Subject: NA Digest, V. 92, # 13 NA Digest Sunday, March 29, 1992 Volume 92 : Issue 13 Today's Editor: Cleve Moler The MathWorks, Inc. moler@mathworks.com Today's Topics: Fast Hartley Transform Mesh Generation Bibliography Who Solves ODEs of the form y''=f(x,y)? Three Dimensional Singular Elements Temporary Change of Address for Jim Bunch New Jersey Section of SIAM Spring Meeting Conference on Approximation and Probability 35th British Theoretical Mechanics Colloquium Parlett-Kahan Meeting XI Parallel Circus Positions at University of Vermont Visiting Positions at Arizona State Contents: SIAM Scientific and Statistical Computing Submissions for NA Digest: Mail to na.digest@na-net.ornl.gov. Information about NA-NET: Mail to na.help@na-net.ornl.gov. ------------------------------------------------------- From: David H. Bailey Date: Thu, 19 Mar 92 09:26:00 -0800 Subject: Fast Hartley Transform Readers of the NA Net Digest may have noted the recent posting of a seminar on the Hartley transform. It is widely believed, and sometimes stated or implied in print, that for purely real inputs the fast Hartley transform (FHT) is twice as fast as the fast Fourier transform (FFT), since the FHT operates with real data instead of complex data. However, there exist well-known variants of the FFT that compute the discrete Fourier transform (DFT) on real input data in only half the operations required for the standard complex FFT. One way is to employ formulas that convert an FFT on real input data into a complex FFT of half the size. An even more efficient scheme is to use Edson's algorithm, which computes the DFT result using real arithmetic, and which does not require a pre- or post-processing step. Almost all vendor FFT libraries include efficient implementations of such schemes, and these routines usually run faster than implementations of the FHT. Thus the FHT is not fundamentally more efficient than well-known variants of the FFT, even for real data, and the Hartley transform itself has no known physical significance. Also, the FHT software as developed at Stanford is patented, and commercial usage requires consent from Stanford. In summary, we frankly do not understand the continuing interest in the FHT. Can anyone offer solid scientific reasons for using the FHT in place of the real-to-half complex FFT? David H. Bailey, NASA Ames Paul N. Swarztrauber, NCAR ------------------------------ From: Stephen Vavasis Date: Mon, 23 Mar 92 14:43:48 -0500 Subject: Mesh Generation Bibliography Over two years ago in this forum I posted a notice that I was interested in writing a bibliography on automatic mesh generation for finite elements. The task quickly overwhelmed me, and I never finished it. The good news is that Marshall Bern of Xerox PARC and David Eppstein of U.C. Irvine have written a very nice illustrated summary of automatic mesh generation. Their summary is in the form of an annotated book chapter, and is also available as a Xerox technical report. The summary looks at the problem primarily from the computer science viewpoint (rather than from the applications viewpoint). The title of the report is "Mesh Generation and Optimal Triangulation"; if you are interested please contact Marshall Bern directly: bern@parc.xerox.com. -- Steve Vavasis ------------------------------ From: John Coleman Date: Thu, 26 Mar 92 15:30:57 GMT Subject: Who Solves ODEs of the form y''=f(x,y)? I would like to hear from you if you solve second-order differential equations of the special form y''=f(x,y), where y and f may be vectors, or if you know of anyone who does. Many numerical methods have been proposed for these equations, in papers which may contain one or two simple problems for which method n+1 is 'better' than method n. I would like to make contact with people who solve such problems to find an answer rather than to test a method. I am particularly interested in the sub-class of problems having oscillatory solutions -- such as orbit problems in classical mechanics, quantum mechanical scattering problems governed by the radial Schrodinger equation, and perhaps others about which I know nothing. In your response please tell me: (a) the form and the origin of the problem, (b) the numerical method or methods currently used, (c) the accuracy required and how you know if you have got it, (d) what difficulties (if any) arise in the solution process, (e) how satisfactory you consider your present approach to be, (f) any relevant references. I believe that the information I seek would be of interest to several people involved in the numerical analysis of ODEs. If I get a good response I intend to submit a summary to the NA Digest. My thanks in advance to all who can contribute in any way. John Coleman. E-mail: John.Coleman @ durham.ac.uk na.jcoleman @ na-net.ornl.gov ------------------------------ From: E.V.Glushkov Date: 26 Mar 92 16:02:58 GMT Subject: Three Dimensional Singular Elements Hello from Russia, As is well-known, use of so-called singular elements in a 2-dim. finite element method improves the convergence and accuracy very much. But it is very difficult to find the order of singularity at the 3-dim. corner points, and therefore to construct the singular elements. In our lab there was eleborated, realized as a computer program and tested a method of the singularity orders at a top of an arbitrary elastic polihedron extracting. The method is based on the spectral points of some integral operators seaching. We are not specialists in FEM and, being restricted in our com- puter's output, cannot carry out a regular and thorough numerical analysis. Untill now we've obtained results for a top of a cube with one fixed side and for some other geometries. So, if somebody is interested in employing 3-dim. singularities, please contact us and we'll discuss how a co-operation could be arranged. Sincerely, Drs. E.V.Glushkov & N.V.Glushkova E-mail: evg@kgu.kuban.su Kuban State University Institute of Mechanics and Applied Mathematics Krasnodar 350640, Russia ------------------------------ From: James R. Bunch Date: Thu, 26 Mar 92 11:53:23 -0800 Subject: Temporary Change of Address for Jim Bunch I will be at the IMA during April 3 - June 6, 1992: Prof. James R. Bunch IMA, Univ. of Minnesota 514 Vincent Hall 206 Church St SE Minneapolis, MN 55455-0436 bunch@ima.umn.edu ------------------------------ From: SIAM Date: Tue, 24 Mar 92 10:25:37 EST Subject: New Jersey Section of SIAM Spring Meeting New Jersey Section of SIAM: Spring Meeting The spring meeting of the New Jersey Section of SIAM will be held on Saturday morning, April 25, 1992 from 8:30 to noon. It will be held on the Busch Campus of Rutgers University in Piscataway. There will be two distinguished speakers who will provide an excellent mix of applied mathematics and computer science: o Diane Souvaine of the Computer Science Department and DIMACS, Rutgers University, on "Finding Maximum Inscribed Triangles and Shortest Aquarium Keeper Tours Using Shortest Paths." o Bruce L. Bush of the Molecular Systems Department, Merck Research Laboratories, Rahway, NJ, on "Some Headaches in Biomolecular Modeling: Are Mathematical Remedies on the Shelf?" Mathematicians in industry are encouraged to attend the meeting. The meeting will provide a unique opportunity for applied mathematicians in local industries, research laboratories, and academic institutions to meet and share problems, methods, and solutions. The meeting is not only for SIAM members but is open to all interested people. You are encouraged to bring along your colleagues. Graduate and undergraduate students are most welcome. For a copy of the schedule of the meeting, titles and abstracts of the talks, and detailed directions, send your mailing address and request for same to siam@siam.org. Richard B. Pelz President, New Jersey Section of SIAM ------------------------------ From: George Anastassiou Date: 24 Mar 92 16:54:59 CDT Subject: Conference on Approximation and Probability The following is of concern primarily to approximation theory and probability people, but some people from numerical analysis might be interested. An international conference on "Approximation,Probability & Related Fields" Place: Univ. California at Santa Barbara Dates: 20,21,22 May 1993 Days: Thurs.,Friday,Saturday Information-Organizers: George Anastassiou Dept. Mathematical Sciences Memphis State University Memphis TN 38152 E-mail anastasg@hermes.msci.memst.edu S.T.Rachev Dept. Statistics & Applied probability program University of california at Santa Barbara Santa Barbara CA 93106 U.S.A E-mail ZARIRACH@BERNOULLI.UCSB.EDU ------------------------------ From: D. Sloan Date: Thu, 26 Mar 92 10:47 GMT Subject: 35th British Theoretical Mechanics Colloquium BRITISH APPLIED MATHEMATICS COLLOQUIUM 35th British Theoretical Mechanics Colloquium 5th - 8th April 1993 Venue: University of Strathclyde in Glasgow Start Time: Approx. 4.00pm on Monday 5th April 1993 Finish Time: Lunch on Thursday 8th April 1993 Invited Speakers Computational Mathematics Professor Bengt Fornberg (Exxon Corporate Research, NJ, USA & University of Strathclyde) Mathematical Biology Professor Bob May, FRS (University of Oxford) Solid Mechanics Professor Ingo Muller (Technische Universitat Berlin) Differential Equations Professor Larry Payne (Cornell University) Rheology Professor Ken Walters, FRS (University College of Wales, Aberystwyth) Stewartson Memorial Lecture Professor Alex Craik (University of St Andrews) Mini- Symposia Liquid Crystals Industrial Mathematics Inverse Problems Nonlinear Dynamical Systems Office Bearers Chairman Professor Frank Leslie Tel: 041 552 4400 EXT 3655 Secretary Dr. Ian Murdoch Tel: 041 552 4400 EXT 3657 Treasurer Dr. John Parkes Tel: 041 552 4400 EXT 3720 E-mail address bamc93@uk.ac.strath Postal Address: British Applied Mathematics Colloquium Department of Mathematics University of Strathclyde Glasgow G1 1XH U. K. ------------------------------ From: James R. Bunch Date: Thu, 26 Mar 92 12:04:13 -0800 Subject: Parlett-Kahan Meeting SECOND ANNOUNCEMENT THREE DECADES OF NUMERICAL LINEAR ALGEBRA AT BERKELEY: A CONFERENCE IN HONOR OF THE SIXTIETH BIRTHDAYS OF BERESFORD PARLETT AND WILLIAM KAHAN A conference will be held at MSRI in Berkeley on Saturday, October 17, 1992, in honor of the 60th birthdays of Beresford Parlett and William Kahan. The Organizing Committee consists of: James Bunch, UC San Diego, jbunch@ucsd.edu (March 28-midJune: bunch@ima.umn.edu)James Demmel, UC Berkeley, demmel@imafs.ima.umn.edu (after midJune: demmel@ wsparc.berkeley.edu) Horst Simon, NASA Ames, simon@nas.nasa.gov The speakers at the conference will be: Scott Baden, UC San Diego James Bunch, UC San Diego James Demmel, UC Berkeley Gene Golub, Stanford Anne Greenbaum, Courant Larry Nazareth, Washington State Bahram Nour-Omid, San Francisco John Reid, Rutherford Appleton Laboratory, England David Scott, Intel G. W. Stewart, Maryland Peter Tang, Argonne There will be a banquet on the evening of October 17. The banquet speaker will be Richard Lau, ONR. There will be a Special Issue of the Journal of Numerical Linear Algebra with Applications dedicated to Parlett and Kahan. The deadline for manuscripts will be October 17, 1992. James Bunch will be the editor of the Special Issue. Anyone interested in submitting a manuscript should obtain a copy of "Guidelines for Contributors" from James Bunch. HOTEL INFORMATION: We have been able to reserve only 40 rooms for the conference. Some other departments are having conferences that same weekend. You must make your own reservation directly with the hotel, but mention the conference. 20 single rooms reserved: Durant Hotel, 2600 Durant Ave, Berkeley 94704; at the campus; 510-845-8981. $75/single. (Only singles are available.) 20 rooms reserved: Marriott Hotel, 200 Marina Blvd., Berkeley; at the Berkeley Marina, 1 1/2 miles from campus, city bus service is convenient. 510-548-7920. $85 flat rate per room for a single double, triple, or quad. Other possibilities: The Women's Faculty Club, 510-642-4175, on campus. The Men's Faculty Club, 510-642-1993, on campus. The Shattuck Hotel, 510-845-7300, downtown Berkeley. (They weren't willing to reserve a block of rooms.) There are various motels in the area. We will get a list later. ------------------------------ From: Gene Golub Date: Thu, 26 Mar 92 20:52:42 CST Subject: XI Parallel Circus XIth Parallel Circus sponsored by the Department of Computer Science, Cray Research, Inc., and the Minnesota Supercomputer Institute April 24-25, 1992 The Department of Computer Science, Cray Research, Inc., and the Minnesota Supercomputer Institute, continuing the tradition started at Yale University in 1986, will be hosting the XIth Parallel Circus at the Minnesota Supercomputer Institute in Minneapolis, MN on Friday and Saturday, April 24-25, 1992. The Parallel Circus is an informal meeting which emphasizes parallel algorithms for scientific computing. There is no set agenda. At the beginning of each session the attendees reach a consensus as to each day's program of presentations. This format allows attendees to discuss the very latest results as well as interesting work in progress. (The Tenth Circus is described in the NA Digest, v 92, # 10, March 8, 1992.) Graduate students are especially welcome to attend. There is modest support from the National Science Foundation for student travel to the Parallel Circus XI. Those students requesting support should give reasons for attending the meeting, and a budget for expenses. The student(s) should indicate their research interests and plans. A letter verifying that the student is in good standing should be sent independently by a faculty adviser. This letter should give the student's GPA. We will be pleased to consider joint proposals which would include the expenses of several students. The dealine for application has been extended to April 1, 1992. Correspondence by e-mail is desirable. FAX: (612) 625-0572 (Write "Student Support" on Fax.) E-mail: circus92@cs.umn.edu REGISTRATION To register for XIth Parallel Circus, contact: Michael Olesen Symposium Administrator Minnesota Supercomputer Institute 1200 Washington Avenue South Minneapolis, MN 55415 Tel: (612) 624-1356 FAX: (612) 624-8861 e-mail: mikeo@s1.msi.edu.umn The registration fee of $25 includes lunches on both days of the symposium as well as a banquet the evening of Friday, April 24, 1992. Registration is required by April 10, 1992. Post deadline registrations will be accepted on a space available basis. TRANSPORTATION Special conference airfares are available through Daisy Travel of St. Paul, MN. For details or to make reservations call toll free (800) 553-1660 (8:00 a.m. to 4:30 p.m. CST) and mention that you are participating in the XIth Parallel Circus at the Minnesota Supercomputer Institute. ACCOMMODATIONS Arrangements have been made for a block of discount rooms at the Days Inn University, 2407 niversity Avenue S.E., Minneapolis, MN 55414, for April 23, 24, and 25. The price per night is $39 for single occupancy and $45 for double. To receive these rates contact them at (612) 623-3999 and mention that you are participating in the XIth Parallel Circus at the Minnesota Supercomputer Institute. PRESENTATIONS Each presentation will be about 30 minutes long, 25 minutes plus 5 minutes for questions and discussions. Their actual length will depend on the number of participants wishing to make presentations. The Circus will begin on Friday morning at 9:00 a.m. Although the program is not yet set, it will probably conclude early Saturday afternoon. Organizers: Gene Golub, Bill Harrad, Ahmed Sameh, Apostolos Gerasoulis Local Committee: Don Truhlar, Michael Olesen ------------------------------ From: U. Vermont Date: Thu, 26 Mar 92 22:08:07 GMT Subject: Positions at University of Vermont TWO TENURE TRACK ASSISTANT PROFESSOR POSITIONS IN COMPUTER SCIENCE THE UNIVERSITY OF VERMONT BURLINGTON, VERMONT Applications are invited for two tenure-track faculty positions in Computer Science at the level of Assistant Professor beginning in the l992-93 academic year. Responsibilities will include instruction in mainstream computer science and the development of a quality research program. Candidates should show promise of excellence in both teaching and research; have demonstrable expertise in networks and distributed systems, parallel algorithms and systems, or database and knowledge base systems; and have a strong interest in interdisciplinary research in the mathematical sciences. Faculty are encouraged to supervise graduate students in related fields as well as in computer science. A doctorate in computer science or a closely related field is required. Applications will be accepted until the positions are filled. Please submit resume and description of current research interests, and have three letters of recommendation sent directly to Dr. Richard Foote, Search Committee Chairperson, 101 Votey Building, College of Engineering & Mathematics, University of Vermont, Burlington, VT 05405. Inquiries may be made by mail to the above address or by email to cssrch@uvm.edu. The University of Vermont is an Affirmative Action/Equal Opportunity employer and encourages applications from women and members of minority groups. ENDOWED CHAIR POSITION IN COMPUTER SCIENCE THE UNIVERSITY OF VERMONT BURLINGTON, VERMONT Applications are sought in Computer Science for the Dorothean Chair in the College of Engineering and Mathematics beginning in the 1992-93 academic year. It is anticipated that the position will be filled at the level of Full Professor. The successful candidate is expected to assume a leadership role, teach mainstream computer science, and develop an externally funded research program. An established record of excellence in teaching and research in computer science is required. Candidates should have demonstrable expertise in networks and distributed systems, database and knowledge base systems, or parallel algorithms and systems, together with a strong interest in interdisciplinary research in the mathematical sciences. Faculty are encouraged to supervise graduate students in related fields as well as in computer science. A doctorate in computer science or a closely related field is required. Applications will be accepted until the position is filled. Please submit resume and description of current research interests, and have three letters of recommendation sent directly to Dorothean Search Committee, 101 Votey Building, College of Engineering & Mathematics, University of Vermont, Burlington, VT 05405. Inquiries may be made by mail to the above address or by email to dorsrch@uvm.edu. The University of Vermont is an Affirmative Action/Equal Opportunity employer and encourages applications from women and members of minority groups. ------------------------------ From: Hans Mittelmann Date: Thu, 26 Mar 92 17:15:20 mst Subject: Visiting Positions at Arizona State VISITING POSITIONS AT ARIZONA STATE UNIVERSITY There is some chance that the Department of Mathematics at Arizona State University may be able to support one or more visiting faculty for the 1992/93 academic year (8/16/92-5/15/93). While these visitors are needed to cover the department's teaching obli- gations and probably will teach 3-4 courses during the year, preference will be given to those candidates that contribute also to the department's research efforts. In particular, anyone interested in cooperation with members of the Computational Mathematics Group (Feldstein, Jackiewicz, Mittelmann, Ringhofer, Welfert; Renaut will be on leave) should send their vita to my attention. By e-mail (TeX, LaTeX, ASCII): na.mittelmann or mittelmann@math.la.asu.edu or by FAX (602) 965 8119. Hans D. Mittelmann Department of Mathematics Arizona State University Tempe, AZ 85287-1804 ------------------------------ From: Beth Gallagher Date: Mon, 23 Mar 92 12:22:15 EST Subject: Contents: SIAM Scientific and Statistical Computing CONTENTS SIAM Journal on Scientific and Statistical Computing July 1992 Volume 13, Number 4 Parallel Methods for Solving Nonlinear Block Bordered Systems of Equations Xiaodong Zhang, Richard H. Byrd, and Robert B. Schnabel Solution of Structured Geomotric Programs in Sample Survey Design Faiz A. Al-Khayyal, Thom J. Hodgson, Grant D. Capps, James A. Dorsch, David A. Kriegman, and Paul D. Pavnica An Efficient Scheme for Unsteady Flow Past an Object with Boundary Conformal to a Circle Mo-Hong Chou Block M-Matrices and Computation of Invariant Tori Luca Dieci and Jens Lorenz Analysis of Initial Transient Deletion for Parallel Steady-State Simulations Peter W. Glynn and Philip Heidelberger An Implementation of the Fast Multipole Method without Multipoles Christopher R. Anderson On the Spectrum of a Family of Preconditioned Block Toeplitz Matrices Takang Ku and C.-C. Jay Kuo Domain Decomposition with Local Mesh Refinement William D. Gropp and David E. Keyes An O(n log n) Time Algorithm for the Minmax Angle Triangulation Herbert Edelsbrunner, Tiow Seng Tan, and Roman Waupotitsch Which Cubic Spline Should One Use? R. K. Beatson and E. Chacko Integrating Products of B-Splines A. H. Vermeulen, R. H. Bartels, and G. R. Heppler ------------------------------ End of NA Digest ************************** -------