[Date Index] [Thread Index] [Author Index]
Call for Participation Intelligent Scientific Computation October 23, 24, & 25, 1992 Royal Sonesta Hotel Cambridge, Massachusetts Intelligent Scientific Computation The purpose of this symposium is to identify the scope of contributions that artificial intelligence has and might make to scientific computing and vice versa. By scientific computing we mean the practical application of computer programs to solve difficult real-world problems such as the super-computer-sized grand challenges (e.g., forecasting severe weather events, predicting superconductors, or energy conservation and turbulent combustion, as proposed in the Office of Science and Technology report "Grand Challenges: High Performance Computing and Communications"). These problems often involve the formulation and use of mathematical models in the form of differential equations. We include also the significant improvement of solution techniques or solution time for engineering problems routinely tackled today, such as seismic modeling or ship design. We use the term "scientific computation" broadly to include engineering computation as well. Although purely theoretical work on small, highly distilled "toy" problems is interesting and useful in its own right, we exclude such work from the scope of this symposium unless it also is applied to solving hard engineering and scientific problems. The result of the symposium should at a minimum be both 1) familiarizing AI researchers with the types of problems faced in scientific computing and the work to date addressing the problems by those outside as well as in AI; and 2) familiarizing those outside AI with the contributions, past and potential, from AI. In addition, we would like to explore ideas about what further leverage could be gained in future research by specifying well defined interfaces between (quasi) independent components of solution tools, by documenting and/or standardizing knowledge bases in enough detail to reuse them or reimplement them quickly, and by working with the targeted user community early enough to understand the problems and gain acceptance for the approaches. Appropriate topics include descriptions of research, completed or in progress, on the application of AI techniques to the various phases of scientific and engineering computing (e.g., problem definition, solution formulation, software construction, experimentation, analysis, revision). Equally important are summaries of work from technologies other than AI that impact on the role of AI in scientific computing. We encourage the submission of summaries of a related technology such as symbolic manipulation systems, compiler optimization, vectorizing/parallelization tools, scientific databases, shared knowledge representations, or numerical libraries. Such a presentation could summarize the current state of the art, the strengths and weaknesses of the approach, and speculate on how that technology could help, replace, or benefit from related AI approaches. Alternatively, a submission might describe contributions of scientific computing to AI, such as a good application problem to work on, or an alternative model of intelligent behavior. Submissions on intelligent scientific computing should address aspects such as: o Specifying specific instances of problems to be solved (e.g., by equations, geometries, construction from primitive components, formal languages, domain-specific languages) o Solution techniques (applying numerical approximations, symbolic manipulation, problem reformulation or specialization, program synthesis, special-purpose simulation generators, adaptive control of the simulation/analysis, knowledge-based methods) o Representing knowledge of physics (directed at scientific computing tools) o Software issues such as performance and reliability, modifying models and algorithms, recording design rationales, documentation, and validation o Visualization or interpretation of results o Experiment guidance and management (parameter set up, model/data management, computational steering, results maintenance/tracking) Submissions can describe implemented systems for directly setting up and/or answering specific questions, simulating classes of problems, or synthesizing programs that solve the problem. Live demos or videotapes may be appropriate for implemented systems. Submissions of AI approaches should note the competing technologies and should address the specific contribution of the AI approach: does it enable problems to be set up and/or solved more efficiently? does it enable the solution of more difficult problems? does it interface to or replace existing technologies such as libraries of numerical routines? In considering all submissions describing the use of symbolic techniques, we will use criteria such as the following. Does the submission demonstrate how its techniques significantly facilitate complex numerical computing? Which phase(s) of the scientific computation process is (are) being addressed? Is the work being done in the context of a realistic science or engineering problem? If a technique claims to replace numeric computing by symbolic reasoning, has the result been demonstrated to be acceptable to a user who would normally use numeric computing? Surveys of the utility of some of these areas might be more useful than details on specific implementations. The symposium will most likely be organized around half a dozen topics, determined by the submissions, with each session consisting of a few brief presentations followed by a discussant and then general discussion. Some sessions will allow for long discussion periods. Potential participants should submit a 2-3 page summary of past and current work in the area organized around the aspects described above. Although a list of related publications should be included, the text description should be thematic rather than simply chronological. In addition, submissions should include a 1-2 page description of a critical issue that the author(s) would like to see addressed by the symposium, including a description of any presentations the author(s) would be willing to make. Additional supporting material in the form of copies of previous papers or extended abstracts is optional. Submit either e-mail (Ascii or Latex) source or 4 paper copies to be RECEIVED BY May 11, 1992 to Elaine Kant Schlumberger Laboratory for Computer Science P.O. Box 200015 Austin, TX 78720-0015 E-mail: kant at slcs.slb.com Phone: 512-331-3737 Fax: 512-331-3760 Acceptances will be mailed by June 8, 1992. Material for inclusion in the working notes of the symposia will be required by August 10, 1992. Symposium Committee: Elaine Kant, Schlumberger Laboratory for Computer Science (kant at slcs.slb.com); Richard Keller, NASA Ames Research Center (keller at ptolemy.arc.nasa.gov); Stanly Steinberg, University of New Mexico, (stanly at crunch.unm.edu). ORGANIZATIONAL INFORMATION This Symposium is part of the AAAI Fall Symposium Series sponsored by the American Association for Artificial Intelligence The topics of the five symposia in the 1992 Fall Symposium Series are: Applications of AI to Real-World Autonomous Mobile Robots; Design from Physical Principles; Intelligent Scientific Computation; Issues in Description Logics: Users Meet Developers; and Probablistic Approaches to Natural Language. Please contact AAAI at the address below for information about other symposia. American Association for Artificial Intelligence 445 Burgess Drive, Menlo Park, CA 94025 (415) 328-3123 fss at aaai.org Most symposia will be limited to approximately 60 participants. Each participant will be expected to attend a single symposium. Working notes will be prepared and distributed to participants in each symposium. A general plenary session will be scheduled in which the highlights of each symposium will be presented and an informal reception will be held on Friday evening, October 23. In addition to invited participants, a limited number of other interested parties will be allowed to register in each symposium. Registration information will be available in July 1992. To obtain registration information write to the address above.