*** I
AHPCRC (n.) Researchers from the US
*** G
Asynchronous Problems(n.)
These are hard to parallelize problems with no natural algorithmic
synchronization between the evolution of linked irregular data points.
Further information and examples can be found in Parallel Computing
Works, chapters I and II.
*** G
atmospheric research (n.)
Studies of the behaviour of the atmosphere involve both large scale
numerical simulation and the analysis of significant quantities of
data.
*** G
back substitution (n.) See LU decomposition.
*** G
battlefield simulations
(n.) The complexity and expense of modern warfare has created the need
for accurate computer simulations, hence allowing commanders to
explore tactical and strategic issues at minimum costs.
*** I The US Army
Research Laboratory (ARL)'s Simulation Technology
Division conducts research into this area.
**** this area goes to battlefield simulation
*** G
biomolecular simulations
(n.) The field of biotechnology has seen spectacular advances
in the recent years. The understanding of some of these processes at
the molecular level has allowed computationally demanding models to be
made. In particular two areas which have received attention are
automated DNA sequencing, and the modeling of protein folding. The
BLAS (n.) Basic linear algebra software: a
suite of very basic linear algebra routines, out of which almost any
other matrix calculation can be built.
See the National
Software Exchange for the BLAS source and
documentation.
*** G
C2I (n.) Command, Control and
Intelligence. See C3I.
*** G
C3I (n.) Command, Control, Communications
and Intelligence (C3I) is one of the most important concepts in modern
US warfare doctrine. The underlying idea is as old as war itself, but
the prevalence of modern technology has allowed new avenues to be
explored in the task of providing the vital eyes, ears, and voices for
the military commanders.
*** I
The US Airforce's Rome Laboratory has a
dedicated C3I
Technology programme,
**** in the C3I area and has developed nonmilitary spin off
*** note deletions contributions in a wide area
of activities, from helping the United Nations' aid effort in the
Former-Yugloslavia, demonstration of the New York Network high speed
network (with NYNEX, Inc), as well as industrial technology transfer
programmes.
*** G
C4I (n.) Command, Control, Communications,
Computers and Intelligence. See C3I.
*** G
CAD (n.) Computer-Aided Design; a term which
can encompass all facets of the use of computers in manufacturing
although the term CAM is also in use.
CAE (n.)Computer-Aided Engineering, like CAD, but usually applied to the use of computers in
fields such as civil and nautical engineering.
CAM (n.) Computer-Aided Manufacturing.
*** G
cardiac fluid dynamics
(n.) The availability of supercomputers has expanded the models
medical researchers are able to make.
*** I
Of particular note in the
anatomical field
**** of cardiac fluid dynamics
is recent work by researchers at New York University
who have used supercomputer resources at the Pittsburgh Supercomputing
Center to produce a full three dimensional
View of the Heart. After 15 years of development, beginning with
two dimensions and for the last five to six years working on the much
more difficult three-dimensional version, the researchers finally
achieved their goal of producing a fully functioning 3-D computational
model of the heart, its valves and the flow of blood through the
nearby major vessels. For this accomplishment, the researchers
C. Peskin and D. McQueen received the 1994 Computerworld Smithsonian
Award for Breakthrough Computational Science.
*** G
CEM (n.) computational electromagnetics;
the simulation or prediction of electromagnetic fields using
computers. Models used include field and particle models and a number
of techniques including: finite differences; finite elements; method
of moments; and discrete methods. Further information is available as
an
article in the Encyclopaedia
Technologica. See also CFD.
*** G
CFD (n.) Computational fluid dynamics; the simulation or
prediction of fluid flow using computers, a field which has generally
required twice the computing power available at any given time.
*** G
chemical
processing plant simulations (n.) The building of modern
chemical plants on an industrial scale prohibits is often orders of
magnitude higher than the cost of a high-end supercomputer. Hence it
is obviously cost effective to simulate the processes within the
prospective plant as closely as possible, for example the behaviour
inside distillation columns, to both prove designs as well as optimise
existing plants.
*** I
Bayer
AG, is one of the larger users of high performance computing for
processing plant simulations.
*** G
chemistry modeling (n.) an
application category that includes simulations of: chemical
potentials; elemental reactions and chemical dynamics. The problems
are often formulated in terms of matrix elements; matrix eigenvalues;
matrix multiplication and matrix inversion.
*** G
classical dynamics (n.) See
molecular dynamics.
*** G
cleanup of groundwater
(n.) This topic combines aspects of CFD, multiphase fluids in porous
medium etc into groundwater flow simulation for the production of
environmental modeling codes. For further information see
Computational Applications of the CRPC.
*** G
climate modeling (n.) The
development of sophisticated numerical models on high performance
computers is currently the only realistic means Man has for answering
questions regarding Earth's future climate. These models are more
demanding than those used for short term, comparatively local,
forecasting because entire planet's atmospheric and ocean circulation
need to be considered, as well as for substantially longer time
periods. The Climate and Global
Dynamics Division (CGD) of NCAR which studies the physical causes
of present and past climates, with particular emphasis on global
change.
****I
Call Community Climate Model and other general climate applications
*** repeat last sentence of climate entry
The division also examines the large-scale dynamics of the
atmosphere and oceans, developing complex supercomputer models,
including NCAR's Community Climate Model. In addition, researchers at
the Lawrence Livermore National Laboratory have demonstrated an
Atmospheric-Oceanic model on a variety of high-performance
platforms.
*** G
climate studies (n.) The
numerical simulation of the long term development of the earths
climate is an area of growing concern. Numerical models bring together
chemistry modeling, fluid
flow. Models are intergrated for long time periods and numerical
errors are a major concern. Advanced methods are adopted in an attempt
to reduce these. See also weather forecasting.
*** G
combined applications
(n.) Here several currently isolated computations are tightly coupled
together to performan advanced computational simulation of a complex
problem. One example would be an engineering problem such a fire
simulation. Current codes make many simplifing assumptions but an
advanced code could have coupled high resolution CFD, structural dynamics, chemistry modeling, environmental modeling, graphics rendering and particle flux transport modeling
computations within one code.See also complex systems.
*** G
combustion process
simulation (n.) Combustion processes dominate current
transportations, as well as playing a major part in many industrial
processes. Both the requirements of energy efficiency and emission
reduction have led industry to increased use of computer simulation to
design combustion devices.
*** I
*** call Specific Combustion Projects
Researchers from the Los
Alamos National Laboratory have worked with those from the
Lawrence Livermore National Laboratory, the Courant Institute for the
Mathematical Sciences, and the University of California, Berkeley, to
develop a new generation of computational tools for simulating
combustion processes on complex three-dimensional geometries, using
state-of-the-art numerical algorithms. This project makes use of
adaptive meshes, and both data-parallel and domain decomposition
algorithms are being developed for the parallel implementation.
Support is from the Department of Energy's HPCC program.
*** G
complex systems (n.) an
applications category where many relatively simple modeling
components are combined into a complex whole systems model. Examples
might include defense simulations; education simulations; multimedia
and virtual reality in entertainment situations; multiuser virtual
worlds; and chemical and nuclear plant operation. Applications are
often event driven and time stepped. Powerful computational engines
are often needed as well as very large database backends. These
systems are often intractive. The MADIC
project concerns the integration of large complex systems for the
aerospace industry. See also CFD, CEM, structural dynamics.
*** G
computational biology
(n.) The simulation of biological molecules and processes.
This combines many application areas including
computational chemistry,
genetic sequencing and
molecular dynamics.
Parallel technologies can assist in the simulation of the processes involved,
as well as the processing and visualisation of the large amount of data
produced.
For further information see
Computational Biology applications of the CRPC.
*** G
computational
biomolecular design (n.) Here aspects of biomolecular
simulations and computational biology are combined with expertise in
chemistry, chemical engineering, computer science and mathematics to
produce models for life processes. For further information see
Computational Applications of the CRPC.
*** G
computational chemistry
(n.) Numerical simulation of chemical reactions.
Associated areas are molecular modeling, classical and
quantum molecular dynamics involving both classical mechanics for
particle-in-cell methods, and Monte Carlo methods for quantum
systems. Manufacturing processes can involve study of applications such
as CFD and multiphase fluids in porous medium flows.
For further information see
Computational Chemistry applications of the CRPC.
*** G
Computational Fluid
Dynamics (n.) See CFD.
*** G
conjugate gradient
method (n.) A technique for solving systems of linear
algebraic equations, which proceeds by minimizing a quadratic residual
error function. The method is iterative but quite powerful: in the
absence of roundoff error, it will converge exactly in M steps, where
M is the order of the system in question.
*** G
data assimilation (n.)
Here the collection and processing of realtime data from sources such as
satellite or enviromental studies are coupled with models of environmental
change such as weather forecasting to improve accuracy. The process
involves using temporal data along with spatial data as part of the input
boundary conditions. For further information see
Computational Applications of the CRPC.
*** G
direct method (n.) Any technique
for solving a system of equations that relies on linear algebra
directly. LU decomposition with back substitution is an example of a
direct method. See also indirect
method.
*** G
Embarrassingly
Parallel Applications(n.) Such applications can employ
complex algorithms but can be parallelized because the evolution of
different points is largely independent. More information and examples
can be found in Parallel Computing Works, I
*** G
energy management (n.)
Here the goal is to provide support for the design, implementation and
management of energy efficent processes. It includes a wide range of
computing technologies from the design cycle and CAD/CAE to the management
of large databases and realtime active control systems which many include
neural networks.
*** G
environmental modeling
(n.) an application category that can include: weather forecasting; climate simulation; oil reservoir simulation;
waste repository simulation. Most work in this field is done using
partial differential equation solvers either using finite difference methods or finite element methods. An
important complication of models used in this field is their
sensitivity to input data in the form of initial conditions and
feedback.
*** G
fast Fourier transform
(n.) See FFT
*** G
fast multipole (n.)
*** G
FFT (n.) The fast Fourier transform is a
technique for the rapid calculation of discrete Fourier transform of a
function specified discretely at regular intervals. The technique
makes use of a
butterfly data structure.
*** G
financial modeling (n.) an
application category wihich is mostly concerned with decision support
and optimisation. Techniques such as Monte
Carlo methods and simulated
annealing are often used. Financial simulation models are
often linked to very large database systems to access pricing and
historical data.
*** G finite difference
method (n.) A direct method for
the approximate solution of partial differential equations on a
discrete grid, by approximating derivatives of the unknown quantities
on the grid by linear differences. See also finite element method.
*** G
finite element method (n.)
An approximate method for solving partial differential equations by
replacing continuous functions by piecewise approximations defined on
polygons, which are referred to as elements. Usually polynomial
approximations are used. The finite element method reduces the
problem of finding the solution at the vertices of the polygons to
that of solving a set of linear equations. This task may then be
accomplished by a number of methods, including Gaussian elimination, the conjugate gradient method and the multigrid method. See also finite difference method.
*** G
forward reduction (n.) See LU decomposition.
*** G
full matrix (n.) A full matrix is
one in which the number of zero elements is small (in some sense)
compared to the total number of elements. See also sparse matrix.
*** G
*** Individual mentioned sites should be listed in instance glossary with
*** some description
galaxy formation (n.) Study of
the formation and evolution of galaxies, using large-scale N-body simulation. Further
information is available from the Grand
Challenge Cosmology Consortium, the University of Washington
HPCC Earth and Space Science Project, and the Los Alamos Theoretical Astrophysics
Group. See also Grand
Challenges, N-body
simulation.
*** G
Gauss-Seidel method (n.) An
iterative method for solving partial differential equations on a grid.
When updating a grid point the new value depends on the current values
at the neighbouring grid points, some of which are from the previous
iteration and some, which have already been updated, are from the
current iteration. So updates are performed by sweeping through the
grid in some user-chosen fashion. The key feature of this algorithm
is that it makes use of new information (updated grid point values) as
soon as they become available.
*** G
Gaussian elimination (n.) A
method for solving sets of simultaneous linear equations by
eliminating variables from the successive equations. The original
equation in the form Ax=b (A is a matrix, b the vector of known
values, and x the unknown solution vector) is reduced to Ux=c, where U
is an upper triangular matrix. The solution vector x can then be
found by back substitution. This
method is usually formulated as LU
decomposition.
*** G
geoscience (n.) The study and
simulation of the Earth's geology. This encompasses groundwater flow
simulation, oil
reservoir simulation, seismic wave simulation, and pollution modeling. Codes often
involve several application areas including CFD, computational chemistry and data assimilation. For further information
see the
Geophysical Parallel Computation Project.
*** G
Grand Challenges (n.) Grand
Challenges are defined by the National Science Foundation as
"fundamental problems in science and engineering with broad economic
and scientific impact, whose solutions require the application of
high-performance computing". For further information, see
a list of Grand Challenges.
*** G
graphics rendering (n.)
An applications category covering computer graphics, computer animation
and virtual reality.
Common rendering algorithms include ray tracing, radiosity, and
polygon-based rendering.
*** G
groundwater flow simulation
(n.) The problem of nuclear waste from highly radioactive
spent reactor fuel rods to rubber gloves from medical uses is becoming
a source of concern in many countries, as available above ground
storage sites are filled. One promising solution is to place the
material in vast underground repository, to be sealed over geological
time scales. For those examining potential sites, the major concern
is whether the radionuclides will escape in unacceptable quantities,
via transport from groundwater. Computer simulations have been
developed to answer these questions, based on the best available
geological data.
**** I
The Edinburgh Parallel Computing Centre has
collaborated with the United Kingdom Atomic Energy Authority to
develop a suite of their
********** all H are G
Hawick (n.) a Scots word meaning a "town
surrounded by a hedge"; also an actual town on the border between
Scotland and England; also my surname. This is not relevant to HPCC except that this is a usful way of ensuring my
email address (hawick@npac.syr.edu) does not get lost from this file
so you can always seek out the latest version of this glossary.
health modeling (n.) an
application category that includes biological modeling with examples
such as simulations of the effects of pollutants such as lead in human
blood. This field typically makes use of empirical models, Monte Carlo methods and histograms and
statistical modeling.
High Performance
Fortran (n.) See HPF.
high speed networks (n.)
The Federal High Performance Computing and Communications (HPCC)
programme has always stressed the need for high speed network
technology, as well as high performance computing machines.
Researchers in the US Navy's Naval Research Laboratory have worked on
the High Speed
Optical Networking (HSON) project which sponsors commercial
development of Asynchronous Transfer Mode (ATM) networking technology.
The intent is to develop an early capability, including supercomputing
and multimedia interfaces, that will be compatible with future ATM
service offerings from public networks.
HPCC (n.) an acronymn for High Performance
Computing and Communications, which is the field of information
addressed by this glossary. A USA
National Coordination Office for HPCC also exists, and other
information on HPCC can be found from the Northeast Parallel
Architectures Center, the Center for Research in
Parallel Computing the National Software
Exchange or the Edinburgh Parallel Computing
Centre depending upon your geography.
HPF (n.) High Performance Fortran, a data
parallel programming language definition developed by the HPF Forum
lead by CRPC. HPF expands
Fortran 90 by adding various directives and other parallel constructs.
For further information see the HPF Applications collection
at the Northeast Parallel
Architectures Center .
*********** all I are G
***** except Icase which is an instance
ICASE (n.) The Institute for Computer
Applications in Science and Engineering (ICASE),
is a center of research in Computational Fluid Dynamics (CFD), providing
mechanisms for interactions among NASA scientists and engineers, the
ICASE staff, universities and related industries.
indirect method (n.) Any
technique for solving a system of equations that does not rely on
linear algebra directly. Successive over-relaxation is an example of
an indirect method. See also direct
method.
InfoMall (n.) a virtual organisation
for the development of HPCC software and development. For further
information see the InfoMall
server at the Northeast Parallel
Architectures Center .
inner product method (n.)
method of matrix multiplication in which one element of the resultant
matrix is computed at a time. See also middle product method
ISO (n.) International Standards
Organization, which, among other things, sets standards for
programming languages.
*** G
Jacobi method (n.) A stationary,
iterative method for solving a partial differential equation on a
discrete grid. The update of each grid point depends only on the
values at neighbouring grid points from the previous iteration. See
also Gauss-Seidel method.
*** G
land cover dynamics (n.) The
analysis of large data volumes from remote sensing sources such as
high resolution satellite images and enviromental monitoring programs.
Large databases need to be search and analysised to detect any changes
and then feed these into biogeochemical cycling, hydrological modeling
and ecosystem response modeling codes to study the possible
impacts. For further information see
Computational Applications of the CRPC.
*** :I
LAPACK (n.) A linear algebra software
package, which has been mounted on a wide range of platforms. It
evolved from the older LINPACK package from Netlib. See also ScaLAPACK.
*** I
LINPACK (n.) A linear algebra software
package, which has been mounted on a wide range of platforms. It has
now been superceded by LAPACK. (n.)
also a set of widely quoted
performance benchmarks based on linear algebra and available from
the National Software
Exchange.
*** G
Loosely Synchronous
Applications (n.) This class of applications are iterative or
time-stepped but unlike the synchronous case, employ different
evolution(update) procedures which synchronize macroscopically, , more
information and examples can be found in Parallel Computing Works,
*** G
***** add cross reference to gaussian elimination
LU decomposition (n.) a
technique where a matrix A is represented as the product of a lower
triangular matrix, L, and an upper triangular matrix U. This
decomposition can be made unique either by stipulating that the
diagonal elements of L be unity, or that the diagonal elements of L
and U be correspondingly identical.
*** I
MADIC (n.) Multidisciplinary Analysis and
Design Industrial Consortium - an industrial consortium of several
aerospace industry partners, developing engineering codes for cross
optimizing design of CFD, CEM and
structural engineering simulation and modeling. For further
information see
MADIC at NPAC.
*** I
**** should have a general entry for manufacturing processes
saying
The Manufacturing Process is receiving growing interest computationally
and is included as part of multidisciplinary analysis which links all aspects of a product in a single simulation (metaproblem) from conception design, testing, manufacturing marketing and maintenance.
But below could be Instance
manufacturing processes
(n.) Researchers at the North
Carolina Supercomputing Center have simulated the
plastics/injection molding process, combined with high-performance
computing resources and visualization techniques.
*** G
materials simulation (n.)
Simulation of materials properties,in order to understand and predict the
structural, magnetic, optic, electrical, and thermal properties of materials,
with the ultimate goal of being able to design and synthesize materials
with specific properties. This requires large-scale simulations
using techniques such as classical potentials, tight-binding models,
and ab initio methods.
*** here reference our grand challenge entry for this project
Further information is available from the
First-Principles Simulation of Materials Properties
Grand Challenge Project.
*** G
medical simulations (n.) The
use of radiotherapy in the treatment of tumours is amenable to the
numerical techniques familiar to physicists. In particular, Monte Carlo methods can be used to
model the transport of neutrons, photons, electrons, or protons
through a three dimensional model of a patient derived from computer
tomography. The aim is to maximise the radiation dosage to the
target, but to minimise damage to the surrounding healthy tissues.
Moreover, the use of Monte Carlo
techniques is particularly suited to parallel architectures.
*** I file under peregrine
The
Lawrence Livermore National Laboratory has a particle transport
code `PEREGRINE', for use in calculating the dose deposition in
patients receiving radiation therapy.
*** G
message passing (n.) The most
generally adopted, portable and high performance parallel paradigm so
far accepted. an early explosion of message passing extension and
languages has come together into the MPI standard.
*** G
Metaproblems(n.)
Such problems are hybrid integration of several subproblems of the other
basic application classes, Synchronous Applications, Loosely Synchronous
Applications, Embarrassingly Parallel Applications, Asynchronous Problems
etc. More information and examples can be found
in Parallel Computing Works, I
*** G
middle product method
(n.)a method of matrix multiplication in which entire columns of the
result are computed concurrently. See also inner product method.
*** G
*** please define as multiple instruction single data and xref mimd and simd
MISD (n.) the problem with Flynn's taxonomy
- no MISD machines exists - at least we have never found one!
*** G
molecular dynamics (n.) an
application category that includes the study of atomic and molecular
motion and dynamics at the level of classical dynamics as well as quantum mechanics.
The dynamics may be formulated in terms of particle methods with long
range forces with or without a cutoff applied or in terms of short
range forces with neighbour interaction lists. Fast multipole methods
allow the simulation of very large systems as do particle-in-cell
methods which employs a combination of particle methods and PDE solver methods.
*** G
Monte Carlo (adj.) Making use of
randomness. A simulation in which many independent trials are run
independently to gather statistics is a Monte Carlo simulation. A
search algorithm that uses randomness to try to speed up convergence
is a Monte Carlo algorithm.
See also
random number generator.
Further information is available in an
Introduction to Monte Carlo Methods from the
Computational Science Education Project.
*** G
multigrid method (n.) A method
for solving partial differential equations in which an approximate
solution on a coarse resolution grid is used to obtain an improved
solution on a finer resolution grid. The method reduces long
wavelength components of the error or residual by iterating between a
hierarchy of coarse and fine resolution grids.
*** G
multiphase fluids in
porous media (n.) Understanding the behaviour of water and
oil in porous rocks (ie, multiphase fluids) is essential for improving
the accuracy of numerical reservoir simulations. The physics here is
the relationship between the microscopic detail of the porous medium,
and the resultant macroscopic response of the fluids. It is this need
for microscopic detail which drives the use of high performance
computing resources.
*** I
Researchers from the Los Alamos National
Laboratory are engaged in simulating multiphase
fluid behaviour in a porous medium, using a lattice Boltzmann
approach for their code.
*** G
**** add fast multipole as G entryreferencing greengard's work
*** add pointers to grand challenges using this
N-body simulation (n.)
Simulation of the motion of interacting particles, such as the
gravitational interactions of stars in galaxy formation. Standard algorithms for N
bodies are O(N^2), however heirarchical tree-code or multipole
methods such as the Barnes-Hut algorithm have been developed which
are O(NlogN).
*** G
nanotechnology (n.) Molecular
systems engineering. An interdisiplinary field where devices are
constructed at, and to work with objects on the molecular scale. This
technology should allowthe construction of very compact and high
performance computing devices and other services.
*** G
**** the separate entries mentioned below should each be I
** could initially have instance tagged to us navy c c and o s
*** keep full text in G entry but replicate navy center
naval warfare (n.) Researchers in
the US Navy use high performance computer technology in many of its
activities, making use of the increased computing power to execute
novel algorithms. The US Navy's Naval Command, Control and Ocean
Surveillance Center have a number of such projects within its Research and
Development Division. These include the demonstration of the
feasibility of using embedded scalable high performance digital and
optical processing for submarine detection and classification in
shallow water, the automatic recognition and targeting of hostile
aircraft and ground vehicles, interferometric synthetic aperture radar
(SAR) processing of terrain, and simulation of the suppression of
acoustic signals.
*** G
network simulations (n.) an
applications category covering optimisation problems for power
distribution networks; telecommunications providers; and other
distributors. The problems are frequently formulated in terms of a
large sparse matrix where the zero
structure is defined by the network.
*** G
neural network (n.) artificial
devices using interconnects and processing capabilities suggested by
models of the cortex are termed neural networks. These systems are
widely used for optimization problems including content addressable
memories and pattern recognition.
*** G
numerical general relativity
(n.) studies are providing a significant
insight into the processess occuring within remote galatic bodies and
hope to predict features which, if observed, will provide verification
of theories regarding gravitational waves. For further information see
Computational Applications of the CRPC.
*** G
numerical tokamak (n.)
numerical computations to study realistic plasma and geometry
parameters for large tokamak experiments. 3D CFD models are using with
additional electromagnetic field equations resulting in a complex
system to be modelled. For further information see
Computational Applications of the CRPC.
*** G
***** there should be an I which is in grand challenge section
ocean modeling (n.) The modeling
of the world's ocean is analogous to that for the atmosphere. The
fluid being modelled obviously has different characteristics, and the
availability of measurements for boundary and initial value conditions
are on a far lesser scale. Nonetheless many research groups, in
particular those specifically concerned with understanding the World's
ocean system, are engaged in this activity. Researchers at the Los
Alamos National Laboratory have developed a Global Ocean
Model to reflect the new generation of massively parallel
computers, under the Department of Energy's Computer Hardware,
Advanced Mathematics and Model Physics and the Federal High
Performance Computing and Communications programs. This high
resolution global circulation model simulates the evolution in time of
the ocean currents and distributions of temperature and salinity. Its
domain is the three-dimensional global ocean, including realistic
bottom topography and coastal boundaries of continents and islands.
Ocean modeling effort by United States Navy research laboratories
include: the Center for
Computational Sciences at the Naval Research Laboratory who have a
project to produce massively parallel versions of operational ocean
prediction and weather forecast models; the High-Performance Computing at
Stennis Space Center at the Naval Oceanographic Office who also
provide oceanographic support to the Department of Defense (DoD)
through a wide range of oceanographic modeling, prediction, and data
collection techniques; and the Naval Postgraduate
School Scientific Visualisation Laboratory in collaboration with
scientists at the Los Alamos National Laboratory.
*** G
OEM (n.) Original Equipment Manufacturer; a
company which adds components to someone else's computers and sells the
result as a complete product.
*** G
oil reservoir simulation
(n.) The numerical simulation of oil reservoirs is used to
optimise the recovery of oil from the porous rock, in particular when
water is pumped in at injection points to maintain the pressure of the
outcoming oil. Given a description of the reservoir from geological
data, such as the permeability of the constituent rocks, the numerical
model attempts to predict the productivity of the reservoir as fluid
is pumped in, and oil removed, over a typical period of years.
Although the petroleum industry was comparatively swift to make use of
numerical simulations, the advent of distributed memory computing
resources -- from workstation clusters to massively parallel
distributed memory machines with gigabytes of core memory -- promises
the ability to simulate large and/or complex reservoirs such as those
found in the Arabian Peninsula.
*** G
The Edinburgh Parallel Computing
Centre has had a long running project with Intera Information to port
the latter's widely used
Eclipse-100 black oil reservoir simulation code.
*** G
OLTP (n.) On-line transaction processing;
handling transactions (such as deposits and withdrawals) as they
occur. An application area of great importance to banks and insurance
companies.
*** G
OSF (n.) Open Software Foundation; an
organization established by a number of the major computer
manufacturers to set software standards.
*** G
Parallel Computing
Works! (n.) Book describing work done at the Caltech Concurrent
Computation Program, Pasadena, California. This project ended in 1990
but the work has been updated in key areas until early 1994. The book
also contains links to some current projects.
A WWW version of the book exists which
contains links to additional and expanded sources of information.
*** G
parallel I/O (n.) Input output
operation on parallel computers can be a major problem in obtaining
both performance and portability. Studies are currently investigating
the behavior of specific application programs to define
application-level methodologies for efficient parallel I/O.
For further information see
Parallel I/O applications of the CRPC.
*** G
particle dynamics (n.) See molecular dynamics.
*** G
particle flux
transport modeling (n.) an application category relevant to
the nuclear and weapons industries. Monte
Carlo methods are often used for the simuylation of models for
neutron transport in a reactor or in an explosion for example.
*** G
PDE (n.) partial differential equation.
For further information see
Some High Performance Computing Issues in Partial Differential Equations
from the
Computational Science Education Project.
*** G
**** these are WRONG references as I said in PCW comments
*** petaflops book is 1994........
petaflops (n.) A performance level
currently being associated with typical applications for early next
century.Several workshops has been help to discuss the challenges such
performance levels will present. Issues addressed included
Applications and Alogorithms, Device Technology, Architectures and
Systems and Software Technology.
Further information is available in the proceedings of the 1992 Workshop
and 1995 Workshop
on System Software and Tools for HPCC.
*** I
*** need a more general G entry
pollution modeling (n.)
Atmospheric pollution, especially from internal combustion engines,
are an acknowledged problem in many cities in the world. Mathematical
models to improve the understanding of the interaction of the
pollutant with the atmosphere are currently being developed by the
United States Environmental Protection Agency (EPA), using high
performance resources from the North
Carolina Supercomputing Center. The aim of this work is to
evaluate the effectiveness of proposed emissions control legislation
in reducing air pollution and acidic deposition. Environmental models,
including the Urban Airshed Model for Clean Air Act compliance.
*** G
QCD (n.) Quantum Chromodynamics; a model of
the behaviour of matter on sub-nuclear scales, the simulation of which
is very hungry of computing power.
See also Grand Challenges and
Monte Carlo.
Quantum Chromodynamics (n.)
See QCD.
*** G
random number generator (n.)
A program or algorithm that produces a series of numbers which can be
used as random variables, for applications such as
Monte Carlo simulation.
*** G
RDMS (n.) Relational Database Management
System; software to manage a database in which data are stored by
attribute.
*** G
relaxation method (n.) A type
of indirect method in which the values
making up a trial solution to an equation are repeatedly updated
according to some rule until convergence criteria are met.
See also direct method
*** G
rendering (n.) See graphics rendering.
*** G
SAXPY (n.) elemental BLAS operation involving "constant (a) times
vector (x) plus vector (y)". The S refers to Fortran Single
precision. SAXPY and related BLAS operations are often implemented by
the hardware manufacturer as part of the system software and the
execution time for such operations has been used a a primitive
benchmark of a high performance computing system.
*** I
ScaLAPACK (n.) A linear algebra
software package, which has been mounted on a wide range of platforms.
This is a version of LAPACK suitable for
distributed memory computer systems. The software
is available from the National HPCC Software
Exchange. See also LAPACK, LINPACK.
*** G
scheduling (n.) an application
category requiring expert systems; neural
networks; simulated
annealing; linear programming. Typical examples include:
manufacturing (and just-in-time situations); transportation ranging
from dairy delivery to military deployment; timetablinga nd university
classes; airline crew and aircraft scheduling.
*** G
seismic wave simulation
(n.) The use of seismic waves has allowed the oil exploration industry
to make detailed predictions of subterranean geology, and is a key
tool in the search for potential oil reserves. However, inverting the
raw data from the reflected acoustic waves into the three dimensional
spatial domain is a known computationally demanding problem.
*** I
Researchers from the Lawrence Livermore National Laboratory has ported
a
3-D Seismic wave propagation code from the French Institut
Francais Du Petrole onto a massively parallel computer (with
sponsorship from the United States Department of Energy's Gas and Oil
National Information Infrastructure project). This code simulates the
propagation in three dimensions of acoustic waves through a region of
orders kilometres along each dimension. From this reflections from
characteristic features such as salt domes can be simulated with
unprecedented time resolution and spatial size.
*** G
seismology (n.) an applications
category that includes oil and gas exploration and recovery as well as
geological prediction studies. Typical applications tend to require
very large quantities of data and often result in relatively simple
calculations on the measured data.
*** I
semiconductor design (n.)
Researchers at the Los Alamos National
Laboratory have worked on theoretical design tools for advanced
semiconductor devices. The application is the modeling of future
semiconductor devices which will be an order of magnitude smaller and
faster than current devices. The code makes use of state of the art
Monte Carlo (MC) simulations for
building better theoretical models, for one and two-dimensional cases.
The use of advanced computing platforms will allow full
three-dimensional MC simulations to be made.
*** G
simulated annealing (n.)
Optimization technique introduced by Kirkpatrick in 1983 which applies
statistical physics methods to find an approximate optimal solution to
a problem. Typically a thermodynamic analogy is used for the model
system under study and the task of finding an optimal solution is
mapped to that of finding the ground state of the thermodynamic
system.
*** G
smart memory (n.) refers to
integrated processors and memories; content addressable memories; and
various application specific memories . Examples of application which
would benefit from such technologies are financial and economic modeling where large
scale databases are accessed. Graphics rendering or direct solution techniques
such as Gaussian
elimination where high memory bandwidth is required.
*** G..
sparse matrix (n.) A matrix with
the majority of its elements equal to zero. See also full matrix.
*** G
SQL (n.) Standard Query Language; a standard
for adding data to, or recovering data from, relational databases.
*** I separate for dyna3d and southampton work
**** need a general entry as well -- please invent
structural analysis (n.)
Modern vehicles and vessels have to meet stringent safety
requirements, through testing of their structural integrity. The
standard numerical technique for meeting these is through the use of
structural analysis codes. Examples here are the PAFEC-FE finite
element package, and the DYNA3D nonlinear response code. Their uses
range from modeling vehicle collisions, earthquake and highway safety
models, and damage assessment to shipping containers. Researchers at
the Southampton Parallel
Applications Center have ported the PAFEC-FE finite element crash
simulation code onto a parallel platform, by parallelising the core
frontal solver. The DYNA3D program, developed by researchers at the
Livermore National Laboratory is an engineering design tool for
modeling the nonlinear, transient response of complex structures. A
developmental version of DYNA3D has been installed on Meiko CS-2 and
the Cray Research Incorporated T3D parallel supercomputers.
*** G
*** I invent instance entry for nastran
structural dynamics (n.)
the use of computational models for simulation in structural and civil
engineering. The most well know commercial code in this field is
NASTRAN. Finite element
methods are the most widely used method in this field.
*** G
Synchronous Applications
(n.) is a simple class of applications which tend to be
regular and characterised by algorithms employing simultaneous
identical updates to a set of points. More information and examples
can be found in Parallel Computing Works,
*** G
telemedicine (n.) The availability
of reliable, high bandwidth nationwide networks promise to remove the
constraints of geographical locations in many professions. One
example which has been investigated is in the area of medicine,
whereby the use of such networks allow medical information and
expertise to be shared across the country.
***** I
Los Alamos National
Laboratory and the National Jewish Center for Immunology and
Respiratory Medicine in Denver, CO have collaborated on such a scheme.
Their project, `TeleMed',
allows the Center's expertise in pulmonary diseases and radiology, the
treatment of tuberculosis and other lung diseases to be made available
throughout the nation. A national radiographic repository located at
Los Alamos National Laboratory, such that participating doctors can
view radiographic data via a sophisticated multimedia interface
without leaving their offices. With the new system, a doctor can
match a patient's radiographic information with the data in the
repository, review treatment history and success, and then determine
the best treatment. Moreover, the availability of high performance
computers in LANL allows sophisticated image processing to be done on
the Computer Tomography data.
*** G
*** add Instances for Dow and dupont
thermochemical
calculations (n.) The ab initio calculations of the
thermodynamic properties of molecules require supercomputer
performances and large amounts of machine memory and disk storage.
The Du
Pont chemical company uses molecular modeling packages such as
`Gaussian92' and `DGauss' in the search for alternatives to
chlorofluorocarbon (CFC) products; in particular in determining their
thermodynamic properties. In a similar vein, the Dow
Chemical Company uses similar techniques to routinely calculate
thermodynamic properties of complex organic molecules.
*** G
traffic simulation (n.) The
popularity of private transport and the limited public tolerance for
major road projects is a dilemma faced by many industrialised cities,
especially in Europe. High performance computers have recently been
introduced to aid the urban planning process, whereby realistic models
of traffic behaviour allows the planner to decide on the best means of
controlling traffic flow.
*** G
The Edinburgh Parallel Computing Centre has
collaborated with SIAS Ltd to develop a parallel version of the
latter's
Paramics microscopic traffic simulation program. The use of a
Thinking Machine CM-200 allows individual vehicles to be simulated
around the roads of Edinburgh, and future effort is directed at
porting the code onto the distributed memory environment of a Cray
T3D.
*** G
unindexed text search (n.)
The majority of databases use indexing to improve performance. If datasets
are highly dynamic or transient is not feasible.
*** G
unstructured grids (n.)
In unstructured grids each node can be in a different topological location
with respect to neighbouring nodes. Such meshes typically use linked lists
and involved indirect addressing. Both features which can inhibit
high performance on vector, cache or distributed memory systems without
some programming effort.
*** G
Video on Demand (n.) See VOD
*** G
visualization (n.) In large
applications where the data set is very large, the problem of
presenting the data can also be a numerically intensive task.
Moreover, some of the well known visualization techniques, such as
ray-tracing and volume rendering, are
naturally amenable to modern parallel architectures.
*** I
Researchers at
the Scientific Visualization Center in the US Army Corps of Engineers
Waterways Experiment Station are developing Parallel
Visualization algorithms for scientific and engineering
applications.
*** G
Also, the US Army Research Laboratory (ARL) has a Scientific
Visualization team engaged in exploiting high performance
computing technology to meet visualization demands from within the
ARL. Some of the applications include computational fluid dynamics, penetration
mechanics, battlefield troop movement and artificial terrain
generation, and explosive effects simulations.
*** G
VOD (n.) Video on Demand, a new
service which will allow the high costs of the information
superhighway infrastructure to be recovered by charging on a pay per
view basis for services which would otherwise be freely
distributed.
*** G
waste minimization (n.)
Current environmental concerns are leading to the legal requirement for
companies and in many cases householded to reduce waste, both in terms of
material management through recycling and energy through programs of energy
management. Here CAD/CAE can be used to produce minimum waste products and
processes. However care must be taken to analyse full the impact of any
changes and this will require large databases and complex financial and
economic modeling
*** G
**** should have european (guy) and us specific instance entries
weather forecasting (n.)
Short and medium term weather forecasting is perhaps the most publicly
accessible application of supercomputers. Essentially the problem is
that of solving the Navier-Stokes equation for fluid flow, subject to
boundary and initial value conditions from physical measurements.
Most of the industrialised countries in the world have their own
centre for developing and running the models most pertinent to them.
In the United States, this responsibility lies with National Weather
Service, a division of the National Oceanic and
Atmospheric Administration (NOAA).
H
I
Glossary Index and
Credits
Ken Hawick, Geoffrey Fox, and the RoadMap team.
Geoffrey Fox gcf@npac.syr.edu, http://www.npac.syr.edu
Phone 3154432163 (Npac central 3154431723) Fax 3154434741