% Geoffrey,
%
% here is the new exec summary with 4-5 lines remove that
% described CSE.
%
% You have 5-6 lines to add more details on PSE or whatever if you
% wish.  Please return ASAP.  Thanks.

The Florida State University has a tradition of excellence in advanced
scientific computing and applied sciences, with particular strengths
in applied numerical mathematics, physics, meteorology, oceanography.
In order to meet the challenges of the twenty-first century, FSU has
established an interdisciplinary School of Computational Science and
Information Technology (CSIT) to support both graduate and
undergraduate concentrations, provide a leading-edge high-performance
computational facility, and contribute to a high level of computational
culture beneficial to the State of Florida and to the nation.
The School's content embraces all of the possible inferences drawn
from its name: its scope will include the science and technology of
performing, analyzing, visualizing, communicating and archiving
large-scale computations over wide-area networks, and its goal will be
to employ large-scale computer and network resources in pursuit of
scientific and technological research goals. The initial thrusts of
the School are in the broad areas of {\bf computing sciences},
computational biology, computational climate dynamics, computational
materials science and computational hydrology. The computing sciences
program  goals, in addition to research in computer and information
science per se, are i) to develop and implement efficient algorithms
(of common interests across disciplines) on the state-of-the-art
architectures, ii) to develop integrated software environment that
simplifies the computational effort of application researchers in
doing high-performance computing, and iii) to carry out research in
cross-cutting themes in close collaboration with discipline experts,
and coordinate its research across diverse disciplines. As such,
the computing sciences program critically supports the applications
research and in turn is driven by it.

A few of the core topics in the computing sciences program (whose
importance is almost universal in application areas, whether these
application areas are in physics, biology, meteorology, engineering
or in the social sciences) addressed in the current infrastructure
research proposal include high-performance numerical algorithms for
the solution of partial differential equations, visualization, data
management, and programming and problem-solving environments. These
topic areas overlap with the core CISE-research at the National Science
Foundation. This choice of the core technologies is dictated in a truly
interdisciplinary fashion by the particular needs of the application
areas where FSU has well-recognized strengths.

The area of advanced numerical algorithms addresses the high-order
and high-performance issues for the discretization of partial-differential
operators, such as compact finite-difference methods or discontinuous
spectral methods. It also addresses iteration acceleration strategies
(inherently amenable to parallelization) in an adaptive mesh refinement
context for speeding up the time-stepping procedure of a numerical method.
The wide variety of physical processes in various disciplines, which are
governed by partial differential equations ensures the importance and
rather universal application of this work.

The second core technology area of scientific visualization is an
integral part of high-performance computing research. The proposed
research effort addresses concurrent interactive visualization of
scalar and vector fields, especially relevant to climate dynamics
and liquid composite molding process. Surrogate models of the
main algorithms and parallelism are exploited to boost interactive
visualization.

The third core technology is the area of problem solving environments --
application-sensitive compilation and prototyping, collaborative
science portals, data-intensive computing in high energy and nuclear
physics.  The number of alternatives a researcher must choose between
in selecting a physical model, an algorithm or a parallel processing
strategy is becoming increasingly larger. While having choices is
desirable, having to implement the choices can be cumbersome. The problem
solving or prototyping environment will relieve the researcher of the
mundane tasks involved in the implementation of these alternatives
as they are selected through a high-level language and the environment
takes care of producing the low-level code, running the simulation and
preparing the results for human interpretation. Synergistically with
this effort is a major effort in Educational Technology linking CSIT
research with FSU's initiative in practical use of distributed and
distance learning. A hallmark of the school will the use of the very best
technology to create productive learning environments.

The programming environments efforts specifically focus on parallel
and distributed programs based on Java with data parallelism that
can be implemented with either threads or message passing interface (MPI).

In true synergistic fashion, research in these CISE core technology
areas is proceeding in close collaboration with research in the
application areas which employ the core technologies. A unique feature
of this proposal is the manner in which the School of Computational
Science and Information Technology provides a structure already in
existence for the facilitation of cross-disciplinary research between
CISE core technology research areas and the application research areas
that they benefit.  Through the management and operation of the requested
infrastructure by the School, technology transfer between different
applications, as well as  between the core technologies and the
applications (in both directions) will take place with an ease that
would be difficult if not impossible in a more traditional academic setting.
The School is constituted in such a fashion as to encourage and promote
multidisciplinary interaction necessary for the success of the proposed
research effort, and the brief overview of the organization and management
of the School provided at the end perhaps makes it clear.

We have chosen four application areas -- i) high-performance computational
models of coupled atmosphere-land-ocean systems, ii) rapid virtual
prototyping and optimization for liquid composite molding process,
iii) turbulent flows and radiated noise, and iv) general earthquake models
computational infrastructure. FSU's expertise in the first three areas
is highly regarded. Geoffrey Fox brings with him the collaborative
arrangement in the fourth area involving experts across the nation.
These projects are only a few of the many research projects conducted
in the vigorous research environment at FSU.

In the first project, a team of computer scientists, mathematicians and
preeminent geophysical modelers proposes to
i) effect modeling and algorithmic improvements to the state of
the art in simulating global circulation with passive and active
transport, and
ii) adapt full application codes (from input to output, not just kernels)
to global distributed memory, local hierarchical memory environments;
and
iii) develop and demonstrate global circulation software and
broadly applicable partitioning and load balancing tools, and
a problem solving environment hospitable to geophysics research
in general, and weather and climate prediction in particular.
Successful completion of this project will push the weather and
climate prediction capability to the next level, whose value
can not be overemphasized.

The second project is concerned with composite materials, specifically
liquid composite molding (LCM) process modeling and design optimization.
In current LCM applications, computer simulations tools have been used
to predict LCM process behavior. Design cycle simulations based on
such tools become prohibitively expensive and extremely time-consuming
to be useful. Thus, there is a need to develop simulations models and
solution algorithms which are both efficient and accurate to conduct
LCM process design in a rapid real-time fashion. This is precisely
the intent of the project. A multidisciplinary team of computer
scientists, mathematical modelers, materials scientists and industrial
engineers propose to exploit the results of the core technology, and
build a robust and problem solving environment for rapid virtual
prototyping and optimization of LCM process.

The third project area is concerned with noise production by
turbulent flows, particularly those associated with jet engines.  The
goal of this work is to minimize jet noise without significantly
reducing engine thrust.  The solution of this problem involves both the
analysis of the turbulent flow field -- an extraordinarily difficult
problem in itself due to its enormous range of scales -- and the analysis
of the production and propagation of the sound waves.  Scientists at FSU
with decades of experience in turbulence and acoustic problems are working
together to develop models for subgrid scale phenomena and exploit advanced
algorithms of high-order accuracy and their efficient implementation on
high-performance platforms (resulting from core research).

It is but proper to provide here an overview of the School of computational
Science and Information Technology to demonstrate that FSU has committed
significant new resources to this initiative. The School spans three
colleges -- College of Arts and Sciences, College of Social Sciences
and College of Engineering. The School comprises a dozen departments including
Computer Science and Mathematics. The mission of the School is to
i) perform basic research in core areas of applied computer science, mathematics
(including numerical modeling) and areas of overlap between different disciplines,
ii) to foster an interdisciplinary research environment conducive to the
participation of scientists from industries, national laboratories and academia,
high-performance computing,
iii) develop a curriculum incorporating the latest research techniques
and results,
iv) acquire and maintain state-of-the-art computing, communication
and visuaization facilities, and
v) adopt a pro-active education outreach policy.
%The term ``Computational Science and Engineering", is
%used to represent an interdisciplinary field comprising a specific
%engineering or applied scientific discipline, applied mathematics (including
%numerical methods), and computer science. In its noblest incarnation, it
%strives to abstract the quantitative process of research itself,
%intelligently linking many tools from analysis, optimization, visualization
%and statistics, and freeing the scientist to hypothesize, correlate,
%investigate, simulate and design from a high-level interface.
Core CSIT courses consistent with the mission are being
planned and implemented. FSU has committed significant resources including
thirty new faculty positions in recognition of the fact that interdisciplinary
computational faculty are critical to any university aspiring to intellectual
and educational leadership in the twenty-first century. Presently, the scope of
its interdisciplinary research is broadly categorized into applied and engineering
sciences, biosciences, geosciences and social sciences.
These projects involve researchers in the College of Engineering, a joint
venture with Florida A \& M University which is a historically
minority-oriented university.

The proposal makes clear that FSU has sufficient expertise, leadership,
programmatic resources, and is committed to engage additional faculty on which
to found this effort. In order to execute a rather ambitious research agenda
that overlaps with CISE interests, it now seeks NSF funds for infrastructure
support with required matching commitments. Note that FSU as part of the
launching of the school, has committed to the allocation of 5 to 8 million
dollars for the the purchase of a massively parallel supercomputer that will
achieve over a Teraflop of processing power over the next two years. A request
for proposals will be sent out at the end of January 2000, and the system is
scheduled for installation by the end of October of this year. Here we ask for
the additional infrastructure to enable the computer science research that that
will create the multi-disciplinary environment linking applications and
computing research. We propose infrastructure of three distinct types which
will enhance the raw computing power of the central system and allow research
in distributed systems and cross architecture algorithms spanning mobile
devices through parallel machines with shared and distributed memory
architectures. This research infrastructure purchase involves a further $1M matching from FSU
and is spread over 5 years. There is a commodity cluster architecture aimed at the computer science
distributed system work and algorithmic research for such systems
which are of growing importance. We are proposing a clustered shared memory
system to be used in architecture comparisons and for application/algorithm development
for the large supercomputer. The shared memory will enable thread parallelim research for our
Java based research. The remaining infrastructure with database and mobile technology
will be used extensively -- especially in our education and problem solving environment research.

\newpage
%\end{document}

**********************************************************************
          Gordon Erlebacher
          School of Computational Science & Information Technology
          150 Dirac Science Library
          Florida State University
          Tallahassee, FL 32306-4120

          Tel: (850) 645-0308
          FAX: (850) 645-0300
**********************************************************************