Abstracts of Participants
Participants
Title: Interpretation of Geodetic Crustal Strains using Massively
Parallel Supercomputer Simulations of Nonlinear Dynamical Models
Objectives: The underlying objective of this basic research is to
understand the fundamental physical processes giving rise to hazards
and risks a variety of critical energy facilities face from several
kinds of tectonic instabilities, notably earthquakes, volcanic
eruptions and landslides, in concert with the International Decade of
Natural Disaster Reduction.
Project Description: A variety of nonlinear dynamical processes operate
within the complex earth system, and are observed to display the
signatures of many of the same phenomena as, for example, neural
networks, driven foams, and magnetic depinning transitions in driven
high temperture superconductors.
In particular, scaling (fractal distributions), nonlinear thresholds,
and spatial interactions are all features possessed by these systems.
Signatures of these processes include the appearance of scaling
(geometric and dynamic fractal distributions), global and local
self-organization, space-time pattern formation, intermittancy
(transitions from "laminar" to "turbulent" behavior), chaos, and the
emergence of coherent space-time structures. The geodynamical effects
observed in earthquake systems, particularly crustal straining,
dynamical segmentation, and intermittant seismicity, are being modeled
in massively parallel simulations in an effort to clarify the origins
of these phenomena. Simulations and theoretical investigations are
particularly aimed at quantifying the limits of predictability for
disasters the occur within the earth system. We are currently
developing the simulation methods for earthquake models and the
statistical mechanical analysis techniques needed to understand and
interpret the results. From these simulations, we will then predict
geodetic and other deformations associated with impending earthquakes,
to be tested against Global Positioning System, Interferometric
Synthetic Aperature Radar, and other field data.
Results: During this third year of the project, we have focused on
extending our recent results on 1) coarse grained models of earthquake
and other driven threshold processes, 2) mean field models for
earthquakes involving Lyapunov functionals, 3) the mechanics of
arrested nucleation in dynamical models, 4) viscoelastic and
poroelastic deformation and stress Green's functions in layered media
for use in interpreting GPS and ISAR data. In addition to these areas,
we have also been focusing recent research on an investigation of why
earthquakes stop, that is to say, why rupture processes terminate, and
the mechanics by which earthquakes nucleate against a background of
heterogeneous stress distribution. In our recent work, we have been
able to show, for example, that the most common earthquake events in
models with a realistic stress tensor are localized within a region
determined by a pre-earthquake nucleation event of an area of high
stress. These nucleation events are triggered by smaller precursor
events localized in a region (Mogi donut) surrounding the location of
the main event. However, there are rare events that do not conform to
this pattern. For example we have shown that randomness in fault
properties controls for example, whether a fault exhibits fault-wide
"characteristic" earthquakes, or remains confined to the high stress
nucleation region. The randomness also controls the nature and
space-time patterns of foreshock and aftershock distributions. We have
also been able toderive, from a theoretical basis for the first time,
1) the commonly observed Gutenberg-Richter b value of 1; and 2) the
Omori relation that describes the time dependence of foreshocks and
aftershocks associated with large earthquakes. Moreover, we are also
making significant progress in our efforts to calculate both three
dimensional stress and deformation changes due to earthquakes in
layered poroelastic media, for comparison particularly to space
geodetic ISAR data. The significance of these projects lies in our
increasing ability to understand the origin of historically observed
phenomena such as the Gutenberg-Richter and Omori scaling relations,
and to relate them to models that also generate realistic patterns of
foreshocks and aftershocks, and realistic deformation and stress
distributions.
Title: High-Resolution Imaging of Electrical Conductivity Using
Low-Frequency Electromagnetic Fields
Objectives: The objective is to develop numerical and field techniques
for high-resolution imaging of electrical conductivity using
magnetotelluric (MT) and controlled-source electromagnetic (CSEM)
methods. Applications of high-resolution conductivity imaging include
the mapping of groundwater, resource exploration and reservoir
characterization, subsurface processes monitoring, and general
geological mapping of the crust of the earth.
Project description: Many fundamental questions relating to
resolution, depth of exploration, required bandwidth in frequency and
spatial sampling rate remain to be answered. To resolve some of these
questions four main tasks have been selected in this project; 1)
improvement of the q-domain imaging method using the wavefield
transform and tomographic inversion, 2)development of an approximate
analysis and imaging method using Born inversion, 3) development of a
rapid and practical 3-D inversion scheme, and 4) development of a
borehole time-domain EM system. Some of these tasks are driven by the
pressing need to interpret an increasing amount of field data
available to us. The data may be in the frequency domain, but can
only be interpreted properly if and when tasks 2 and 3 become
successful. Also, the evident success of the q-domain tomographic
imaging process strongly argues for the development of a suitable
wideband borehole system (task 4).
Results: Experiments were carried out using the graphite based scale
model system to establish the validity of the wavefield transform.
The experiment has shown that time-domain electromagnetic data can be
acquired in a laboratory environment with a signal-to-noise ratio of
72 dB and the data can be successfully transformed to wavefields. The
experiment included a surface-to-borehole experiment which gave
surprisingly accurate wave domain results considering that no
theoretical formulation for such a situation exists. This project has
also made it possible the laying down of minimum requirements for data
acquisition systems required for wavefield analysis procedures.
Further efforts have been directed at improving the data processing
procedure to make the technique more robust. An innovative weighting
scheme and the application of wavelet theory have been tested.
Development of an approximate algorithm for efficiently
simulating and interpreting EM responses of 3-D conductivity anomalies
has been continued. The forward simulation is based on the non-linear
approximation ( Habashy et al., 1993) modified to include non-uniform
conductivity distributions. By performing the calculation in spatial
harmonic domain, a reasonable accuracy is obtained while the algorithm
can be two orders more efficient compared to another program based on
the integral equation method. This algorithm will be used to invert
3-D EM data on an ordinary workstation. Inversion routines for both
crosshole and surface-to-borehole configurations is underdevelopment.
A new 3-D parallel nonlinear inversion using global integral and
local differential equation algorithm (GILD) has been developed and
tested successfully (Xie and Li, 1997). The method appears to be a
major step forward from conventional approaches. The GILD algorithm
consists of five parts: 1) The domain is decomposed into subdomains SI
and SII. 2) A new global magnetic integral equation is used in SI and
local magnetic differential equation is used in SII together to obtain
the magnetic field in the modeling step. 3) In the inversion step EM
parameters are updated using the same arrangement as in step 2. 4)
Subdomain SII is uniformly decomposed into smaller cubic domains;
whose sparse matrices can be eliminated separately in parallel. 5) A
new parallel multiple hierarchy substructure algorithm will be used to
solve the smaller full matrices in SI in parallel.
Title 1: Seismic Resonance Characterization of Fractured Geologic
Structures
Objectives: The principal objectives of this research are to develop a
basic understanding of the resonance characteristics of fractured rock
and to develop the associated theory required to determine the
mechanical properties and geometries of fractures in rock from
resonance measurements.
Project Description: Research will begin with laboratory resonance
tests on rocks samples with material anisotropy and attenuation.
Subsequent resonance measurements will focus primarily on the effects
of fracture properties and geometries on the resonance frequencies and
mode shapes. Techniques for selective excitation and detection of
modes will be developed and used to estimate the normal and shear
stiffnesses of dry and fluid-saturated fractures with heterogeneity
and anisotropy in fracture stiffness. Numerical codes will be
developed in parallel to assist in the understanding and
interpretation of the laboratory resonance measurements.
Results: During FY97, the effects of (1) rock anisotropy and (2)
single and multiple fractures on the shifts in resonance frequencies
and resonance peak broadening (i.e., attenuation) of simple
one-dimensional rock bars were investigated. The anisotropy work
focused primarily on transversely isotropic rocks in which the
anisotropy is caused by either bedding structure (sandstone) or
preferentially oriented microcracks (granite). Resonance measurements
were performed on cube-shaped samples under free-vibration conditions.
A numerical inversion technique was developed and used to determine
the five elastic constants of the rocks by matching the observed and
computed resonance frequencies of the samples. Elastic moduli derived
from the resonance measurements were in general smaller than those
determined from ultrasonic transmission tests and larger than those
measured by static stress-strain tests. This result may be attributed
to frequency-dependent wave velocities resulting from scattering off
microcracks and grain contacts.
The frequency spectra computed using the elastic properties from the
inversion showed excellent agreement with the measured spectra. Good
agreement was also observed between the computed mode shapes and those
measured experimentally with a scanned laser vibrometer. The
effects of fractures and fracture viscoelasticity on the resonance of
rocks were investigated using a one-dimensional propagator code and
laboratory experiments on rock bars. The introduction of fractures
altered the resonance frequencies and mode shapes of the rock. The
magnitude of these effects were observed to be largest for modes with
nodes located on or near the fractures. Introduction of water into the
fractures resulted in viscoelastic attenuation. All these effects can
be accurately modeled with a propagator code in with the fractures are
modeled as a displacement-discontinuity boundary condition with a
complex fracture stiffness. These results demonstrate that fractures
in rock have distinct resonance signatures that can potentially be
used for fracture location and stiffness determination using
mode-dependent frequency shifts, and for fracture rheology
determination using peak broadening.
Title 2: Effects of Micro- and Macro-Scale Interfaces on Seismic Wave
Propagation in Rock and Soil
Objectives: This research investigates the effects of multi-scale
heterogeneity in materials such as rocks and sediments on propagating
seismic waves. The heterogeneity at different scales leads to
frequency-dependent behavior of the seismic waves such as the velocity
dispersion and attenuation. The focus of the research will be on the
effects of particular types of the heterogeneity such as fractures and
grain contact in geologic materials.
Project description: This research examines the effects of
multi-scale heterogeneity in sediments and rock on seismic waves. The
program is a joint effort with University of Notre Dame. The research
at LBNL will use laboratory transmission measurements, numerical
finite difference and boundary element modeling, and theoretical model
development to examine frequency-dependent wave phenomena in: (1)
fractured rock with variable mechanical properties and with multiple
fractures, and (2) granular geologic materials for a range of pore
fluid properties and consolidation conditions. Particular emphasis
will be on understanding the basic mechanics of elastic wave
propagation in these complex systems and identifying potential
diagnostic features in the wavefield that can be used to evaluate the
presence and conditions of micro- and macro-scale discontinuities. The
focus of research during FY97 will be numerical simulation of wave
propagation across and along fractures with variable stiffness, and
laboratory investigation of the transmission characteristics of
fluid-saturated, granular media for a range of consolidation
conditions.
Results: Research on wave propagation in fractured rock has utilized
numerical finite difference modeling to examine the effects of multiple
aligned fractures on the anisotropic characteristics of waves generated
by a localized source. Wavefields from these simulations were compared
to those for a transversely isotropic (TI) medium with equivalent
static elastic properties for the fracture model. These simulations
indicate that homogenization of a fracture system into its equivalent
TI properties may not be appropriate, particularly as the fracture
stiffness is reduced, as the number of fractures becomes large, and as
the frequency of the wave is increased. Work is now focused on
developing a dynamic anisotropy theory for fractured rock.
Research on granular media has focused on ultrasonic wave transmission
tests on water-saturated granular media (glass beads and silica sand)
for a range of confining stresses (i.e., consolidation). These tests
have revealed that the spectral content of the transmitted waves
through the samples show marked changes that cannot be explained by
conventional theories for the early stages of consolidation. An
increase in confining stress attenuates the propagating waves for
relatively low stresses. At higher stresses, the low frequency
spectral components of the wave recovers slowly with increasing
stress. This recovery was observed to occur at lower stresses for
packings with small grain diameters. The conventional attenuation
mechanisms based on viscous shear loss between solid and fluid phases
such as Biot and squirt flow models by saturating the packing under
low confining stress with colloidal gel. The transmitted waves showed
virtually no changes in its spectral amplitude before and after the
colloid transformed from its liquid state(finite viscosity) to solid
state (infinite viscosity). Work is presently focused on theoretical
model development and the design of a larger confining vessel that
will enable these tests to be performed at lower frequencies relevant
to crosshole and logging acoustics (10 to 200 kHz).
Title: Physical and Experimental Studies of Magma Rheology,
Sedimentary Basins and Molecular Dynamics of Silicates
Objectives : Our aims are: (1) numerical modelling of dynamics of
sedimentary basins and lithospheric processes, (2) mixing processes in
rheological fluids undergoing convection, (3) numerical modelling of
magmatic underplating and dynamics of plume ascent (4) determination by
molecular dynamics (MD) simulations of the structure and property of
melts and mesoscale fluid dynamics and phase transitions in sedimentary
basins.
Project Description: This collaborative project with F.J. Spera at the
U. of California , Santa Barbara, will help our understanding of the
thermal, chemical, dynamical , mechanical and rheological states of the
continental crust and subcurstal lithosphere with particular emphasis
on the nonlinear interactions among the various subsystems. Our
workplan encompasses the following (1) determination of the structure
and property of melts by MD simulations and thermal convection of
multicomponent systems. (2) Molecular dynamics of mesoscale
fluid-dynamical phenomenon in the range of tens of microns ,
specifically for two-fluid systems (3) mixing processes of rheological
fluids in convection and visualization of complex processes (4) shear
zones in faulting from grain-size rheology (5)dynamical effects of
lithospheric phase transitions on uplifts of sedimentary basins
(6)development of stress fields and faulting (7) numerical modelling of
heat and mass transport driven by thermal and chemical
heterogeneities.
Results: The results reported below are the U. Minnesota part of this
project. Additional results can be found in the summary of activities
by the Univ. California team led by F.J. Spera. MD simulations of
Rayleigh-Taylor instabilities with up to 3x10**6 particles have been
carried out on massively parallel computers ( CRAY-T3E and Convex-HP
systems).The results have been published in Vol. 5 , Annual Rev.
Computational Physics, 1997. We are beginning to put in some rules for
chemical reactions into the MD simulations for studying some reacting
flows with MD. We have finished the first stage of our study of on
mixing in thermal convection with temperature-dependent and for both
Newtonian and non-Newtonian rheology. A theory based on fractal
analysis has been developed , showing the transition from fractal type
of mixing on its route to homogenization. Two papers have come out, one
in Geophys, Res. Lett., 1996 and the other in the February issue 1997
of Earth Planet. Sci. Lett. Recently we have developed a line technique
for studying entrainment and mixing in which very high resolution is
attained by linking 10**5 to 10**6 particles per line. With a dozen
lines, we can map out all of the wrinkles in the line distorted by the
flow and can construct cumulative measures of mass flux from different
source regions by using line-integral technique. The positive feedback
by both non-Newtonian and temperature-dependent rheology on
accelerating fast plumes has been shown in high-resolution simulations.
Twopapers on this topic have appeared in Earth Planet. Sci. Lett. and
Geophys. Res. Lett. this year. We are continuing on work on
two-dimensional wavelets based on the adaptive , collocation formalsm.
One paper on wavelets as applied to solving partial differential
equations will appear in Computers in Physics. We are continuing our
work to study the rise of diapirs in variable viscosity medium using
wavelets. We have used the model of phase transition in sedimentary
basins to explain (1) domal uplift at early stages of rifting, (2)
small amount of crustal stretching for large subsidence , (3.)
stratigraphic onlap. This is being written up now for Tectonophysics.
Advanced state of the art all-electron local density methods (including
total energy and atomic force determinations) are applied to the study
of the structural, electronic and magnetic properties of ceramics and
metal ceramic interfaces. As indicated in this proposal, good progress
has been made in the previous period on a number of problems,
including: (i) The first principles determination of the structure and
bonding at the noble metal-ceramic interface, Ag/Cd(001) and the
structural, electronic and magnetic properties of transition
metal-ceramic interfaces (notably Pd, Rh and Ru on MgO(001). In the
latter we demonstrated that MgO can be used, in principle, as a benign
substrate to realize for the first time the magnetism of Ru and Rh.
(ii) The first principles computational modeling of the leakage
behavior of Metal/BST (barium-strontium-titanate) thin film devices, in
collaboration with Texas Instruments. Here the issues of how to achieve
low leakage current (through calculation of the Schottky barrier
height), bonding mechanism and structure of the interface are
determined. (iii) The magnetism in colossal magnetoresistance (CMR)
materials for which we have (a) determined the effects of lattice
distortions on the competition between double exchange and super
exchange from first principles (b) studied the possible formation of
spin polarons in the double exchange model and its interesting
consequences; and (c) we have initiated a new program to calculate the
transport properties of the CMR materials from a first principles
approach based on linear response theory; results have already been
obtained for the magnetoresistance in some Fe/transition metal
superlattices. (iv) Started investigating the electronic structure and
properties of a ceramic material based on Si and transition metal
carbides and nitrides with first studies of Ti3SiC2 and the possible
role of C substitutional defects in its solid solutions. (v) Finally,
computational methodological developments have been made in
parallelizing the FLAPW and FLMTO codes so as to exploit the growing
power of new parallel architectures.
Our proposed research plans in the next period include: (i) Extending
the search for optimum low leakage metal/BST devices to include some 4d
metals and first replacing BaTiO3 by SrTiO3 and then modelling the
Ba5Sr5TiO3 system using a supercell approach. (ii) Expanding our
investigations of CMR materials by (a) relating the strong coupling of
lattice, magnetic and electronic degrees of freedom to a number of new
features and (b) determining the single electron magneto-transport
properties in the LaMnO3, (La,Ca)MnO3 and (La,Sr)MnO3 systems, and
(iii) Investigating the unusual (antiferro) magnetism and
ferroelectricity of YMnO3.
Title: Spin Dependent X-ray Spectroscopy Theory
We propose to develop relativistic, spin-dependent generalizations of
our ab initio x-ray spectroscopy codes FEFF to improve near edge
calculations and to make possible calculations for magnetic and
heavy-atom systems. Such spectroscopies are important for structural
studies using synchrotron radiation x-ray sources and include polarized
x-ray absorption fine structure (XAFS), and near edge structure
(XANES), x-ray magnetic circular dichroism (XMCD), and others.
One of our long term research goals has been to attain a quantitative
theory of deep core x-ray spectroscopies. These spectroscopies include
x-ray absorption fine structure (XAFS), x-ray absorption near edge
structure (XANES), x-ray magnetic circular dichroism (XMCD),
diffraction anomalous fine structure (DAFS), photoelectron diffraction
(PD), and others. All are used extensively at modern synchrotron
radiation facilities. Since they share in common the same excited
state electronic structure, these spectroscopies have similar
theoretical underpinings: they all can be described in terms of a
curved wave multiple scattering (MS) formalism. These spectroscopies
are important probes of local atomic structure, especially in
non-crystalline materials where conventional diffraction techniques are
inapplicable. However their interpretation generally requires accurate
theoretical models or experimental reference standards. With the
development of our fast XAFS codes FEFF, we have taken a giant step
toward these goals. Our MS codes are generally equivalent or better
than experimental standards, and have been recognized as the best of
those available. Indeed they are used by scientists throughout the
world and have revolutionized the field. By and large EXAFS (extended
XAFS) can now be regarded as a solved problem. The summary
presentation at the 1994 International XAFS Conference in Berlin
mentions FEFF prominently and concludes that EXAFS analysis is no
longer limited to first neighbors. We have been invited to review
these advances in Reviews of Modern Physics. Our project is important
to the DOE mission for several reasons: scientific interest in the
theory and applications; the increasing use of synchrotron radiation
facilities by physicists, chemists and other scientists; and the
significant technology transfer provided by our work. Our project had
an overall rating of "outstanding" (9.0) by the DOE Assessment Office,
which is within the highest priority category for continued funding.
Why is still more work in this area necessary? First, new
spectroscopies continue to be developed requiring extensions of the
theory or improved precision. For example, there is now a substantial
worldwide interest in magnetic scattering and related spectroscopies
such as XMCD. Another relatively new spectroscopy is DAFS, which
combines the information in both absorption and diffraction, and is
well suited for the new generation of high brilliance synchrotrons.
Second, many theoretical challenges to a complete theory of x-ray
absorption remain. Despite considerable success with EXAFS, a
quantitative theory of XANES has been elusive, and theories of
spin-dependent absorption are incomplete. Relativistic corrections are
only included at the semi-relativistic level in FEFF. Third, despite
progress in understanding many-body processes in x-ray spectroscopies,
most codes are still based on a one-electron approximation, in which
many body effects are handled crudely or ignored.
The goal of our renewal project is to overcome such limitations and
develop a relativistic, spin dependent x-ray absorption spectroscopy
(XAS) theory. Such a theory would thus unify the theory of EXAFS and
XANES [we introduced the term XAFS to cover both regimes], as well as
related spectroscopies, like XMCD. Our approach has been to develop in
parallel with formal theory, fast state-of-the-art codes that that test
various approximations, and permit high precision comparisons with
experiment. With the success of the ab initio codes FEFF the problem
of EXAFS (i.e., extended XAFS) is now well understood, though certain
many-body effects remain. For example, the EXAFS spectra calculated
with FEFF for close-packed metals converge to full-MS band-structure
calculations, and also agree with experiment to high accuracy. The
name FEFF refers to the effective curved-wave scattering amplitude
f_eff for each scattering path in the fast Rehr-Albers MS theory.
FEFF was carefully designed to be an accurate user-friendly code for
fast calculations in arbitrary systems. Although adequate for its
original purposes, some of the approximations are significant
limitations for XANES and XMCD calculations and require a major effort
to rectify. We now plan to address these limitations. Each is an
important topic in itself, and all are of significant current interest.
Multimillion atom molecular dynamics simulations of nanostructured materials
on parallel computers
Executive summary
In this summary we provide an overview of our DOE supported research on
"Multimillion Atom Molecular Dynamics Simulations of Nanostructured
Materials on Parallel Computers" for the period 1996-1999. The
research completed under the Department of Energy grant for the
duration 1992-1995 is described. A list of 46 DOE supported
publications and 54 invited talks on DOE supported research is given,
and the first pages of reprints and preprints are included.
An outline of the proposed research, knowledge/technology transfer
activities, and interactions with DOE laboratories and industry are
given below. Nanostructured ceramics are excellent materials for
applications requiring extreme operating conditions. The capability to
withstand high temperatures, combined with their high strength and low
weight, makes them highly desirable for high thrust-to-weight ratio
turbine engines, hypersonic aerospace vehicles, and advanced energy
applications. These materials are also very useful for surface
transportation, electronics, and advanced manufacturing industries.
Nanophase materials are usually synthesized by the gas-condensation
method using an ultra high vacuum (UHV) system. The flexibility of the
gas- condensation method allows the synthesis of a wide variety of
nanophase metals and alloys, ceramics, and intermetallic compounds in
both crystalline and amorphous states. Nanophase materials have
superior properties -- fracture characteristics of these materials are
found to be much better than those of their coarse-grained
counterparts. Nanostructured ceramics exhibit ductility at
temperatures where conventional ceramics show brittleness.
The fundamental issue concerning nanophase ceramics is the relationship
between the microscopic structure, atomic diffusion, and mechanical and
thermal properties at high temperatures and stress. An understanding
of this relationship on the atomic scale is critical for the design of
nanophase materials with application-specific properties. A highly
efficient, parallel molecular-dynamics (MD) approach will be used to
determine the structural, dynamical, thermal, and mechanical properties
of nanophase ceramics (Si3N4, SiC, TiO2, and Al2O3).
Large-scale MD simulations will be performed on parallel machines using
a multiresolution approach which involves the fast multipole method
(FMM) and reduced cell multipole method (RCCM) for the long-range
Coulomb interaction. Non-coulombic interactions will be calculated
with a multiple time-scale (MTS) approach based on an adaptive
long-time MD algorithm. In addition, a separable tensorial scheme will
be used to efficiently compute three-body covalent interactions. With
the multiresolution MD approach, the execution time scales linearly
with the system size and is inversely proportional to the number of
processors. Moreover, the computation time dominates the communication
time. A single MD time step for a 4.2 * 10**6 atom silicon nitride
system takes only 4.8 sec on the 512-node Intel Touchstone Delta
machine at Caltech. On a 512 node IBM SP2 system a single MD time step
should take about 0.5 second! Our parallel MD code has been adapted to
run with several message-passing standards including Intel NX,
Chameleon, Parallel Virtual Machine (PVM), and Message Passing
Interface (MPI).
Molecular-dynamics simulations will be performed on massively parallel
machines to investigate: i) Atomic structure and sintering; ii) atomic
diffusion and relaxation processes in grain boundaries; iii) scaling
behavior associated with interface roughening; iv) phonon spectra and
thermal transport; and v) mechanical properties, including flaw
initiation dynamics and fracture and the microscopic mechanism of
ductility. Increased ductility of nanophase ceramics is an intriguing
issue because grain sizes in the range of a few nanometers could
inhibit the generation and motion of dislocations. MD simulations will
allow the investigation of atomic processes near crack tips, and they
will provide a microscopic picture of how ultrafine grain sizes affect
the emission and extension of dislocations. Simulations will also be
performed to investigate the effect of cluster-size distribution on
mechanical properties of nanophase ceramics under extreme conditions of
pressure and temperature. The proposed work will involve multimillion
atom simulations, since each nanometer size cluster has in excess of
103 atoms and the condensed phase will require on the order of a
thousand nanoclusters.
From a technological standpoint, the high temperature thermal
conductivity of a nanophase ceramic is an important physical property.
We will use a nonequilibrium MD approach in which the equations of
motion are modified to include the coupling of the perturbation to the
energy density. This non-equilibrium MD approach allows the use of
periodic boundary conditions while providing thermal conductivity in
the long wave length limit. The proposed research will involve
immersive and interactive 3D scientific visualization in a CAVE
automatic virtual environment. (From the Louisiana Board of Regents,
we are seeking support to build a CAVE.) The basic control unit of the
CAVE, a Power Onyx, will be connected to our 40 node Digital Alpha
system by ATM (Asynchronous Transfer Mode) and HiPPI (High Performance
Parallel Interface).
Graduate students in this project will receive an interdisciplinary
education in high performance computing. They will be enrolled in a
new dual- degree program which allows students to receive a Ph.D. in
Physics and a M.S. from the Department of Computer Science in five
years.
Finally, the proposed research will involve collaborations with a
number of scientists at DOE laboratories and industry (Boeing
Corporation, Nanophase Technologies and Mobil Research and
Engineering).
Your comments and
suggestions are appreciated.
[Previous]
[Wilkins Home
Page]
Computational
Science Initiative
To cite this page:
Recent/Selected Papers of Participants
<http://www.physics.ohio-state.edu/~wilkins/doe/abstracts.html>
Edited by: wilkins@mps.ohio-state.edu [January 1998]