|
Last modified: July 25, 1996 4:15 PM Large-Scale Modeling, Computation, and Visualization in Medicine Christopher R. Johnson Computer Science Department University of Utah Salt Lake City, Utah Computer modeling and simulation of problems in medicine often require a researcher to apply diverse skills in confronting the challenges associated with large data sets, three-dimensional complex geometries, large-scale computing, and numerical analysis. The speaker presented an overview of these distinct, but interrelated aspects of large-scale problems in computational medicine, concentrating on bioelectric field applications in cardiology and neuroscience. David H. Bailey NASA Ames Research Center Moffett Field, California Oddly, the field of pure mathematics has been one of the last scientific disciplines to embrace the computer as an essential research tool. This is changing with the emergence of powerful symbolic math packages and a new generation of computer-literate mathematicians. This presentation described the discovery of new mathematical identities by means of numerical computations on workstations and parallel supercomputers. Peter M. Goorjian NASA Ames Research Center Moffett Field, California Goorjian described algorithm development and computed results in computational nonlinear optics. The computer simulations he discussed are made by solving the full-vector, nonlinear Maxwell equations without any approximations. For the light bullet calculations, the Maxwell equations are coupled to nonlinear ordinary differential equations that model linear dispersion as well as nonlinear effects. For the calculations modeling semiconductors, the Maxwell equations are coupled to the semiconductor Bloch equations, without any approximations. Evan Mitsoulis Department of Chemical Engineering Univeristy of Ottawa Ottawa, Ontario, Canada In recent years progress has been achieved in the viscoelastic simulation of polymeric liquids by using integral constitutive equations. This is particularly true for the modelling of polymer solutions and melts with a spectrum of relaxation times. Novel numerical techniques based on the finite element method (FEM) have been developed to handle the necessary particle tracking in the general case of flows with open and closed streamlines. Test cases for commercial polyethylene melts have shown the adequacy of integral models to capture memory phenomena and strong viscoelastic behaviour, not seeing with simple models of the Upper-Convected Maxwell or Oldroyd-B types. Considerable success in modelling viscoelastic flows has been achieved with a particular type of integral constitutive equation, proposed by Papanastasiou, Scriven, and Macosko, and referred to as the PSM model. This equation has been used to predict several well-known viscoelastic phenomena for polymer melts, such as extrudate swell, difference in behaviour between different polymer melts in flows through contractions, etc. It has also been used in the modeling of several polymer processes, such as fiber spinning, film casting, film blowing, etc. This talk made an effort to explain from the numerical point of view past and current achievements in viscoelastic simulations of polymer fluid flows, mainly for two-dimensional, steady-state cases. Mitsoulis also discussed future developments and challenges in the field of computational non-Newtonian fluid dynamics. Stanley Osher Mathematics Department University of California Los Angeles, California In 1987, this speaker, together with J.A. Sethian devised a new numerical procedure for capturing fronts and applied it to curves and surfaces whose speeds depend on local curvature. The method has been applied to a very general class of problems in computational fluid dynamics. The technique handles topological merging and breaking, works in any number of space dimensions, does not require that the moving surface be written as a function, captures sharp gradients and cusps in the front, and is relatively easy to program. Many applications and extensions have recently been found and were discussed in the talk. Irwin Kuntz Pharmaceutical Chemistry Department University of California San Francisco, California Macromolecular structures contain detailed information that can be used to guide the design of low molecular weight compounds that bind specifically to “active sites” and other surface features. Macromolecule-macromolecule interactions can also be studied. The basic computational tools are rigid and flexible docking programs that scan large databases of commercially available compounds or virtual libraries of compounds that are synthetically accessible. Professor Kuntz described the computational challenges in developing successful docking strategies with special emphasis on the new issues posed by the use of combinatorial libraries. Olaf O. Storaasli Langley Research Center Hampton, Virginia This speaker described the development and performance of a new family of NASA-developed equation solvers used for large-scale (e.g., 551,705 equations) structural analysis. To minimize computer time and memory, the solvers are divided by application and matrix characteristics (sparse/dense, real/complex, symmetric/nonsymmetric, size: in-core/out of core) and exploit the hardware features of current and future computers. In this talk, the equation solvers, which are written in FORTRAN and are therefore easily transportable, were shown to be faster than specialized computer library routines utilizing assembly code. Kisa Matsushima Department of Supercomputer Systems FUJITSU Ltd. Tokyo, Japan This speaker presented two topics involving computational fluid dynamics applications. One concerned the flow field prediction of a future flight vehicle by vector parallel computation; the other was a direct design method in which a flow simulation solver and an inverse design solver are incorporated. The first topic dealt with compressible viscous flows past a space plane, one of Japan’s SST/HST candidates, simulated on a Japanese vector parallel supercomputer. The second topic dealt with the aerodynamic design of wing sections whose geometry realizes a specified surface pressure distribution with a given flow condition. In addition, several examples of aerodynamic design of wing systems were discussed. Steve Reinhardt Cray Research Eagan, Minnesota The Minnesota Supercomputer Center Inc. is scheduled to install a Cray T3E Massively Parallel Processing supercomputer in the Spring of 1997. Steve Reinhardt, the T3E Project Director at Cray Research Inc. was the feature presenter at a recent gathering of the Minnesota Supercomputer Institute and Minnesota Supercomputer Center Inc. He presented an historical account of Cray Research's Massively Parallel Processing project and provided some of rationale behind its design decisions for the T3D and subsequent T3E project. He stated that the T3D design emphasized high bandwidth between processors as well as small latencies for interprocessor communication. These design parameters were also an important part of the T3E design, but there was a much larger emphasis on single processor performance. He overviewed Cray's hardware and software strategy for accomplishing increase single processor performance. He also emphasized the specific application areas where the T3E is expected to perform well. |
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. |
||