|
he "1999 International Conference on Preconditioning Techniques for Large Sparse Matrix Problems in Industrial Applications" was held at the Hubert H. Humphrey Center in Minneapolis, Minnesota on June 1012, 1999. The conference was sponsored by the University of Minnesota Supercomputing Institute for Digital Simulation and Advanced Computation, the University of Waterloo, and the Lawrence Livermore National Laboratory. Eighty-seven participants came from all over the world‹39 international participants from Africa, Australia, Asia, Europe, and South America and 48 participants from the United States and Canada. The conference was timed to coincide with the biannual "Householder Numerical Algebra Symposium" held in Whistler, British Columbia the week before.
The conference aimed at bringing together researchers to discuss the latest developments in preconditioning methods and the use of iterative procedures in scientific and engineering applications. The organizing committee, Yousef Saad (University of Minnesota), Daniel Pierce (Boeing Company), and Wei-Pai Tang (University of Waterloo), strived for an equal array of talks on applications and algorithms. This goal was reached with 15 of the 32 talks falling in the applications category.
 |
 |
| Peter Forsyth (left)of the University of Waterloo, Canada and Martin Gutknecht (right) of ETH-Zentrum, Switzerland.
|
Masha Sosonkina (left) of the University of Minnesota Duluth and Iain Duff (right) of the European Centre for Research and Advanced Training in Scientific Computation, France.
|
 |
 |
| Edmond Chow (left) of Lawrence Livermore National Laboratory and A. Yu. Yeremin (right) of the Russian Academy of Sciences.
|
Wei-Pai Tang (left) of the University of Waterloo, Canada and Wing Lok Wan (right) of Stanford University.
|
The opening talk, in the category of applications, was given by Charbel Farhat of the University of Colorado at Boulder. His talk, "10 years of FETI (Finite Element Tearing Interconnections)," was an overview of work done by a team at the University of Colorado at Boulder in solving a large variety of problems in mechanics.
Peter Forsyth of the University of Waterloo, Canada presented an introduction to problems in computational finance and discussed his experience in dealing with these problems. Professor Forsyth presented some experiments in using iterative methods for the solution of Partial Differential Equations resulting from the problems in computational finance.
Presentations on recent improvements or applications to the class of approximate inverse methods were given by Alex Yeremin from the Russian Academy of Sciences, Russia, Edmond Chow from Lawrence Livermore National Laboratory, Michele Benzi from Los Alamos National Laboratory, Jun Zhang from the University of Kentucky, and Thomas Huckle from the Technical University of Munich, Germany. Overall, these talks reflected the tremendous progress made in this area. Some of the talks presented experiments in which these methods are now starting to be competitive with standard ILU preconditioners. A few years ago, such methods were met with much skepticism. A great deal of the progress was made in finding new inexpensive strategies for obtaining a good pattern for the inverse. It was reported, for example, that taking the pattern of A2 was generally sufficient to achieve the efficiency of most sophisticated techniques.
|
From top left to bottom right: Venansius Baryamureeba of the University of Bergen, Norway and Teresa Vespucci of the Universitá degli Studi di Bergamo, Italy; Willy Schilders of Phillips Research Labs, The Netherlands and Menno Verbeek and Henk van der Vorst of Utrecht University, The Netherlands; Charbel Farhat of the University of Colorado, Jane Cullum of Los Alamos National Laboratory, and John Lewis of Boeing Computer Service; Peiru Wu of Parke-Davis Pharmaceutical Research and Yousef Saad of the University of Minnesota; Andy Wathen of Oxford University, United Kingdom and Howard Elman of the University of Maryland; and Serge Goossens of the Katholieke Universiteit Leuven, Belgium and Wim Bomhof of the University of Utrecht, The Netherlands.
|
Presentations on the use of iterative methods in Circuit Simulation were given by Willy Schilders from Philips Research Laboratories, The Netherlands, Henk van der Vorst from the University of Utrecht, The Netherlands, and Wim Bomhof also from the University of Utrecht, The Netherlands. These presentations represented some of the first discussions in which positive results have been reported on the use of iterative techniques in this specific application. The method presented used one sort of independent set reordering very successfully. Two other presentations on applications in which iterative methods are relatively new were given‹one by Venansius Baryamureeba of the University of Bergen, Norway on the solution of large-scale linear programming problems and a second by Masha Sosonkina from the University of Minnesota at Duluth on the use of iterative methods for solving problems in tire design.
A number of speakers presented preconditioning techniques for specific applications or specific classes of problems. Preconditioning methods for various saddle-point problems or constrained problems were presented by Andy Wathen from Oxford University, United Kingdom, Howard Elman from the University of Maryland, Ilaria Perugia from the University of Pavia, Italy, and Matthew Knepley from Purdue University. Daniele Bertaccini from the University of Firenze, Italy and Raymond Honfu Chan of the Chinese University of Hong Kong, Peoples Republic of China presented preconditioners for P-circulant matrices and Toeplitz matrices respectively. Uri Peskin from the Technion-Israel Institute of Technology, Israel presented a parallel method based on Kronecker sums and products of operators for an important problem that arises in electro-optics and nanoelectronics. Herbert De Gersem from the Katholieke Universiteit Leuven, Belgium considered Schur complement type methods for solving systems arising from field-circuit coupled electromagnetic models. Martin Gander from the Vienna University of Technology, Austria presented a factorization technique (in the PDE sense) specifically geared toward parabolic PDEs. Finally, methods for applications that lead to dense matrices were discussed by Iain Duff from the European Centre for Research and Advanced Training in Scientific Computation, France, Jussi Rahola from the Center for Scientific Computing, Finland, and Menno Verbeek from Utrecht University, The Netherlands.
|
From top left to bottom right: Matthias Bollhöfer of the Chemnitz University of Technology, Germany and Thomas Huckle of the Technical University of Munich, Germany; Charbel Farhat of the University of Colorado and Willy Schilders of Philips Research Labs, The Netherlands; Matthew Knepley of Purdue University and David Keyes of NASA Langley Research Center; and Jussi Rahola of the Center for Scientific Computing, Finland and Andreas Pomp of the Swiss Federal Institute of Technology.
|
Robert Bridson of the University of Waterloo, Canada discussed new types of preconditioners in his presentation on multiresolution approximate inverse technique. Additionally, Brian Suchomel from the University of Minnesota and Lutz Grosz from The Australian National University, Australia discussed strategies using independent sets (ARMS and ILUM respectively). The methods proposed by Andreas Pomp from the Swiss Federal Institute of Technology, Switzerland, which separates the unknowns in two subsets treated differently, also fell into this series of presentations. Esmond Ng of Lawrence Berkeley National Laboratory discussed a number of variants of incomplete Cholesky methods and suggested that preconditioning techniques should benefit from widely used ideas in sparse direct methods.
A number of talks were geared specifically toward parallel implementations or parallel algorithms. This category consisted of presentations by David Keyes of Old Dominion University and ICASE at the NASA Langley Research Center, Maya Neytcheva of the University of Nijmegen, The Netherlands, Wing-Lok Wan of Stanford University, and Serge Goossens of the Katholieke Universiteit Leuven, Belgium. These presentations related to multilevel and Domain Decomposition methods.
The conference also consisted of a good set of poster presentations: twelve in the category of algorithms and others and seven in the category of applications.
The talks and posters were of high quality, and in spite of the rather heavy two and half day schedule, participation and discussion were excellent. A special issue of "Numerical Linear Algebra with Applications" is currently being planned.
|
From top left to bottom right: Raymond Honfu Chan of The Chinese University of Hong Kong and Esmond Ng of Lawrence Berkeley National Laboratory; Uri Peskin of the Technion-Israel Institute of Technology and Michele Benzi of Los Alamos National Laboratory; A. Yu. Yeremin of the Russian Academy of Sciences and Robert Numrich of SGI; Herbert De Gersem of the Katholieke Universiteit Leuven, Belgium and Ilaria Perugia of the Univesità di Pavia, Italy; Wing Lok Wan of Stanford University and Robert Bridson of the University of Waterloo, Canada; and Brian Suchomel of the University of Minnesota, Maya Neytcheva of the University of Nijmegen, The Netherlands, and Lutz Grosz of The Australian National University.
|
|
|
This information is available in alternative formats upon request by
individuals with disabilities. Please send email to
alt-format@msi.umn.edu
or call 612-624-0528.
|
|
URL: http://
|
|
|
This page last modified on
|
|
Website related questions or problems should be directed to
webmaster@msi.umn.edu
The Supercomputing Institute does not collect personal information on
visitors to our website. For the University of Minnesota policy, see
www.privacy.umn.edu.
|
|
| |
