternoon of May 5, 1996 Price: $US 65.00 Venue: The Workshop will be held at the conference hotel. Submission of Workshop Papers: Four copies of extended abstracts of position papers (1-2 pages) plus an electronic copy, should be submitted by 11 March 1996 to: Fay Sudweeks Key Centre of Design Computing (G04) University of Sydney NSW 2006 Australia Email: fay [at] arch [dot] su.edu.au Electronic copy may be emailed or on disk, in ASCII, RTF, or Word 5 file format. Abstracts will be reviewed by the International Advisory Committee. Notification of acceptance will be sent by 20 March. Final submission will be short position papers of 3-5 pages by 15 April. Position papers will be included in the Workshop Proceedings which will be distributed to participants at the Workshop. Overview: The Interval Techniques (IT) workshop is part of MSO'96 conference. This workshop is intended to provide a forum for provocative discussion related to the theory, application and evolution of methods and techniques of Interval Analysis. Since Moore's publication on Interval Analysis appeared in 1966, the academic research community has been promoting the use of interval techniques to achieve guaranteed bounds for the results of numerical computations. Conventional numerical algorithms compute an estimate for an answer and an estimate for computing error. To be able to judge about the accuracy of the estimated answer user needs also extensive, expensive error analysis. The quality of scientific software packages has improved dramatically, but there are still problems for which catastrophically incorrect answers are returned to the user with no warning. Interval techniques compute an interval in which the correct answer is validated to lie. If interval techniques compute an answer [100.953456, 100.953457], then the result is within this interval, that means that 7 decimal places are known to be correct. If as a result an algorithm yields the interval [-10^30, +10^30], then it means that the answer is unknown. Consequently, interval computations include an assurance of their quality. >From the field of reliable computations interval approach entered the area of modeling, simulation and optimizaion. The advantages of the interval techniques for model building are the "wide" initial assumptions (methods require only that the unknown true value lies within the interval obtained from the experiments) and easy to perform hypothesis testing. In the early days, it was not very convinient to apply interval methods because suitable software was not widely available. Now there are several commercially available systems, others are under development. However, interval technique is known mostly within a narrow community of researchers in computer science and applied mathematics. Therefore, its application is relatively immature and as such presents exciting and challenging issues to the scientists from different disciplines, related to modeling, simulation and optimization. Workshop Material: Each participant will receive a copy of workshop proceeding. The information on this and other interval-related events is available from the Interval Computations website: URL http://cs.utep.edu/interval-comp/main.html In particular, you may click on the Forthcoming Conferences to get the information about future interval conferences and workshops. From owner-reliable_computing Fri Feb 23 13:24:26 1996 Received: by interval.usl.edu id AA05737 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 23 Feb 1996 05:27:38 -0600 Received: from rztsun3.rz.tu-harburg.de by interval.usl.edu with SMTP id AA05730 (5.65c/IDA-1.4.4 for ); Fri, 23 Feb 1996 05:27:30 -0600 Date: Fri, 23 Feb 96 12:24:26 +0100 From: rump@tu-harburg.d400.de ( Prof.Dr.S.M. Rump) Message-Id: <9602231124.AA26590 [at] ti3sun [dot] ti3.tu-harburg.de> To: reliable_computing [at] interval [dot] usl.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 554 X-Lines: 19 People like to have free interval S/W, so here is my add: We distribute PROFIL, a very fast library for interval arithmetic. Many people use it, and everybody is welcome to join. PROFIL is growing, currently available for many PCs and workstations. Information and source code is available via anonymous ftp ti3sun.ti3.tu-harburg.de in directory /pub/profil Speed and comparisons can be found in O. Knueppel: PROFIL/BIAS - A Fast Interval Library, COMPUTING, Vol. 53, No. 3-4, p. 277-287. Best regards, Siegfried M. Rump From owner-reliable_computing Fri Feb 23 03:34:00 1996 Received: by interval.usl.edu id AA06017 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 23 Feb 1996 09:34:14 -0600 Received: by interval.usl.edu id AA06007 (5.65c/IDA-1.4.4 for reliable_computing); Fri, 23 Feb 1996 09:34:00 -0600 Date: Fri, 23 Feb 1996 09:34:00 -0600 From: "Kearfott R. Baker" Message-Id: <199602231534.AA06007 [at] interval [dot] usl.edu> To: reliable_computing Subject: interval software in "C" Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 36 X-Lines: 2 ----- Begin Included Message ----- From rump@tu-harburg.d400.de Fri Feb 23 05:28:05 1996 Date: Fri, 23 Feb 96 12:24:55 +0100 From: rump@tu-harburg.d400.de ( Prof.Dr.S.M. Rump) To: rbk5287 [at] interval [dot] usl.edu Content-Length: 590 Status: O X-Lines: 23 People like to have free interval S/W, so here is my add: We distribute PROFIL, a very fast library for interval arithmetic. Many people use it, and everybody is welcome to join. PROFIL is growing, currently available for many PCs and workstations. Information and source code is available via anonymous ftp ti3sun.ti3.tu-harburg.de in directory /pub/profil Speed and comparisons can be found in O. Knueppel: PROFIL/BIAS - A Fast Interval Library, COMPUTING, Vol. 53, No. 3-4, p. 277-287. Best regards, Siegfried M. Rump ----- End Included Message ----- From owner-reliable_computing Fri Feb 23 01:51:19 1996 Received: by interval.usl.edu id AA06128 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 23 Feb 1996 09:51:28 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA06122 (5.65c/IDA-1.4.4 for ); Fri, 23 Feb 1996 09:51:23 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA18987; Fri, 23 Feb 96 08:51:19 MST Date: Fri, 23 Feb 96 08:51:19 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602231551.AA18987 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: let's add intervals to Java: a proposal Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 4673 X-Lines: 114 LET'S ADD INTERVALS TO JAVA: A PROPOSAL WE WANT COMPUTERS TO GENERATE GUARANTEED COMPUTATION RESULTS. Our interval community unites researchers and practitioners who understand that the existing numerical computations techniques often produce "educated guesses" rather than guaranteed results, and that in many applications, it is vitally important to have GUARANTEED results. For example, when we launch a spaceship to the Moon, it is not sufficient to know that the approximately computed trajectory leads to the Moon; we must be 100% sure that the actual trajectory brings the spaceship there. THEREFORE, WE NEED COMPUTER SUPPORT FOR INTERVAL COMPUTATIONS. Guaranteed means, e.g., that in addition to the approximate value, we must know the bound on its accuracy; in other words, we must know an INTERVAL that is guaranteed to contain the actual value. Hence, we need computer support for interval computations. CURRENTLY, THE MAINSTREAM PROGRAMMING LANGUAGES DO NOT SUPPORT INTERVAL COMPUTATIONS. As a result, lots of effort went into producing interval computation packages and interval versions of the mainstream languages (for detailed information, see, e.g., interval computations website http://cs.utep.edu/interval-comp/main.html). IDEALLY, INTERVAL COMPUTATIONS SHOULD BE PART OF MAINSTREAM LANGUAGES. In particular, R. Baker Kearfott and other researchers are currently promoting the idea of including interval computations in the new version of Fortran. Another language is now appearing on the horizon: JAVA: THE LANGUAGE OF THE FUTURE? Currently, there is a new language called Java that is spreading like wildfire. It is, crudely speaking, a version of C++ that is tailored towards the World Wide Web and parallel computations. This language is getting more and more popular with people doing applications, e.g., with chemical engineers, industrial engineers, etc., because with Java, they can not only run their programs, but they can also produce the results in a nice dynamic format directly into the Web page. This language is actively supported by Sun Company, that distributes it for free. The detailed information about Java can be found on the Web,at http://java.sun.com. Many researchers believe that it is the language of the future. LET'S ADD INTERVALS TO JAVA It seems reasonable to add interval computations to Java. A specific feature of Java that makes interval computations especially important for Java is that Java is a platform-independent language. Running the same program on different platforms with different computer precisions, etc., often leads to different results. If we do not have intervals available to bound numerical errors, users will not know, when they get different results on different platforms, whether the differences are the result of a bug of some sort, or just numerical instability. AT SUN, THERE HAS ALREADY BEEN AN INTERNAL EFFORT TO ADD INTERVALS TO JAVA. This effort is mainly promoted by Bill Walster, Sun's leading interval researcher. HELP IS NEEDED. IDEAS ARE NEEDED. SUPPORT IS NEEDED. However, the language is still, basically, in the design stage. Language designers need feedback from the interval community: * examples and arguments to help them make an intelligent decision on whether adding intervals to Java is a good idea; * specific (and general) ideas of what and how to add; * ideally, proposals for help and collaboration with the Sun Java design team. It is especially important that the Sun team hears from potential users (customers) who are interested in applying interval computations to real-life problems. Sun is making a serious effort to accommodate potential users, and it will definitely respond to the customer input. WHO TO CONTACT. If you have specific ideas and/or suggestions, please send them to Bill Walster at Bill.Walster [at] eng [dot] sun.com and to David Hough, the most senior floating point person at Sun, at david.hough [at] eng [dot] sun.com (both are on reliable computing mailing list). Also, please mail a copy of your mail to Vladik Kreinovich at vladik [at] cs [dot] utep.edu. On Bill Walter's suggestion, and with his help, we will try prepare a formal summary of the interval community's opinion that will be sent to the Sun team. If you simply want to express support and/or you feel that your ideas are too specific to send them to Sun team leaders, please mail your support/information to Vladik Kreinovich for inclusion into the summary. The more input, the better. THANKS FOR YOUR HELP. P.S. This plead for help is also posted on the Interval Computations website, in the Languages section. From owner-reliable_computing Fri Feb 23 02:32:57 1996 Received: by interval.usl.edu id AA06495 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 23 Feb 1996 10:33:06 -0600 Received: from cs.utep.edu by interval.usl.edu with SMTP id AA06487 (5.65c/IDA-1.4.4 for ); Fri, 23 Feb 1996 10:33:00 -0600 Received: from earth.cs.utep.edu by cs.utep.edu (4.1/SMI-4.1) id AA19606; Fri, 23 Feb 96 09:32:57 MST Date: Fri, 23 Feb 96 09:32:57 MST From: vladik [at] cs [dot] utep.edu (Vladik Kreinovich) Message-Id: <9602231632.AA19606 [at] cs [dot] utep.edu> To: reliable_computing [at] interval [dot] usl.edu Subject: interval software on the Web Cc: mkosh [at] cs [dot] utep.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1930 X-Lines: 54 1) Just a reminder for those who have access to the Web: links to different interval software packages (including PROFIL) and different interval languages can be found on the interval computations website URL http://cs.utep.edu/interval-comp/main.html This URL gives a contents page, from where clicking on "Interval Software" or "Languages on Interval Analysis" brings you to the corresponding pages. 2) If you have (or know of) interval software or language that is not mentioned or linked in our website please send me the info and/or URL. 3) This website also has the Personalia section (links to the homepages of all interval researchers) and the section with links to homepages of interval research centers. If you know of any interval researcher whose homepage is not mentioned please send me the info. Arnold Neumaier also suggested that these two lists would be even more useful if they were equipped with short descriptions of the specific fields of interest of a researcher (center). I would therefore ask those whose homepages are there to please send me such short descriptions that most adequately describe your own research areas. Once again, we would like to thank everyone who helped to maintain the pages. Thanks for your help. Yours Vladik Vladik Kreinovich co-maintainer, Interval Computations website > People like to have free interval S/W, so here is my add: > > We distribute PROFIL, a very fast library for interval > arithmetic. Many people use it, and everybody is welcome to join. > PROFIL is growing, currently available for many PCs and > workstations. > > Information and source code is available via anonymous ftp > > ti3sun.ti3.tu-harburg.de in directory /pub/profil > > Speed and comparisons can be found in > > O. Knueppel: PROFIL/BIAS - A Fast Interval Library, > COMPUTING, Vol. 53, No. 3-4, p. 277-287. > > Best regards, Siegfried M. Rump > > From owner-reliable_computing Fri Feb 23 09:24:18 1996 Received: by interval.usl.edu id AA07111 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Fri, 23 Feb 1996 13:24:37 -0600 Received: from helix.nih.gov by interval.usl.edu with SMTP id AA07105 (5.65c/IDA-1.4.4 for ); Fri, 23 Feb 1996 13:24:31 -0600 Received: (from ray@localhost) by helix.nih.gov (8.6.13/8.6.12) id OAA08940; Fri, 23 Feb 1996 14:24:18 -0500 Date: Fri, 23 Feb 1996 14:24:18 -0500 (EST) From: Raymond Mejia To: reliable_computing [at] interval [dot] usl.edu Cc: ray [at] helix [dot] nih.gov Subject: Re: let's add intervals to Java: a proposal (fwd) In-Reply-To: Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 642 X-Lines: 18 A word of caution about Java. Although perhaps the wave of the future in computing languages, I believe that it is an interpreted language (as opposed to a compiled language). This may make it especially inefficient for numerical computation. You may want to look further if significant multiprocessing is considered. Regards, Raymond Mejia Laboratory of Kidney and Electrolyte Metabolism, NHLBI, and Mathematical Research Branch, NIDDK BSA/350 National Institutes of Health Bethesda, MD 20892-2690 Tel: (301)496-9972 Fax: (301)402-0535 E-mail: ray [at] helix [dot] nih.gov WWW: http://mrb.niddk.nih.gov/ray/ From owner-reliable_computing Mon Feb 26 04:54:24 1996 Received: by interval.usl.edu id AA29073 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Mon, 26 Feb 1996 10:55:08 -0600 Received: by interval.usl.edu id AA29062 (5.65c/IDA-1.4.4 for reliable_computing); Mon, 26 Feb 1996 10:54:24 -0600 Date: Mon, 26 Feb 1996 10:54:24 -0600 From: "Kearfott R. Baker" Message-Id: <199602261654.AA29062 [at] interval [dot] usl.edu> To: verdonk [at] uia [dot] ua.ac.be Subject: Re: How to get previous correspondence to reliable-computing Cc: jpd [at] ucs [dot] usl.edu, reliable_computing Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1338 X-Lines: 44 Brigitte, > Concerning mails posted to reliable_computing [at] interval [dot] usl.edu, I was > wondering if these are centrally logged somewhere ? I also read that > there is a reliable_computing-digest. Can these digests be consulted > (long) after they have been sent? > There are both a "digest" and a monthly archive. These are all kept. The majordomo software should allow you to access these. For a command summary, send a message to majordomo [at] interval [dot] usl.edu The body of the message should consist of the single line help Archives from a particular month, say, from January, 1996, are stored as reliable_computing.9601 Thus, to get the January, 1996 archive, you would send a message with the following single line as its body: get reliable_computing reliable_computing.9601 The majordomo software will then send you the file reliable_computing.9601, and send you notification that it has done so in a separate message. Best regards, Baker --------------------------------------------------------------- R. Baker Kearfott, rbk [at] usl [dot] edu (318) 482-5346 (fax) (318) 482-5270 (work) (318) 981-9744 (home) URL: ftp://interval.usl.edu/pub/interval_math/www/kearfott.html Department of Mathematics, University of Southwestern Louisiana --------------------------------------------------------------- From owner-reliable_computing Tue Feb 27 06:53:47 1996 Received: by interval.usl.edu id AA00851 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Tue, 27 Feb 1996 12:53:49 -0600 Received: by interval.usl.edu id AA00841 (5.65c/IDA-1.4.4 for reliable_computing); Tue, 27 Feb 1996 12:53:47 -0600 Date: Tue, 27 Feb 1996 12:53:47 -0600 From: "Kearfott R. Baker" Message-Id: <199602271853.AA00841 [at] interval [dot] usl.edu> To: reliable_computing Subject: Proceedings of APIC'95 now fully electronic on WWW Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 886 X-Lines: 19 Vladik Kreinovich has just informed me that the abstracts of APIC'95 (international workshop on Applications of Interval Computations) are now accessible from the interval home page at http://cs.utep/edu/interval-comp/main.html >From this page, select "previous conferences," then select "Proceedings of APIC'95". The individual abstracts, in either LaTeX or Postscript, can now be retrieved by clicking on lines of the table of contents for the Proceedings. Thank you to Vladik and his son Misha. --------------------------------------------------------------- R. Baker Kearfott, rbk [at] usl [dot] edu (318) 482-5346 (fax) (318) 482-5270 (work) (318) 981-9744 (home) URL: ftp://interval.usl.edu/pub/interval_math/www/kearfott.html Department of Mathematics, University of Southwestern Louisiana --------------------------------------------------------------- From owner-reliable_computing Tue Feb 27 08:22:16 1996 Received: by interval.usl.edu id AA02106 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Tue, 27 Feb 1996 18:26:14 -0600 Received: from cronopio.ibase.br by interval.usl.edu with SMTP id AA02100 (5.65c/IDA-1.4.4 for ); Tue, 27 Feb 1996 18:26:09 -0600 Received: from fama.ibase.br (fama.ibase.br [200.18.178.14]) by cronopio.ibase.br (8.6.12/Revision: 1.203 ) with SMTP id VAA18361 for ; Tue, 27 Feb 1996 21:22:15 -0300 From: Received: (from cnc) by fama.ibase.br (8.6.12/Revision: 1.6 ) id QAA10644 for reliable_computing [at] interval [dot] usl.edu; Tue, 27 Feb 1996 16:22:16 -0800 Date: Tue, 27 Feb 1996 16:22:16 -0800 Message-Id: <199602280022.QAA10644 [at] fama [dot] ibase.br> To: reliable_computing [at] interval [dot] usl.edu Subject: Unsubsciption help Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 40 X-Lines: 1 How dHow do i unsuscribe from this list From owner-reliable_computing Wed Feb 28 15:33:23 1996 Received: by interval.usl.edu id AA03167 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 28 Feb 1996 07:37:35 -0600 Received: from rztsun3.rz.tu-harburg.de by interval.usl.edu with SMTP id AA03160 (5.65c/IDA-1.4.4 for ); Wed, 28 Feb 1996 07:37:11 -0600 Date: Wed, 28 Feb 96 14:33:23 +0100 From: rump@tu-harburg.d400.de ( Prof.Dr.S.M. Rump) Message-Id: <9602281333.AA02295 [at] ti3sun [dot] ti3.tu-harburg.de> To: reliable_computing [at] interval [dot] usl.edu Sender: owner-reliable_computing Precedence: bulk Status: RO Content-Length: 3368 X-Lines: 80 subject: back to the roots In the discussion about global optimization methods it has been mentioned that recently developed and very effective methods are i) the combination of local search methods with branch-and-bound techniques together with interval methods, and ii) the use of the so-called expansion principle. The second method proves existence and uniqueness of a stationary point first in a small interval, and then expands this interval again proving that there is one and only one stationary point in the larger interval. Therefore the difference between both sets cannot contain a stationary point and can be discarded. If something proves to be useful and one has used it frequently enough, it seems to be natural to forget about the roots. Therefore I would like to mention that to all of my knowledge in this case Dr. Christian Jansson is the origin. He developed many interesting algorithms for the one- and n-dimensional case and gave a number of talks about this starting in 1990. To my knowledge he was the first who combined local search algorithms with validation algorithms, and he invented the expansion principle. There is a written report on both techniques dated November 1991. The reference is C. Jansson: A Global Minimization Method: The One-Dimensional Case, Technical Report 91.2, Forschungsschwerpunkt Informations- und Kommunikationstechnik, Technical University Hamburg-Harburg, 1991. In this report Jansson showed that on well-known examples from the literature his method with verified bounds for the global optimum is comparable in speed with purely numerical methods (without any verification). In fact, sometimes his methods with verification was even faster. He compared to the well-known algorithms by Zilinskas, Brent, Batishev, Strongin and others. A large collection of numerical results is contained in C. Jansson and O. Knppel: A Global Minimization Method: The Multi-Dimensional Case, Technical Report 92.1, Forschungsschwerpunkt Informations- und Kommunikationstechnik, Technical University Hamburg-Harburg, 1992. C. Jansson and O. Knppel: Numerical results for a Self-Validating Global Optimization Method, Technical Report 94.1, Forschungsschwerpunkt Informations- und Kommunikationstechnik, Technical University Hamburg-Harburg, 1994. The reports contain numerical results of about 50 test problems, including the test set of Hansen. Jansson achieves a remarkable speed up compared to existing algorithms. Even the global optimum of the Griewank function for dimension 50 (!) is calculated with verification (!) in less than 3 minutes on a SUN Sparc Station 1 with accuracy better than 15 decimal places. There is only one reference of a pure floating point algorithm in the literature being able to provide a solution for that problem for dimension 10 . Further references include C. Jansson: On Self-Validating Methods for Optimization Problems, in J. Herzberger (ed.): Topics in Validated Computations, Elsevier Science B.V., pp. 381-438, 1994. C. Jansson and O. Knppel: A Branch and Bound Algorithm for Bound Constrained Optimization Problems without Derivatives, Journal of Global Optimization 7, pp. 297-333, 1995. Regards, S.M. Rump From owner-reliable_computing Wed Feb 28 03:12:52 1996 Received: by interval.usl.edu id AA03538 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 28 Feb 1996 09:12:54 -0600 Received: by interval.usl.edu id AA03528 (5.65c/IDA-1.4.4 for reliable_computing); Wed, 28 Feb 1996 09:12:52 -0600 Date: Wed, 28 Feb 1996 09:12:52 -0600 From: "Kearfott R. Baker" Message-Id: <199602281512.AA03528 [at] interval [dot] usl.edu> To: reliable_computing Subject: history of using local optimizers Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 5149 X-Lines: 125 While we are on the subject, I would like to point out (at the risk of sounding self-serving) that I proposed a method to eliminate approximate roots from the search region in 1989. The idea was that, for singular roots, an interval Newton method wouldn't work, so regions around such roots needed to be eliminated. I used a classical Newton method to first locate such roots approximately, then expanded around such approximate roots to eliminate the region. This is explained in the section dealing with "trisection" in the following reference (in BiBTeX form): @ARTICLE{, AUTHOR = {Kearfott, R. B.}, TITLE = {On Handling Singular Systems with Interval Newton Methods}, JOURNAL = {IMACS Annals on Computing and Applied Mathematics}, VOLUME = "1.2", PAGES ="653--655", PUBLISHER ="J. C. Baltzer AG", BOOKTITLE = {IMACS Annals on Computing and Applied Mathematics Numerical and Applied Mathematics}, EDITOR ={Brezinski, C.}, YEAR = "1989" } Of course, I am not claiming these Fortran 77 experiments on 386 PC approach the tour de force of Janssen and Kn{\"u}ppel. Do Kaj Madsen or Ole Caprani have any comments? Best regards, --------------------------------------------------------------- R. Baker Kearfott, rbk [at] usl [dot] edu (318) 482-5346 (fax) (318) 482-5270 (work) (318) 981-9744 (home) URL: ftp://interval.usl.edu/pub/interval_math/www/kearfott.html Department of Mathematics, University of Southwestern Louisiana --------------------------------------------------------------- > Date: Wed, 28 Feb 96 14:33:47 +0100 > From: rump@tu-harburg.d400.de ( Prof.Dr.S.M. Rump) > > subject: back to the roots > > > In the discussion about global optimization methods it has been > mentioned that recently developed and very effective methods are > > i) the combination of local search methods with > branch-and-bound techniques together with interval methods, > and ii) the use of the so-called expansion principle. > > The second method proves existence and uniqueness of a stationary > point first in a small interval, and then expands this interval > again proving that there is one and only one stationary point in > the larger interval. Therefore the difference between both sets > cannot contain a stationary point and can be discarded. > > If something proves to be useful and one has used it frequently > enough, it seems to be natural to forget about the roots. > > Therefore I would like to mention that to all of my knowledge > in this case Dr. Christian Jansson is the origin. He developed > many interesting algorithms for the one- and n-dimensional case > and gave a number of talks about this starting in 1990. > > To my knowledge he was the first who combined local search > algorithms with validation algorithms, and he invented the > expansion principle. There is a written report on both techniques > dated November 1991. The reference is > > C. Jansson: A Global Minimization Method: The One-Dimensional > Case, Technical Report 91.2, Forschungsschwerpunkt > Informations- und Kommunikationstechnik, Technical > University Hamburg-Harburg, 1991. > > In this report Jansson showed that on well-known examples from > the literature his method with verified bounds for the global > optimum is comparable in speed with purely numerical methods > (without any verification). In fact, sometimes his methods with > verification was even faster. He compared to the well-known > algorithms by Zilinskas, Brent, Batishev, Strongin and others. > > A large collection of numerical results is contained in > > C. Jansson and O. Knppel: A Global Minimization Method: The > Multi-Dimensional Case, Technical Report 92.1, > Forschungsschwerpunkt Informations- und > Kommunikationstechnik, Technical University > Hamburg-Harburg, 1992. > > C. Jansson and O. Knppel: Numerical results for a > Self-Validating Global Optimization Method, Technical > Report 94.1, Forschungsschwerpunkt Informations- und > Kommunikationstechnik, Technical University > Hamburg-Harburg, 1994. > > The reports contain numerical results of about 50 test problems, > including the test set of Hansen. Jansson achieves a remarkable > speed up compared to existing algorithms. Even the global optimum > of the Griewank function for dimension 50 (!) is calculated with > verification (!) in less than 3 minutes on a SUN Sparc Station 1 > with accuracy better than 15 decimal places. There is only one > reference of a pure floating point algorithm in the literature > being able to provide a solution for that problem for dimension > 10 . > > Further references include > > C. Jansson: On Self-Validating Methods for Optimization > Problems, in J. Herzberger (ed.): Topics in Validated > Computations, Elsevier Science B.V., pp. 381-438, 1994. > > C. Jansson and O. Knppel: A Branch and Bound Algorithm for > Bound Constrained Optimization Problems without > Derivatives, Journal of Global Optimization 7, pp. > 297-333, 1995. > > > > Regards, S.M. Rump > > From owner-reliable_computing Wed Feb 28 06:27:34 1996 Received: by interval.usl.edu id AA03832 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 28 Feb 1996 10:27:46 -0600 Received: from helix.nih.gov by interval.usl.edu with SMTP id AA03826 (5.65c/IDA-1.4.4 for ); Wed, 28 Feb 1996 10:27:43 -0600 Received: (from ray@localhost) by helix.nih.gov (8.6.13/8.6.12) id LAA17488; Wed, 28 Feb 1996 11:27:35 -0500 Date: Wed, 28 Feb 1996 11:27:34 -0500 (EST) From: Raymond Mejia To: reliable_computing [at] interval [dot] usl.edu Subject: Re: history of using local optimizers In-Reply-To: <199602281512.AA03528 [at] interval [dot] usl.edu> Message-Id: Mime-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 2866 X-Lines: 84 Yes Baker, history does tend to repeat itself! There are continuation methods that have been around for well over 10 years that come to mind in the context of singular roots. Some may want to look at the following references and in turn their references: \by H. J\"urgens, H-O. Peitgen, D. Saupe \inbook Analysis and Computation of Fixed points \page 139 \publ Academic Press \yr 1980 \by R. Mejia \yr 1986 \paper CONKUB: A Conversational Path-Follower for Systems of Nonlinear Equations \jour J. Comput. Phys.\vol 63 \pages 67--84 \by R. Mejia \paper Interactive Program for Continuation of Solutions of Large Systems of Nonlinear Equations \inbook Computational Solution of Nonlinear Systems of Equations (E. L. Allgower and K. Georg, Eds.) \vol 26 \bookinfo Lectures in Applied Mathematics \publ Amer. Math. Soc. \publaddr Providence, RI \pages 429--449 \yr 1990 Best regards, Ray Raymond Mejia Laboratory of Kidney and Electrolyte Metabolism, NHLBI, and Mathematical Research Branch, NIDDK BSA/350 National Institutes of Health Bethesda, MD 20892-2690 Tel: (301)496-9972 Fax: (301)402-0535 E-mail: ray [at] helix [dot] nih.gov WWW: http://mrb.niddk.nih.gov/ray/ On Wed, 28 Feb 1996, Kearfott R. Baker wrote: > While we are on the subject, I would like to point out (at the > risk of sounding self-serving) that I proposed a method to > eliminate approximate roots from the search region in 1989. > The idea was that, for singular roots, an interval Newton > method wouldn't work, so regions around such roots needed to > be eliminated. I used a classical Newton method to first > locate such roots approximately, then expanded around such > approximate roots to eliminate the region. This is explained > in the section dealing with "trisection" in the following > reference (in BiBTeX form): > > @ARTICLE{, > AUTHOR = {Kearfott, R. B.}, > TITLE = {On Handling Singular Systems with Interval Newton > Methods}, > JOURNAL = {IMACS Annals on Computing and Applied Mathematics}, > VOLUME = "1.2", > PAGES ="653--655", > PUBLISHER ="J. C. Baltzer AG", > BOOKTITLE = {IMACS Annals on Computing and Applied Mathematics > Numerical > and Applied Mathematics}, > EDITOR ={Brezinski, C.}, > YEAR = "1989" > } > > Of course, I am not claiming these Fortran 77 experiments on 386 > PC approach the tour de force of Janssen and Kn{\"u}ppel. > > Do Kaj Madsen or Ole Caprani have any comments? > > Best regards, > > > --------------------------------------------------------------- > R. Baker Kearfott, rbk [at] usl [dot] edu (318) 482-5346 (fax) > (318) 482-5270 (work) (318) 981-9744 (home) > URL: ftp://interval.usl.edu/pub/interval_math/www/kearfott.html > Department of Mathematics, University of Southwestern Louisiana > --------------------------------------------------------------- > From owner-reliable_computing Wed Feb 28 04:36:29 1996 Received: by interval.usl.edu id AA03904 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 28 Feb 1996 10:36:52 -0600 Received: from leonhard.ma.utexas.edu by interval.usl.edu with SMTP id AA03890 (5.65c/IDA-1.4.4 for ); Wed, 28 Feb 1996 10:36:40 -0600 Date: Wed, 28 Feb 96 10:36:29 CST From: koch [at] fireant [dot] ma.utexas.edu Posted-Date: Wed, 28 Feb 96 10:36:29 CST Message-Id: <9602281636.AA03572 [at] leonhard [dot] ma.utexas.edu> Received: by leonhard.ma.utexas.edu (4.1/5.51) id AA03572; Wed, 28 Feb 96 10:36:29 CST To: reliable_computing [at] interval [dot] usl.edu Cc: koch [at] leonhard [dot] ma.utexas.edu Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 468 X-Lines: 20 Can somebody please get me off this mailing list. Thanks - Hans Koch Date: Wed, 28 Feb 1996 10:24:27 -0600 Message-Id: <199602281624.AA03805 [at] interval [dot] usl.edu> To: koch [at] fireant [dot] ma.utexas.edu From: Majordomo [at] interval [dot] usl.edu Subject: Majordomo results: unsubscribe reliable_computing Reply-To: Majordomo [at] interval [dot] usl.edu Status: R -- >>>> unsubscribe reliable_computing **** unsubscribe: 'koch [at] fireant [dot] ma.utexas.edu' is not a member of list 'reliable_computing'. From owner-reliable_computing Wed Feb 28 18:30:04 1996 Received: by interval.usl.edu id AA04523 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Wed, 28 Feb 1996 12:29:12 -0600 Received: from Post-Office.UH.EDU by interval.usl.edu with SMTP id AA04517 (5.65c/IDA-1.4.4 for ); Wed, 28 Feb 1996 12:29:09 -0600 Received: from happy.dt.uh.edu by Post-Office.UH.EDU (PMDF V5.0-6 #8380) id <01I1QW75THXM0004ZD@Post-Office.UH.EDU>; Wed, 28 Feb 1996 12:29:04 -0600 (CST) Received: by happy.dt.uh.edu (5.0/SMI-SVR4) id AA21680; Wed, 28 Feb 1996 12:30:04 +0600 Date: Wed, 28 Feb 1996 12:30:04 +0600 From: hu [at] happy [dot] dt.uh.edu Subject: Interval BLAS To: reliable_computing [at] interval [dot] usl.edu Cc: dongarra [at] cs [dot] utk.edu Message-Id: <9602281830.AA21680 [at] happy [dot] dt.uh.edu> Content-Transfer-Encoding: 7BIT X-Sun-Charset: US-ASCII Sender: owner-reliable_computing Precedence: bulk Status: O Content-Length: 1208 X-Lines: 30 As a member of the interval computations community, I've attended the BLAST Forum in Nashville, TN, organized by Jack Dongarra last week. The Forum is trying to establish technical standards for BLAS, ParallelBLAS, XtendedBLAS in the next 1-2 years. Special issues decussed there involve functionalities, interface layers, language specifications, etc. I mentioned interval BLAS at the meeting to the attendees, and suggested form an INTBLAS (INTerval BLAS) subgroup associated with the BLAST Forum. Professor Baker Kearfott supports the INTBLAS idea and suggests to solicit members through the reliable_computing mailing list. Researchers interested in joining the INTBLAS subgroup, please send me a reply, I'll form a mailing list. Thanks for your attention! Prof. Chenyi Hu Center for Computational Sciences and Advanced Distributed Simulation University of Houston-Downtown Houston, Texas 77002 P.S. Information on the BLAST can be found in NA Digest, V.96, #06. For background information of the forum see: http://www.netlib.org/utk/papers/sblas-meeting.html. For additional information of the forum see: http://www.netlib.org/utk/papers/blast-forum.html. From owner-reliable_computing Thu Mar 1 06:52:38 1996 Received: by interval.usl.edu id AA06812 (5.65c/IDA-1.4.4 for reliable_computing-outgoing); Thu, 29 Feb 1996 12:52:43 -0600 Received: from bp.ucs.usl.edu by interval.usl.edu with SMTP id AA06806 (5.65c/IDA-1.4.4 for ); Thu, 29 Feb 1996 12:52:41 -0600 Received: from rbk5287.usl.edu by bp.ucs.usl.edu with SMTP id AA12321 (5.65c/IDA-1.4.4 for ); Thu, 29 Feb 1996 12:52:38 -0600 Date: Thu, 29 Feb 1996 12:52:38 -0600 Message-Id: <199602291852.AA12321 [at] bp [dot] ucs.usl.edu> X-Sender: rbk5287 [at] pop [dot] usl.edu X-Mailer: Windows Eudora Light Version 1.5.2 Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" To: reliable_computing [at] interval [dot] usl.edu From: Kaj Madsen (by way of "R. Baker Kearfott" ) Subject: history of using local optimizers Sender: owner-reliable_computing Precedence: bulk Content-Length: 814 X-Lines: 19 Status: O Ole Caprani and I published the idea of using a real iteration in connection with interval expansion in the paper Iterative Methods for Interval Inclusion of Fixed Points. BIT 18 (1978), pp 42-51. In that paper we cited the following elegant formulation of J.H. Wilkinson (SIAM Rev. 13 (1971), pp 548-568): "In general it is the best in algebraic computations to leave the use of interval arithmetic as late as possible so that it effectively becomes an a posteriori weapon." In 1993 we published the idea in connection with optimization without knowing about Jansson's results. It was at the Lafayette meeting and in its proceedings, Use of a Real-Valued Local Minimum in Parallel Global Optimization. Interval Computations 2 (1993), pp 71-82. Kaj Madsen