Oliver McBryan - US grants

Nonlinear phenomena are important across all of science. They constitute a broad and poorly understood area which draws on a diverse range of mathematical disciplines. This research program addresses to an area of nonlinear analysis which has recently shown considerable progress: nonlinear waves in hyperbolic systems and in coupled hyperbolic-elliptic systems. The investigators intend to develop computational methods which are based on this mathematical theory and to participate in the development of the mathematical theory of these waves. The goal of this work is to develop the mathematical theory and to utilize it to the maximum extent possible within the computational process. The success of front tracking shows that this goal is attainable. One of the investigator's major accomplishments has been the proof of scientific principle for the front tracking method. This method is now accepted, on the basis of evidence developed within the research program. The success of front tracking has been confirmed by independent work of others. It parallels in two dimensions the previous success which has been achieved by the Random Choice Method in one dimension. The development and use of mathematical theory in the computational process was likewise a characteristic feature of the Random Choice Method.

This award will support collaborative research between US and French scientists on the topic of massively parallel processing of computational fluid dynamics and large-scale coupled computational fluid and solid dynamics problems. The US investigators are Dr. Charbel Farhat and Dr. Oliver McBryan, University of Colorado at Boulder, and Dr. Elaine Oren and Dr. Jay Boris, Naval Research Laboratory, Washington, DC. The French participants are Dr. Alain Dervieux and Dr. Bernard Larrouturou, French National Institute for Computer Science and Automation (INRIA), Sophia-Antipolis, France. The proposed research includes five components: steady and unsteady flow computations on unstructured meshes, direct and large-eddy simulations on structured grids, multigrid methods applied to flow problems, and transient fully coupled fluid/structure simulations. The application problems, real-gas flows at high Mach number, transonic combustion, transonic combustion, detonation physics, turbulent flows and aeroeleastic analysis of full configura- tion aircraft structures, are chosen because of their technological importance and in order to demonstrate to the engineering community the potential of the emerging massively parallel technology in solving problems that are presently beyond the reach of sequential supercomputers. The research team at Colorado will bring their extensive "hands on" experience with massively parallel finite element/difference computations using novel adn intrinsi- cally parallel numerical algorithms. This team will also be responsible for the coordination of the project and the integration of the fluid and structure engineering software. The research group at INRIA will contribute expertise in steady flow finite element computations with unstructured meshes. The NRL team will bring their advances in unsteady flow parallel computations using unstructured grids, as well as their expertise in massively parallel computing.

High performance Computing and Communications (HPCC) is evolving as a critical enabling technology for the sciences. We propose a program to integrate HPCC into selected research areas in the physical sciences. The program will be based around a core group of faculty who collectively span the required knowledge base and who are currently involved in joint interdisciplinary research programs. The primary mechanism will be the joint supervision of a student's research by two of the PI's. The PI's for this proposal will be faculty members in three departments: Aerospace Engineering Sciences (AES), Astrophysical, Planetary and Atmospheric Sciences (APAS) and Computer Science (CS), joined by a Co-I in Physics, J. Dreitlein. These four departments are nationally recognized by their active and productive graduate programs. Inter-departmental collaboration has increased with the award of a prestigious, multi-disciplinary HPCC Grand Challenge Applications Grant by NSF in 1992. HPCC awards from ARPA and NASA have further increased opportunities for interaction and outreach. These awards have provided state-of-the-art HPCC facilities available to GRT trainees. The graduate trainees will spend substantial residence periods at national supercomputer centers, government and industry research laboratories and international HPC centers.

McBryan 9406582 The goal of the proposal research is to develop effective parallel sparse linear systems solvers for solving the algebraic equations arising in the finite element discretization of elliptic coupled systems of partial differential equations. These solvers will be based on domain and subspace decomposition methods for use in the solution of linear coupled multicomponent problems, such as the fluid-structure interaction problems. The emphasis will be scalable, transportable algorithms.

9502956 Schnabel This award provides support for high-speed networking equipment, high-performance multiprocessor workstation servers. a high-speed disk array, and multimedia devices for collaboration, exposition and visualization. The University of Colorado department of Computer Science will be addressing issues of data movement in Grand Challenge and National Information Infrastructure (NII) applications. This in conjunction with the floating point needs of Grand Challenge computations, leads to their request for high-speed computation and communication infrastructure. The research associated with the infrastructure falls into four inter-related projects in the area of scientific computing, distributed systems, and resource discovery. These four projects are data movement in Grand Challenge computing, global optimization algorithms for molecular configuration problems, compiler-assisted network runtime systems, and resource discovery and information sharing.