Applications Accomplishments (1997 CRPC Annual Report):



Director: Geoffrey Fox



Because essentially all application work is funded outside the base CRPC

NSF grant, it has a different organization than other CRPC research

activities. Applications borrow from and provide a proving ground for

nearly every CRPC technology. Included in this work are the fifteen (# may
change ? 14)

Grand Challenge activities briefly described in Table 1.1. The CRPC's role
is

typical of the computer science component of this major national activity:

to ensure that the best algorithms and software technologies are available

to application scientists and to monitor and evaluate these technologies.

This style of application work will continue to be a key aspect of CRPC
activities although it will change in detail. For instance during 1996-98,
several of the original Grand Challenge projects will phase out. However
CRPC has strong involvement with the two NSF PACI centers and this will lead
to natural collaboration with some of the dozen major application activities
associated with these programs. DoE has started a new round of Grand
Challenges where Argonne, Texas(? Check Mary) and to a lesser extent
Syracuse will participate. Further Argonne, Illinois and Caltech have major
application programs as part of the new ASCI Centers of Excellence programs.
(Kathy please check DoE GC and ASCI) CRPC has a major role in the DoD
modernization program where the PET (Programming Environments and Training)
is strongly driven by application requirements which tend to be broadly
based and not focused on major "Grand Challenge" like projects. Finally
there are several significant applications outside the national programs
such as the Syracuse work on Financial Modeling with Industry and on
earthquake prediction with Los Alamos and an international group of
geoscientists. Note the application work of CRPC is actually increasing for
the national programs are moving into major deployment mode.

We have also successfully expanded our PCE-TECH activities, which have taken

computer science research projects into usable prototypes and educational

outreach programs to a broader class of high-performance computer user than

that represented by the scientists involved in Grand Challenge projects.

Visit http://www.crpc.rice.edu/CRPC/research/applications.html for more

detail on CRPC applications activities. The work on the National High
Performance software Exchange has important links to our application work



Table 1.1: CRPC Involvement in Grand Challenge Projects



1. High-Performance Computational Chemistry: (DOE---Argonne)

Using Fortran M to develop highly parallel modeling systems appropriate for

the chemistry of hydrocarbons, the chemistry of clay materials, and the

rational redesign of biodegradative enzymes.



2. Convective Turbulence and Mixing in Astrophysics: (NASA---Argonne)

Developing a numerical code for massively parallel MIMD architectures that

combines a higher-order Godunov method with a nonlinear multigrid elliptic

solver to advance the equations of compressible hydrodynamics with

temperature-dependent thermal conduction.



3. Advanced Computing Technology Applications to SAR Interferometry and

Imaging Science: (NASA---Caltech)

Metacomputing for Earth Science applications; produce portable scalable code

for processing Synthetic Aperture Radar signals and for analyzing resultant

data.



4. High-Performance Computational Biology: (DOE---Caltech)

Using CC++ to develop a tree-search program to compute three-dimensional

protein structures, with initial implementations on networks of workstations

and on the Intel Paragon.



5. Parallel I/O Methodologies for Grand Challenges: (NSF---Caltech)

Using Pablo 5.0, which includes software for capture and analysis of

input/output data, the basis for the input/output data characterization

effort at Caltech.



6. Grand Challenge Computational Cosmology: (NSF---Indiana)

Using parallel computation to explore the origin of large-scale structure

in the universe and how galaxies form. The Pablo Group is using svPablo,
part of Pablo 5.0, to study the behavior of HPF on the SGI Origin 2000.



7. Numerical Tokamak: (DOE---Los Alamos, Caltech (JPL))

Parallel algorithms for MIMD and SIMD plasma particle-in-cell codes (JPL,

LANL); visualization; C++ (Caltech, LANL).



8. HPC for Land Cover Dynamics: (NSF---Maryland)

Tools, C++ class libraries, and compiler run-time support for distributed-

memory and disk-based hierarchical data structures, with emphasis on

efficient support for hierarchical data structures used in geographic

information systems.



9. Computational Aerosciences: (NASA---Rice)

Solving computationally-intensive engineering design problems by using

cheaper, less accurate simulations with strategic reality checks requiring

the full simulation (with Boeing and IBM).



10. Texas Center for Advanced Molecular Computation: (NIH---Rice, Houston)

Simulation of 130,000-atom biomolecular system, acetylcholinesterase in

solvent, using a scalable version of the widely used code GROMOS; comparison

of explicit parallel code with Fortran D version.



11. Numerical General Relativity: (NSF---Syracuse)

Development of toolkit and interoperable modules with common data
structures;

Fortran 90 integrated with DAGH to create a highly effective visualization
package, SciVis, which physicists have used extensively.



12. Material Sciences: (DOE---Tennessee)

Implementing algorithms and software, which are being used in a number of

computationally intense applications such as material science, for the

parallel solution of core problems.



13. Parallel Simulation for Modeling Cleanup of Groundwater: (DOE---Texas)

Formulating and implementing on distributed platforms accurate, efficient,

parallel algorithms for solving multi-phase flow and transport with

biological and chemical kinetics in a heterogeneous porous media.



14. Computational Chemistry of Nuclear Waste Characterization and
Processing: (DOE---Syracuse, Argonne) Relativistic Quantum Chemistry of
Actinides. Researchers will implement relativistic quantum chemical methods
on massively parallel computers and, for the first time, provide
capabilities for modeling heavy-element compounds similar to those currently
available for light-element compounds.



Neither of Following is a Grand Challenge in federal sense. Both are
technically very

Similar to Grand Challenges but funded differently. So remove from table and

Add to applications page?



15. General Earthquake Modelling, GEM (DOE---Syracuse, Los Alamos)

Geoscientists and physicists are developing an international project to
model earthquakes as the dynamics of interacting fault segments. Part of an
emerging field that uses a variety of the fast multipole algorithms
developed by Caltech and Los Alamos for astrophysics.



As above This is NOT a GC – it is funded by industry

GCF: Is the following a Grand Challenge or is it an application that might
be added to our applications web site? If it is a GC, is it sponsored by the
DOE? Or someone else? Are other CRPC sites involved?



16. Financial Modelling (Syracuse)

Through the NPAC InfoMall technology transfer program researchers are
developing small business novel High Performance software systems (involving
monte carlo and neural nets) to predict behavior of complex financial
instruments.