PHYSICS OF PLASMAS VOLUME 9, NUMBER 5 MAY 2002

Advanced computations in plasma physicsa…

W. M. Tangb)
Princeton University, Plasma Physics Laboratory, Princeton, New Jersey 08543
共Received 26 November 2001; accepted 14 February 2002兲
Scientific simulation in tandem with theory and experiment is an essential tool for understanding
complex plasma behavior. In this paper we review recent progress and future directions for
advanced simulations in magnetically confined plasmas with illustrative examples chosen from
magnetic confinement research areas such as microturbulence, magnetohydrodynamics, magnetic
reconnection, and others. Significant recent progress has been made in both particle and fluid
simulations of fine-scale turbulence and large-scale dynamics, giving increasingly good agreement
between experimental observations and computational modeling. This was made possible by
innovative advances in analytic and computational methods for developing reduced descriptions of
physics phenomena spanning widely disparate temporal and spatial scales together with access to
powerful new computational resources. In particular, the fusion energy science community has
made excellent progress in developing advanced codes for which computer run-time and problem
size scale well with the number of processors on massively parallel machines 共MPP’s兲. A good
example is the effective usage of the full power of multi-teraflop 共multi-trillion floating point
computations per second兲 MPP’s to produce three-dimensional, general geometry, nonlinear particle
simulations which have accelerated progress in understanding the nature of turbulence
self-regulation by zonal flows. It should be emphasized that these calculations, which typically
utilized billions of particles for thousands of time-steps, would not have been possible without
access to powerful present generation MPP computers and the associated diagnostic and
visualization capabilities. In general, results from advanced simulations provide great
encouragement for being able to include increasingly realistic dynamics to enable deeper physics
insights into plasmas in both natural and laboratory environments. The associated scientific
excitement should serve to stimulate improved cross-cutting collaborations with other fields and
also to help attract bright young talent to plasma science. © 2002 American Institute of Physics.
关DOI: 10.1063/1.1467985兴

I. INTRODUCTION cal scaling. This is made possible by the impressive advances

Plasmas comprise over 99% of the visible universe and
are rich in complex, collective phenomena. A major compo-
nent of research in this area is the quest for harnessing fusion
energy, the power source of the sun and other stars, which
occurs when forms of the lightest atom, hydrogen, combine
to make helium in a very hot 共100 million degrees centi-
grade兲 ionized gas or ‘‘plasma.’’ The development of a se-
cure and reliable energy system that is environmentally and
economically sustainable is a truly formidable scientific and
technological challenge facing the world in the twenty-first
century. This demands the basic scientific understanding that
can enable the innovations to make fusion energy practical.
In this sense, as depicted in Fig. 1, fusion plasma science fits
the ‘‘Pasteur’s Quadrant’’ paradigm,1 where there exists a
strong coupling between basic science 共‘‘Bohr’s Quadrant’’兲
and technological impact 共‘‘Edison’s Quadrant’’兲. Accord-
ingly, future research will require the accelerated develop-
ment of computational tools and techniques that vitally aid
the acquisition of the scientific understanding needed to de-
velop predictive models which can prove superior to empiri-
in high performance computing which will allow simulations
of increasingly complex phenomena with greater fidelity.
Viewed in this way, advanced scientific computing, in tan-
dem with theory and experiment, is a powerful new tool for
discovery. This paper reviews recent progress and future di-
rections for advanced simulations in magnetically-confined
plasmas with illustrative examples chosen from magnetic
confinement research areas such as microturbulence, magne-
tohydrodynamics, magnetic reconnection, and others. The in-
tent here is not to provide a comprehensive survey of the
many significant advances in computational plasma physics
but rather to emphasize the message that plasma science is
both effectively utilizing and contributing to the exciting
progress in information technology and scientific computing.
This has played a key role in producing a more stimulating
environment for carrying out the type of research with great-
est promise for accelerating scientific understanding and in-
novation.
II. ADVANCES AND CHALLENGES IN
COMPUTATIONAL PLASMA SCIENCES

Paper TR1 1, Bull. Am. Phys. Soc. 46, 321 共2001兲.

a兲 Although the fundamental laws that determine the be-
b兲
Invited speaker. havior of plasmas, such as Maxwell’s equations and those of

1070-664X/2002/9(5)/1856/17/$19.00 1856 © 2002 American Institute of Physics

Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Phys. Plasmas, Vol. 9, No. 5, May 2002 Advanced computations in plasma physics 1857

FIG. 1. 共Color兲 Fusion energy science is in Pasteur’s Quadrant.

classical statistical mechanics, are well known, obtaining

their solution under realistic conditions is a scientific prob-
lem of extraordinary complexity. In a magnetically-confined
plasma, the interplay between the complex trajectories of
individual charged particles and the collective effects arising
from the long-range nature of electromagnetics forces lead to
a wide range of waves, oscillations, and instabilities charac-
terizing the medium. As a result, there is an enormous range
of temporal and spatial scales involved in plasmas of inter-
est. As illustrated in Fig. 2, the relevant physical processes
can span over ten decades in time and space. The effective
prediction of the properties of energy-producing fusion
plasma systems depends on the successful integration of
many complex phenomena spanning these vast ranges. This
is a formidable challenge that can only be met with advanced
scientific computing in tandem with theory and experiment.
Magnetically-confined plasmas are naturally subject to
both large and small-scale disturbances 共‘‘instabilities’’兲
which thermodynamically relax the system to a lower energy
state. In order to generate more power output than it takes to
keep the plasma hot in fusion experiments, it is necessary to
first understand these complex, collective phenomena, and
then devise the means to control them. The larger-scale
共‘‘macro’’兲 instabilities can produce rapid topological
changes in the confining magnetic field resulting in a cata-
strophic loss of fusion power density. In addition, smaller-
scale 共‘‘micro’’兲 instabilities can also prevent efficient hot
plasma confinement by causing the turbulent transport of en-
ergy and particles. Because of the complexity of the kinetic,
electromagnetic, and atomic physics equations describing the
FIG. 2. 共Color兲 Widely disparate temporal and spatial scales in fusion en-
behavior of fusion plasmas, researchers in this field have a ergy sciences research.
long history of productive use of advanced computation and
modeling. Beginning with the establishment of the predeces-
sor to the Department of Energy 共DOE兲 National Energy confinement and the interactions of plasma with its surround-
Research Scientific Computing Center 共NERSC兲 25 years ings. The recognition of this fact has been underscored more
ago, the U.S. fusion community has been actively and pro- recently by the inclusion of Fusion Energy Science as an
ductively engaged in the simulation and modeling of plasma active member of the new DOE Office of Science ‘‘Scientific

Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
1858 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 3. 共Color兲 Scientific computing is critical to discovery in many scientific disciplines in the DOE Office of Science research portfolio.
Discovery through Advanced Computing 共SciDAC兲’’ Pro-
gram. This is illustrated in the SciDAC Program’s own figure
共Fig. 3兲, which clearly states that advanced computing is
critical to discovery in DOE science programs.2 The lead role
for coordinating the plasma science component of this new
program is the charge given to the Plasma Science Advanced
Computing Institute 共PSACI兲 by the Office of Fusion Energy
Science. The Fusion SciDAC projects include most of the
key topical areas discussed in this review.
As described in the recent National Research Council
report3 assessing plasma physics, the scientific challenges re-
lated to magnetically-confined plasmas can be categorized
into four areas: macroscopic stability, wave–particle interac-
tions, microturbulence and transport, and plasma–material
interactions. Because charged particles, momentum, and heat
move more rapidly along the magnetic field than across it,
magnetic fusion research has focused on magnetic traps in
which the magnetic field lines wrap back on themselves to
cover a set of nested toroidal surfaces called magnetic flux
surfaces 共because each surface encloses a constant magnetic
flux兲. Macroscopic stability is concerned with large-scale
spontaneous deformations of magnetic flux surfaces. These
major displacements or macroinstabilities are driven by the
large electrical currents flowing in the plasma and by the
plasma pressure. Wave–particle interactions deal with how
particles and plasma waves interact. A detailed calculation of
particle motions in background electromagnetic fields is
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
needed to assess the application of waves to heat the plasma
as well as address the dynamics of energetic particles result-
ing from intense auxiliary heating and/or alpha particles
from the fusion reactions. Microturbulence and the associ-
ated transport come from fine-scale turbulence, driven by
inhomogeneities in the plasma density and temperature,
which can cause particles, momentum, and heat to leak
across the flux surfaces from the hot interior to be lost at the
plasma edge. Finally, plasma–material interactions deter-
mine how high-temperature plasmas and material surfaces
can co-exist. Progress in scientific understanding in all of
these areas contributes in an integrated sense to fusion device
design considerations and demands significant advances in
physics-based modeling capabilities. Indeed, advanced scien-
tific codes are a realistic measure of the state of understand-
ing of all natural and engineered systems.
The developmental path for modern high performance
computational codes as validated tools for scientific discov-
ery can be visualized as shown in Fig. 4. This multi-
disciplinary collaborative process begins with basic theoret-
ical research laying the foundations for the mathematical
formulation of the physical phenomena of interest observed
in experiments. Computational scientists produce the codes
which solve these equations using the best possible algo-
rithms which efficiently utilize modern high-performance
computers. They do so in partnership with applied mathema-
ticians who provide the basic mathematical algorithms and
Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 4. 共Color兲 Development of high performance codes as validated tools for scientific discovery.
the computer scientists who provide the requisite computer
systems software. The computational scientists must then en-
gage the research applications scientists in the critical scien-
tific code validation phase where the newly computed results
are compared against experimental/observational data. This
is a major challenge involving a hierarchy of benchmarking
criteria which begin with cross-checks against analytic
theory, empirical trends, and suggestive ‘‘pictorial’’ levels of
agreement. It then graduates to sensitivity studies, where
agreement is sought when key physical parameters are si-
multaneously varied in the simulation and experiment/
observation. At the next level, richer physics validation is
dependent on the availability of advanced experimental diag-
nostics which can produce integrated measurements of key
physical quantities such as spectra, correlation functions,
heating rates, and other variables of interest. If the
simulation/experimental data comparisons are unsatisfactory
at any of these validation levels, the work flow moves back
to 共i兲 the theorists 共in consultation with experimentalists兲 if
the problem looks to be with the mathematical model; and
共ii兲 computational scientists 共in consultation with applied
mathematicians and computer scientists兲 if the problem ap-
pears to be with the computational method. Even when the
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1859
theory/experiment comparisons prove satisfactory, code per-
formance criteria for speed and efficiency could dictate an-
other round in the computational science box. If all criteria
are met, then the new ‘‘tool for scientific discovery’’ can be
effectively utilized for interpreting experimental data, de-
signing new experiments, and even predicting new phenom-
ena of interest. This cycle of development will of course be
repeated as new discoveries are encountered.
Simulation domains have both minimum and maximum
limits on spatial and temporal resolution so that any given
plasma simulation can only address a finite range of space
and time scales. In the past, this limitation has been dealt
with by deriving simplified sets of equations, or ‘‘reduced
equations,’’ that are valid for only limited ranges of time and
space scales. Examples of these are the ‘‘gyrokinetic
equations’’4 or the ‘‘MHD equations.’’5 While the reduced
equations have enabled progress in the past, they have fun-
damental restrictions on their regions of validity. In actual
laboratory or natural plasmas, phenomena occurring on dif-
ferent time and space scales interact and influence one an-
other. It thus becomes essential to utilize more general equa-
tions which are valid on a wider range of space and time
1860 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 5. 共Color兲 Categorization of macroscopic simulation capabilities and corresponding applications topics.
scales and which will accordingly increase the simulation
domains.
At the most fundamental level a plasma can be described
by kinetic equations for the distribution function within a
six-dimensional 共plus time兲 phase-space of each particle spe-
cies. These kinetic equations are coupled to each other
through self-consistent electric and magnetic fields. The
simulation techniques used in plasma physics fall into two
broad categories: kinetic models and fluid models. The most
prominent kinetic approach is the particle-in-cell method,
which proceeds by integrating a 共possibly reduced兲 kinetic
equation in time by advancing marker-particles along a rep-
resentative set of characteristics within the 共possibly re-
duced兲 phase space. Simulation techniques such as ‘‘finite
sized particles’’6 共to reduce the ‘‘noise’’ due to discrete
marker particles兲, ‘‘gyro-kinetics’’7 共a reduction of the full
kinetic equation to a five-dimensional phase space which re-
moves high-frequency motion not important to turbulent
transport兲, and ‘‘delta-f’’8 共a prescription for integrating the
gyrokinetic equation along characteristics which further re-
duces the discrete particle noise兲 have been developed over
the last 20 years. These advances have served to reduce the
requirements on the number of ‘‘particles’’ necessary to
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
faithfully represent the physics and contributed to dramati-
cally increasing the accuracy and realism of the particle-in-
cell simulation technique.
The fluid models proceed by advancing velocity mo-
ments of the kinetic equation in time. The best known of
these are the extended-magnetohydrodynamics 共MHD兲
models9 which represent the plasma as one or more interact-
ing conducting fluids. This higher-level description frees the
model of many fine-scale resolution requirements and makes
feasible the simulation of large-scale motions and instabili-
ties. Extensive theoretical analysis over the years has led to
refinements of the fluid model and improved the closure re-
lations so that many nonfluid effects, such as particle motion
and wave–particle resonances, can be represented at some
level. This is illustrated in Fig. 5 which depicts the different
levels of macroscopic simulation capabilities along with
some representative topics to which they are applicable.10
In the following, representative examples of large-scale
simulations that have been performed in the macroscopic
stability, wave–particle interaction, and microturbulence and
transport areas are presented. Future research directions will
be discussed along with the need and readiness for enhanced
computational resources.
Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 6. 共Color兲 Simulation of high plasma pressure induced disruption in
the TFTR tokamak experiment.
A. Macroscopic stability simulations
Many key macroscopic simulation problems in fusion
science share the common features of extreme temporal and
spatial stiffness, severe spatial anisotropy, and complex
boundary conditions. These characteristics make them
among the most challenging problems in computational
physics. Some examples of advances of macroscopic compu-
tations which have contributed significantly to advancing un-
derstanding of toroidal plasmas include simulations of 共i兲
large-scale disruptions in the actual fusion-energy-producing
Toroidal Fusion Test Reactor 共TFTR兲 tokamak; 共ii兲 nonlinear
evolution of magnetic islands; and 共iii兲 nonlinear evolution
of an internal reconnection event in the National Spherical
Torus Experiment 共NSTX兲 plasma. TFTR is an example of a
conventional tokamak or ‘‘doughnut-shaped’’ torus, while
NSTX is an example of a spherical torus with a hole through
the middle like a ‘‘cored-apple.’’
One of the most dramatic events to occur in fusion plas-
mas is known as the major disruption. This is a catastrophic
event that can lead to the near instantaneous breakup of the
magnetic flux surfaces and loss of confinement. Macroscopic
simulations have addressed the challenge of predicting the
conditions under which the disruption occurs and the conse-
quences of the disruption on the surrounding plasma struc-
ture. Figure 6 depicts results from a nonlinear macroscopic
simulation of a disruption event induced by high plasma
pressure which prevented the achievement of higher fusion
power output in the TFTR tokamak.11 The three-dimensional
structure of select pressure contours shown here is associated
with a complex highly nonlinear process that involved two
distinct plasma instabilities interacting to further destabilize
one another nonlinearly. Note that the constant pressure con-
tours resemble fingers in each poloidal plane, but look like
ribbons in the toroidal direction. These distortions cause
magnetic field lines to become stochastic which in turn
causes a rapid loss of thermal energy and a subsequent rapid
loss of plasma current to occur. Insights gained from these
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1861
studies helped motivate subsequent experiments to further
advance knowledge in this key area.
The nonlinear evolution of ‘‘magnetic islands’’ is a true
multi-scale problem in that these topological changes in the
magnetic field simultaneously involve the slow time-scale
associated with the growth of the islands together with the
rapid Alfvén time-scale of the underlying waves. Specifi-
cally, in units of a normalized time for an Alfvén wave to
traverse the system, the islands evolve over a time period
that is given approximately by the ratio of the magnetic dif-
fusion time to the Alfvén transit time 共the Lundquist number,
S兲. When the longer mean-free-path dynamics resulting from
toroidal geometry effects are taken into account, the resultant
‘‘neoclassical tearing modes’’ are found to limit the maxi-
mum attainable pressure in long-duration tokamak experi-
ments. Figure 7 illustrates the complex magnetic topology
structure that can develop in associated nonlinear MHD
simulations12 where such plasma instabilities cause the mag-
netic surfaces to break into islands when the winding num-
bers on those surfaces are ratios of small integers. This type
of simulation is presently carried out with unrealistic small
values of the Lundquist number (S⬃10 4 ) to accommodate
computational requirements. In addition to the major compu-
tational challenge of dealing with more realistic values 共i.e.,
S⬃10 6 ), a self-consistent closure for the neoclassical fluid
equations is under active analytic development.
In addition to the validation of advanced codes against
analytic theory and against each other, comparisons against
experimental results have become increasingly compelling.
Aided by the successful implementation of unstructured
mesh algorithms and the application of advanced visualiza-
tion resources to deal with complex three-dimensional toroi-
dal structures, recent nonlinear simulations of an internal
magnetic reconnection event in the NSTX experiment pro-
vide an excellent example of such progress. As depicted in
Fig. 8, the sequence of images depicts the evolution of inter-
nal MHD instability showing how the hot inner region of the
plasma 共depicted in red兲 interchanges with the cold outer
region 共green兲 via magnetic reconnection.9,13,14 Comparisons
between soft x-ray measurements of the thermal response
against results from the simulations indicate reasonably en-
couraging agreement with good prospects for improvement
when these calculations are carried out at a higher resolution
on more powerful computing platforms.
The preceding examples of simulations of macroscopic
dynamics involved toroidal 共tokamak兲 applications. Progress
in modeling capabilities are of course also very important for
many other types of magnetic confinement configurations.
For example, reversed field pinch 共RFP兲 plasmas exhibit a
nonlinear dynamo that sustains the toroidal flux against re-
sistive diffusion. The dynamo results from the nonlinear in-
teraction between long wavelength, low-frequency dynami-
cal modes. The associated magnetic fluctuations determine in
large part the confinement properties of the RFP configura-
tion. With the aid of advanced computational models, critical
questions for the future of this concept will be addressed.
These include multi-scale issues such as how the dynamo
activity and associated magnetic fluctuations scale with the
1862 Phys. Plasmas, Vol. 9, No. 5, May 2002 W. M. Tang

FIG. 7. 共Color兲 The complex magnetic topology structure developed as a result of resistive MHD instabilities.

Lundquist number and how this system behaves for times
much longer than the resistive decay time of the first wall.
In general, accommodating the space and time resolution
demands embodied by extended-MHD models and low Lun-
dquist number resistive MHD investigations will require sig-
nificant increases in computing power and improved algo-
rithms. For a model of a given mathematical complexity, the
computational requirements will depend on how effective
methods such as implicit time-stepping and adaptive mesh
refinement 共AMR兲 prove to be. Expected improvements in
single-processor optimization, algorithmic advances, and
available compute power should continue to stimulate im-
pressive progress in this area. Enhanced computational re-
sources should also have a dramatic effect on enabling the

FIG. 8. 共Color兲 Nonlinear simulation of internal magnetic reconnection event in the National Spherical Torus Experiment.

Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 9. 共Color兲 Two-dimensional field solutions for conversion of fast ion
cyclotron waves to ion Bernstein waves in the heating of the DIII-D toka-
mak plasma.
codes utilizing the physical models involving particle clo-
sures to resolve velocity space. As these computationally de-
manding models are exercised with sufficient spatial resolu-
tion and for longer times, improved understanding of the
physical differences between fluid and particle closures will
lead to a new generation of MHD ‘‘tools for discovery.’’
B. Wave–particle interactions simulations
Electromagnetic waves play a fundamental role in the
dynamics of plasma either as externally-driven waves or as
self-generated instabilities. Realistic calculations of particle
motions in background electromagnetic fields are needed to
assess the efficacy of various applications of waves to heat
the plasma and also to address the impact of energetic par-
ticles which result from strong auxiliary heating and/or the
generation of alpha particles from fusion reactions. To re-
spond to this challenge, simulation codes of wave–plasma
interactions are being developed which make increasingly
effective use of terascale computers.15 An illustrative recent
example is the application of MPP power to resolve short
wavelength structures in mode conversion studies. In this
complex wave-heating process, different wave modes, which
co-exist at widely separated wavelengths, can interact due to
strong spatial variations within plasmas. Figure 9 illustrates
the computation16 of important wave features in 2D that had
previously been largely carried out in 1D. Targeting fully
2-D field solutions across the entire cross-section of the
DIII-D tokamak plasma, these results for the conversion of
fast ion cyclotron waves to ion Berstein waves 共IBW兲 clearly
highlight the fact that IBW’s are excited with strong up–
down asymmetry. Further insights into the physics of this
asymmetry will place greater demands on the resolution of
the ion Bernstein wavelength. Since the present computa-
tions already involved the solution of the wave equation—
120 000 coupled complex linear equations for 200⫻200
Fourier modes, future demands for more processors and
memory will be severe. Such requirements may be signifi-
cantly alleviated with more efficient advanced algorithms
such as adaptive mesh refinement approaches. Scientific
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1863
FIG. 10. 共Color兲 Flux tube or annular region of a 3-D toroidal plasma.
progress in this area will be important for the future
development of wave-based techniques for plasma control
needed to help access and then control plasma behavior in
advanced confinement regimes. In addition to fusion
applications, advances in these computational models will
also be of interest to areas such as 共i兲 radio-frequency-
driven plasma sources for plasma processing of materials
including semiconductors;17 共ii兲 advanced space-propulsion
approaches;18 and 共iii兲 the possible relevance of ion-
cyclotron damping of Alfvén waves to the high-temperature
solar wind.19
C. Microscopic turbulent transport simulations
Understanding turbulent plasma transport is not only an
important practical problem but is generally recognized as a
true scientific grand challenge problem which is particularly
well-suited to be addressed by terascale MPP computational
resources. Building on the progress from the ‘‘Numerical
Tokamak Turbulence Project 共NTTP兲’’ within the DOE High
Performance Computing and Communications 共HPCC兲
Grand Challenge Program, improved models with efficient
grids aligned with the magnetic field have now been devel-
oped to address realistic 3-D 共toroidal兲 geometry with both
global and local approaches.20 The ‘‘flux tube’’ codes can
concentrate on the fine-scale dynamical processes localized
to an annular region depicted in Fig. 10. Global codes have
the more imposing multi-scale challenge of capturing the
physics both on the small scale of the fluctuations 共microin-
stabilities兲 and the large scale of the equilibrium profile
variations. In addition to improved implementation of the
well-established gyrokinetic particle-in-cell algorithms, gy-
rokinetic continuum 共Vlasov兲 approaches have also been
productively advanced.21
Computer simulations are being counted upon to ad-
vance the scientific understanding of the turbulence respon-
1864 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 11. 共Color兲 Turbulence reduction via sheared plasma flow compared to case with flow suppressed.
sible for the increased 共‘‘anomalously large’’兲 transport of
particles, momentum, and heat, which are experimentally ob-
served to be significantly greater than levels expected from
the collisional relaxation of toroidally-confined plasmas
共‘‘neoclassical theory’’兲. This is particularly important for fu-
sion because the effective size 共and therefore cost兲 of an
ignition experiment will be determined largely by the bal-
ance between fusion self-heating and turbulent transport
losses. With the advent of teraflop-scale MPP computers,
high resolution simulations of the fundamental equations
governing turbulent transport begins to become feasible. If
reliably implemented, this would be a cost-effective means
to address key phenomena that would otherwise require ex-
pensive empirical exploration of a huge parameter space. Al-
though the progress in capturing the ion dynamics has been
impressive, the description of the electrons is still being up-
graded to include important kinetic effects such as trapping
in equilibrium magnetic wells, drift motions, and wave–
particle resonances. Significant challenges also remain in ex-
tending the present capabilities for dealing with electrostatic
perturbations to include magnetic perturbations—especially
in cases where they are sufficiently large to alter the actual
geometric properties of the self-consistent magnetic field. In
such circumstances, microinstabilities can drive currents par-
allel to the equilibrium magnetic field, which in turn produce
magnetic perturbations in the perpendicular direction. These
kinetic electromagnetic waves can modify the stability prop-
erties of the electrostatic modes or act as separate instabili-
ties, such as kinetic ballooning modes,22 which can alter the
magnetic topology. In this sense, the kinetic simulations
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
would encounter the major multi-scale task of also dealing
with the larger-scale phenomena associated with the afore-
mentioned MHD studies.
A good example of progress in kinetic simulations in-
volves studies of electrostatic turbulence suppression pro-
duced by self-generated zonal flows within the plasma. Re-
sults from particle-in-cell global gyrokinetic simulations
show that the suppression of turbulence caused by prominent
instabilities driven by ion temperature gradients 共ITG兲 is pro-
duced by a shearing action which destroys the finger-like
density contours which promote increased thermal transport
in a 3-D toroidal system.23 This dynamical process is de-
picted by the sequences shown in Fig. 11. The lower panels,
which show the nonlinear evolution of the turbulence in the
absence of flow, can be compared against the upper panel
sequence which illustrates the turbulence decorrelation
caused by the self-generated E⫻B flow. This is also a good
example of the effective use of one of the fastest nonclassi-
fied supercomputers in the world 共the 5 teraflop IBM-SP at
NERSC兲. The most recent simulations24 of this type used one
billion particles with 125 million grid-points over 7000 time-
steps to produce significant new physics results. In particular,
large-scale simulations have been carried out to explore
some of the key consequences of scaling up from present-
day experimental devices 共around 3 meters radius for the
largest existing machines兲 to those of reactor dimensions
共about 6 meters兲. As shown in Fig. 12, transport driven by
electrostatic ITG turbulence in present scale devices can
change character in larger systems. This transition from
Bohm-like scaling to Larmor-orbit-dependent ‘‘gyro-Bohm’’
Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 12. Full torus simulation of turbulent transport scaling.
scaling is a positive trend, because simple empirical extrapo-
lation of the smaller system findings would be more pessi-
mistic. Some experimental observations in a number of rep-
resentative present-day experiments indicate that the relative
level of turbulent heat loss increases with plasma size while
the size of these eddies remains the same.25 However, excep-
tions to this trend, where gyro-Bohm-like scaling is sensitive
to plasma rotation, have also been reported in certain high-
confinement 共‘‘H-mode’’兲 cases.26 Such experimental inves-
tigations of confinement scaling properties remain a chal-
lenging area of investigation. Nevertheless, for the larger
sized reactor-scale plasmas of the future, the present simula-
tions indicate that the relative level of turbulent heat loss
from electrostatic turbulence does not increase with size. The
underlying question of why such a transition might occur
around the 400 gyroradii range indicated by the simulations
is being explored. Although this is clearly a very favorable
prediction, the fidelity of the analysis needs to be further
examined by investigating additional physics effects which
might alter the present predictions. The analysis of associ-
ated scientific issues will naturally demand more comprehen-
sive physics models within microturbulence codes. In addi-
tion to addressing experimental validation challenges, the
interplay between analytic theory and advanced simulations
will be increasingly important. For example, impressive
progress in physics understanding of the nonlinear processes
associated with zonal flow dynamics has resulted both from
directions provided simulations by analytic theory as well as
by simulation results which have inspired new analytic
models.27–30
A key multi-scale challenge for kinetic simulations is the
realistic implementation of complete electron 共‘‘nonadia-
batic’’兲 physics and electromagnetic dynamics. On the imme-
diate horizon, present capabilities for dealing with electro-
static perturbations are being successfully extended to
include nonadiabatic electrons.31 Much more challenging, es-
pecially for global simulations, are the electromagnetic
waves, which can alter the stability properties of the electro-
static modes and also generate separate instabilities associ-
ated with deformations of magnetic surfaces. In fact, answer-
ing the long standing question about what causes the
ubiquitously-observed anomalously large electron thermal
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1865
transport is likely linked to the ability to deal with magnetic
perturbations. They can potentially cause a great increase in
electron heat flux either through transient deformations of
the magnetic field 共‘‘magnetic flutter’’兲 or by producing an
ergodic region in which the magnetic field lines no longer
rest on nested flux surfaces but wander instead through a
finite volume ‘‘breaking’’ the flux surfaces.
The most favorable actual confinement of fusion-
relevant plasmas occurs in highly localized regions associ-
ated with the spontaneous formation of transport barriers.
These regions, in which the turbulent transport is greatly
reduced or completely vanishes, are associated with regions
of high electric field shear and very steep pressure gradients.
Open questions include identifying the physics behind their
radial structure and propagation as well as accounting for
their experimentally observed power threshold. Since experi-
ments indicate that electron transport can be large in such
regions where the longer-wavelength modes 共e.g., ITG兲 are
predicted to be absent, the possible relevance of electron
temperature gradient 共ETG兲 instabilities has been actively
pursued.32 These are extremely short-wavelength modes
共scale-lengths between the electron gyroradius and the colli-
sionless skin depth兲 which demand a kinetic description of
the electron species. If such modes prove to be as important
as the more familiar longer wavelength instabilities, dealing
with the associated electron/ion mass ratio issues 共e.g., si-
multaneously resolving ITG and ETG turbulence兲 would in-
troduce a huge spatial scale problem that would be most
imposing even with terascale simulation capabilities.
While much work remains before a realistic computa-
tional model, which captures the key physics behind
microturbulence-driven transport, is at hand, encouraging
progress has been made. For example, flux-surface-averaged
models of heat transport, developed by fitting the ion thermal
conductivity to results from nonlinear microturbulence simu-
lations, have achieved reasonable agreement when compared
against actual magnetic confinement experiments.33
D. Plasma boundary physics simulations
Understanding plasma boundary physics, which includes
plasma–wall interactions and turbulent transport in the
geometrically-complex plasma edge region, is a most formi-
dable challenge. Advances in the capability to reliably pre-
dict interactions between the plasma and its material inter-
faces have come primarily in tokamak divertor research. This
has necessitated the development of models dealing with
atomic physics and the presence of neutral gas which is able
to penetrate into the plasma edge. Improved understanding
will require proper formulation of the basic equations de-
scribing complex experimentally-observed processes such as
filamentation and bursting behavior. The generalization of
present laminar models to more realistic 3-D descriptions
will require a strong interplay between advanced computing
and benchmarking against theoretical models motivated by
experimentally-observed trends. Progress on improved mod-
eling of the atomic physics at the plasma edge will also be
actively pursued with advanced computational models.34
1866 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 13. 共Color兲 Initial k-spectrum comparisons between recent gas-puff
imaging experimental measurements and 3-D electromagnetic fluid code
simulations of edge turbulence on the Alcator C-Mod tokamak.
With regard to transport at the plasma boundary, the chal-
lenge is to develop a better physics description of edge tur-
bulence. This is of course related to issues discussed in the
preceding section. Magnetically confined plasmas terminate
at a last closed flux surface, which is defined by either the
presence of a material boundary, or by a ‘‘separatrix’’—a
magnetic surface separating the closed, nested flux surfaces
from the open ones that strike a material surface. Particles,
momentum, and heat are rapidly conducted along open field
lines to the material surface. This results in the formation of
a boundary layer about the last closed flux surface where the
radial gradients steepen until the cross-field transport can
compete with the rapid transport parallel to the equilibrium
magnetic field. An analysis of this boundary layer is greatly
complicated by the convergence of scale lengths 共i.e., the ion
orbit width, which provides the characteristic scale of the
turbulence, is actually comparable in magnitude to the scale
length of the density and temperature gradients兲. The associ-
ated absence of small expansion parameters obviously com-
plicates the analysis. Nevertheless, substantial progress has
been made using electromagnetic fluid codes to simulate the
edge plasma turbulence.35 As shown in Fig. 13, initial com-
parisons of the wave-number 共k兲 spectrum from 3-D electro-
magnetic fluid code simulations of edge turbulence with re-
sults from very recent gas-puff imaging experimental
measurements on the Alcator C-Mod tokamak show a favor-
able level of agreement which should encourage further
studies.36 This is also a good illustration of the richer physics
validation described in Sec. II, which is dependent on the
availability of advanced experimental diagnostics capable of
producing integrated measurements of key physical quanti-
ties such as spectra, correlation functions, heating rates, and
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
FIG. 14. 共Color兲 3-D gyrokinetic particle-in-cell codes have demonstrated
excellent scaling as the number of processors is increased.
other variables of interest. Other outstanding challenges fac-
ing plasma boundary physics simulations in the future in-
clude 共i兲 self-consistent coupling between the plasma bound-
ary and core heat and particle fluxes; and 共ii兲 advancing
beyond fluid equations to deal with kinetic dynamics.
III. TERASCALE COMPUTING CHALLENGES
Addressing the scientific issues described in this review
requires advanced code development efforts which are im-
portant for most areas of research. The primary task involves
enhancing the physics models and developing more efficient
algorithms to efficiently deal with the associated problems.
Challenges include 共i兲 multi-scale physics such as kinetic
electromagnetic dynamics which have been discussed in the
previous section; 共ii兲 improved algorithms; and 共iii兲 scalabil-
ity of codes. With regard to item 共ii兲, in addition to making
sure presently working algorithms are scalable to new com-
puting platforms, innovative numerical algorithms will have
to be invented to make progress. In addition, powerful ap-
proaches such as adaptive mesh refinement for higher dimen-
sionality phase-space need to be actively pursued. Item 共iii兲
deals with the ability to efficiently implement existing codes
on the most powerful MPP supercomputers to enable simu-
lations of larger problems. In fact, for virtually all of the
topical areas covered in this review, the plasma science com-
munity has had success in developing codes for which com-
puter run-time and problem size scale well with the number
Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 15. 共Color兲 3-D particle-in-cell code simulations of dynamics in the
magnetic reconnection current layer leading to the generation of localized
regions of an intense anti-parallel electric field 共‘‘electron holes’’兲.
of processors on massively parallel machines. An example of
this, as illustrated in Fig. 14, is the microturblence code,
GTC, which has recently demonstrated full scalability for
2000 processors.
It should be emphasized that a consequence of the effec-
tive utilization of supercomputers is the tremendous amount
of data generated. The microturbulence example described
earlier in this paper alone produced over 2 terabytes of data.
Hence, coordinated efforts will be essential not only to help
accelerate progress on the impressive state-of-the-art physics
advances but also to provide the necessary tools for data
visualization, mining, and manipulation. Means of dealing
with the ‘‘data glut’’ in the interactive exploratory visualiza-
tion of terascale simulations, including image rendering,
must be developed. Another key priority involves developing
a set of diagnostic and visualization tools that will allow
real-time interaction with the simulated data. This is impor-
tant for assisting applications scientists in testing theories/
hypotheses and in answering specific questions about the key
physics within the computational models. Recent efforts us-
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1867
FIG. 16. 共Color兲 Massively parallel supercomputers effectively utilized to
design future experimental facilities via optimization of confinement and
constructability properties.
ing VTK have significantly advanced this capability in the
microturbulence simulation area.20 Also, in order to realize
the benefits from advancements in understanding, it will be
necessary to periodically update existing integrated models
to ensure that they reflect the fresh insights gained from
these new ‘‘tools for discovery.’’
Improvements in plasma diagnostic techniques have
made it increasingly feasible to demonstrate more in-depth
correlations between experimental results and theoretical
models. For example, the development of diagnostic instru-
ments capable of making high resolution measurements of
electric and magnetic fields and cross-sectional measure-
ments of turbulent fluctuations have greatly improved the
1868 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 17. 共Color兲 Interfacing of MPP simulation of microwave reflectometry diagnostic with large-scale microturbulence simulation results.
basic understanding of the mechanisms controlling plasma
confinement. As in dealing with the output from terascale
simulations, maximizing the effectiveness of simulation/
experiment comparisons will also necessitate addressing
critical computer science and enabling technology 共CSET兲
issues in the area of data management and visualization. In-
deed, the power of advanced computing to solve challenging
problems can be fully exploited only if a capable infrastruc-
ture is established and effective software tools are made
available. Terascale computing requires complementary soft-
ware that scales as well as the hardware and which provides
an efficient code development environment. In general, net-
working, database, visualization, and other infrastructure
tools are critically needed to strengthen the coupling between
terascale simulations with theory and experiment.
The applications development challenge for terascale
computing will invariably involve multiple research institu-
tions. Hence, efficiency will require system integration and
the availability of middleware software that allows maximal
use of available computing platforms to support joint appli-
cation development projects. Accordingly, modern object-
oriented code development methods are needed to facilitate
sharing code development efforts among collaborators from
numerous research groups.
As emphasized throughout the course of this review, it is
critically important for advanced simulations to be rigorously
validated against experiments to ensure the fidelity of the
science. Progress in this direction requires strong collabora-
tion between the CSET community and application areas
such as plasma science. An encouraging step toward the fu-
ture has been the establishment of a national fusion energy
sciences collaboratory project which targets advances in con-
nectivity to experimental data and to state-of-the-art tools for
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
data visualization, mining, and manipulation. As part of the
new DOE SciDAC Program, this project forges for the first
time a partnership between computational applications scien-
tists at the three largest U.S. magnetic fusion experimental
facilities at General Atomics 共San Diego, CA兲, PPPL 共Prin-
ceton, NJ兲, and MIT 共Cambridge, MA兲 together with com-
puter scientists at Argonne National Laboratory, Lawrence
Berkeley National Laboratory, and the Computer Science
Departments at the University of Utah and Princeton Univer-
sity. This cross-disciplinary effort37 has set goals including
共i兲 more efficient use of experimental facilities via powerful
new between-pulse data analysis capabilities; 共ii兲 better ac-
cess by researchers to analysis and simulation codes, data,
and visualization tools; and 共iii兲 the creation of a standard
tool set for remote data access, security, and visualization.
IV. CROSS-DISCIPLINARY OPPORTUNITIES
The increased emphasis on advanced computations in
fusion energy sciences carries an exciting vision for the fu-
ture which can serve the vitally important role of helping to
attract and educate the bright young people essential for the
future technological health of this field. It can also stimulate
mutually beneficial scientific alliances with other applica-
tions areas. For example, effective modeling of global sys-
tems, combustion, and fusion devices all deal with complex,
3-D, nonlinear fluid flows and associated kinetic dynamics,
albeit in very different parametric regimes. They share the
common computational challenge to rapidly develop ad-
vanced integrated modeling capabilities capable of treating
complex dynamical systems covering many decades in time
and space. The computational challenges posed by nonlinear
Phys. Plasmas, Vol. 9, No. 5, May 2002 Advanced computations in plasma physics 1869

plasma fluid problems share many common features with

computational fluid dynamics 共CFD兲 issues faced in global
systems modeling, materials sciences, the DOE’s ASCI 共de-
fense兲 program, and numerous other areas. As for non-DOE
agencies, plasma physics has long-standing programs with
NASA and NSF in the space physics and basic science are-
nas which can be further strengthened. Additionally, ad-
vanced computational modeling capabilities will also stimu-
late progress in nonfusion plasma applications, including the
manufacture of microelectronic components, plasma display
panels, plasma thrusters for satellites, and the use of plasmas
for waste remediation. Realistic physics-based plasma mod-
els will also help accelerate the pace of breakthroughs in
plasma science applications to energetic beams, to space
physics, to solar physics, and to astrophysics. Numerous at-
tractive opportunities currently exist not only for cross-
disciplinary interactions between a scientific application area
and applied math and computer science, but also for cross-
cutting collaborations among various application disciplines.
Some representative examples include the following.
共1兲 Magnetic Reconnection Simulations: Magnetic re-
connection is a research topic of great interest in plasma
physics as well as other areas such as space physics and
astrophysics.38 This is a fundamental dynamical process
which enables a magnetized plasma system to convert mag-
netic energy into high speed flows, thermal energy, and en-
ergetic particles. In order to understand the kinetic processes
occurring at very small scales which control the release of
magnetic energy on a global scale, it is necessary to address
physics coupling across vastly disparate spatial and temporal
scales. This basically involves dynamics in narrow layer re-
gions where long-mean-free path kinetic dynamics are im-
portant and distant regions where a macroscopic description
with proper boundary conditions enforced is appropriate. Ac-
cordingly, the associated research for this multi-scale grand
challenge problem involves the application of both MHD
and kinetic analysis capabilities. In addition to significant
theoretical advances,39 dedicated laboratory experiments are
now actively exploring the structure of the current sheet
which defines the ‘‘dissipation region.’’40 In the advanced
simulation area, a promising new cross-cutting initiative
within the plasma science component of the SciDAC Pro-
gram involves a collaboration between plasma scientists at
the University of Iowa and the University of Texas at Austin
with computational astrophysics researchers at the Univer-
sity of Chicago.41 A single-fluid MHD capability has recently
been implemented in the University of Chicago’s ‘‘Flash
Code,’’42 which solves the fully compressible Navier–Stokes
equations and has been successfully utilized on a large and
diverse scope of applications including astrophysics-relevant
problems. With attractive modern features including adaptive
mesh refinement capabilities implemented, this fully parallel
code, which won the Supercomputing 2000 Gordon Bell
FIG. 18. 共Color兲 3-D particle-in-cell simulation of electron–proton two-
Prize, will now target two-fluid simulations of magnetic re- stream instability.
connection dynamics with focus on Hall current physics is-
sues. Regarding plasma physics advances impacting other nomena in magnetospheric and solar physics research. This
areas of research, computational progress in magnetic recon- includes topics such as solar flares, coronal mass ejection,
nection studies in laboratory plasmas are having an increas- and magnetic sub-storms in the magnetosphere. A good ex-
ingly significant impact on the exploration of associated phe- ample comes from the collaboration led by the University of

Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
1870 Phys. Plasmas, Vol. 9, No. 5, May 2002
FIG. 19. 共Color兲 Example of cross-disciplinary graduate program targeting education in areas of computational science applications 共including plasma
physics兲 together with applied math and computer science.
Maryland and involving researchers at the Max Planck Insti-
tute for Plasma Physics and Dartmouth College. They have
recently carried out particle simulations 共using 70M particles
and 20M grid points in 3-D simulations with periodic bound-
ary conditions兲 in which the large reduction of the magnetic
reconnection current layer width is associated with the rapid
acceleration of electrons.43 As illustrated in Fig. 15, the re-
sultant ‘‘electron beam’’ is found to produce a two-stream
instability which nonlinearly evolves into ‘‘electron
holes’’—localized regions of intense anti-parallel electric
field. The authors have pointed out that these new simulation
results are quite possibly relevant to recent polar satellite
observations of qualitatively similar phenomena in regions
where magnetic reconnection could be present.
共2兲 Design of Advanced Experiments. In the recent na-
tional effort to design the National Compact Stellarator
Experiment,44 the efficient utilization of massively parallel
processing 共MPP兲 computations was essential for the optimi-
zation of the key stability and transport properties in this
complex plasma configuration 共see Fig. 16兲. Moreover, the
MPP computations were critical for addressing ‘‘constructa-
bility’’ issues which involved the proper design of the mag-
netic coil system that could produce the plasma of interest.
共3兲 Modeling of Advanced Diagnostics: To gain a better
understanding of the complex processes impacting plasma
turbulence, theorists, computer scientists, and experimental-
ists have recently developed a MPP simulation of the actual
microwave reflectometry diagnostic.45 As illustrated in Fig.
17, the new MPP code models the actual incoming micro-
waves reflecting off a target plasma which evolves as its
turbulence grows. This is interfaced with the latest large-
scale microturbulence simulation results described
earlier.23,24 The characteristic turbulent correlation length can
be extracted from this reflectometer simulation and com-
pared with that deduced from the original microturbulence
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
W. M. Tang
simulation. This innovative new capability can cost-
effectively aid the interpretation of the reflectometry data
from experiments. It should also help to more efficiently
utilize the existing diagnostic capabilities and to design
improved diagnostic capabilities for the future.
共4兲 High Energy and Accelerator Physics: Another ex-
ample of some recent computationally-driven advances in
plasma physics which impact progress in other disciplines
involves the high energy and accelerator physics area. Uti-
lizing a particle simulation approach 共‘‘delta-f’’ method兲 pio-
neered in magnetic confinement studies,8 a new three-
dimensional particle code for both ions and electrons has
successfully simulated electron-proton two-stream instabili-
ties from initial noise.46 Results obtained from tracking about
one million particles over one million lattice periods 共beta-
tron oscillations兲 are shown in Fig. 18. Developed within the
workscope of the Heavy Ion Fusion Virtual National Labo-
ratory 共LBNL, LLNL, PPPL兲,47 this code has addressed a
key problem of interest to the high energy and accelerator
community—the presence of unwanted electron populations
observed in the LANL Proton Storage Ring Experiment and
expected to be an issue for the Spallation Neutron Source
Project. A valuable finding from these new simulations car-
ried out on powerful MPP platforms is that this potentially
troublesome instability can be eliminated with modest axial
momentum spread.
Finally, it is important to emphasize the fact that the area
of computational plasma science needs to actively develop
and nurture opportunities for attracting, educating, and re-
taining bright young talent essential for the future of the
field. While it is encouraging that many of the key advances
cited in this review have involved major contributions from
young scientists, greater efforts are required to educate the
next generation of researchers with capabilities that cross-cut
Phys. Plasmas, Vol. 9, No. 5, May 2002
traditional boundaries. Accordingly, some major universities
are beginning to respond with graduate educational programs
that target the development of capabilities that ‘‘bridge’’ vari-
ous areas of computational science applications 共including
plasma physics兲 together with applied math and computer
science. An example is a NSF-supported graduate program in
integrative computer and applications science 共‘‘PiCASso’’兲
recently begun at Princeton University. As illustrated in Fig.
19, the collaborating departments and participants foster in-
terdisciplinary research for the mutual benefit of all partici-
pating areas.
V. CONCLUSION
Advanced computations are cost-effectively aiding
progress toward gaining the physics knowledge needed to
harness fusion energy by making key contributions to all
areas of plasma science. The plasma physics community is
both effectively utilizing and contributing to the exciting ad-
vances in information technology and scientific computing.
This has produced a more stimulating environment for trans-
forming research and accelerating scientific understanding,
innovation, and discovery. New resources in this area have
already begun to enable productive partnerships involving
laboratories, universities, and industries, and also a sharing
with other fields of the insights gained in the process of
obtaining successful solutions to scientific problems of ex-
traordinary complexity. Indeed, advanced computation is a
natural bridge for fruitful collaborations between scientific
disciplines which will be of mutual benefit to all areas. The
advanced computational achievements and approaches high-
lighted in this review have the potential to revolutionize
plasma physics research by improving scientific understand-
ing of experimental data, by stimulating new theoretical
ideas, and by helping produce innovations leading to the
most attractive and viable designs for future facilities. Com-
putational plasma science is also helping to attract, educate,
and retain bright young talent essential for the future of the
field.
ACKNOWLEDGMENTS
The author is most grateful for the excellent input mate-
rial provided by the many members of the plasma science
advanced computational community. This review is a testi-
monial to their outstanding productivity which has enabled
an increasing appreciation in the general scientific commu-
nity for both the high quality and future promise of advanced
computations in plasma physics.
This work was supported by the U.S. Department of En-
ergy Contract No. DE-ACO2-76CH03073.
1 36
D. E. Stokes, Pasteur’s Quadrant: Basic Science and Technological Inno-
vation 共Brookings Institution Press, Washington, DC, 1997兲.
2 37
Department of Energy Office of Science’s ‘‘Scientific Discovery through
Advanced Computing 共SciDAC兲 Program’’ 共Department of Energy, Wash-
ington, DC, 2001兲; T. Dunning 共private communication, 2001兲.
3
‘‘An Assessment of the Department of Energy’s Office of Fusion Energy
Sciences Program,’’ National Research Council, Fusion Science Assess-
ment Committee, Final Report 共National Academy Press, Washington,
DC, 2001兲.
4
P. H. Rutherford and E. A. Frieman, Phys. Fluids 11, 569 共1968兲; J. B.
Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp
Advanced computations in plasma physics 1871
Taylor and R. J. Hastie, Plasma Phys. 10, 479 共1968兲; P. J. Catto, W. M.
Tang, and D. E. Baldwin, ibid. 23, 639 共1981兲; E. A. Frieman and L. Chen,
Phys. Fluids 25, 3 共1982兲.
5
J. P. Freidberg, Ideal Magnetohydrodynamics 共Plenum, New York, 1987兲;
R. B. White, Theory of Toroidally Confined Plasmas 共Imperial College
Press, London, 2001兲.
6
J. M. Dawson, Rev. Mod. Phys. 55, 403 共1983兲.
7
W. W. Lee, Phys. Fluids 26, 556 共1983兲; J. Comput. Phys. 72, 243 共1987兲.
8
A. M. Dimits and W. W. Lee, J. Comput. Phys. 107, 309 共1993兲; S. E.
Parker and W. W. Lee, Phys. Fluids B 5, 77 共1993兲.
9
W. Park, E. V. Belova, G. Y. Fu, X. Z. Tang, H. R. Strauss, and L. E.
Sugiyama, Phys. Plasmas 6, 1796 共1999兲.
10
S. C. Jardin, Principal Investigator, Princeton Plasma Physics Laboratory,
J. Callen, W. Park, DOE Fusion SciDAC Extended Magnetohydrodynamic
Modeling Project, 2001.
11
W. Park, E. D. Fredrickson, A. Janos, J. Manickam, and W. M. Tang, Phys.
Rev. Lett. 75, 1763 共1995兲.
12
A. H. Glasser et al., Plasma Phys. Controlled Fusion 41, A747 共1999兲; C.
R. Sovinec, J. M. Finn, and D. del-Castille-Negrete, Phys. Plasmas 8, 475
共2001兲.
13
W. Park, E. V. Belova, and G. Y. Fu, Phys. Plasmas 6, 1796 共1999兲; L. E.
Sugiyama, W. Park, and H. R. Strauss, Nucl. Fusion 41, 739 共2001兲.
14
T. Hayashi, T.-H. Watanabe, and T. Sato, Phys. Plasmas 6, 1253 共1999兲.
15
D. Batchelor, Principal Investigator, Oakridge National Laboratory, C. K.
Phillips, P. T. Bonoli, DOE Fusion SciDAC Wave-Plasma Interactions
Project, 2001.
16
E. F. Jaeger, L. A. Berry, and E. D’Azevedo, Phys. Plasmas 8, 1573
共2001兲.
17
M. Lieberman, ‘‘Design of high-density sources for materials processing,’’
in Plasma Sources for Thin Film Deposition and Etching, edited by M. H.
Francombe and J. L. Vossen 共Academic, San Diego, 1994兲.
18
F. Chang-Diaz and G. Emanuel, Aerosp. Am. 38, 12 共2000兲.
19
S. R. Cranmer, Astrophys. J. 532, 1197 共2000兲.
20
W. Nevins, Principal Investigator, Lawrence Livermore National Labora-
tory, R. E. Waltz, W. W. Lee, DOE Fusion SciDAC Plasma Microturbu-
lence Project, 2001.
21
R. E. Waltz, Bull. Am. Phys. Soc. 46, 322 共2001兲.
22
W. M. Tang, J. W. Connor, and R. J. Hastie, Nucl. Fusion 20, 1439 共1980兲.
23
Z. Lin, T. S. Hahm, and W. W. Lee, Science 281, 1835 共1998兲.
24
T. S. Hahm, Bull. Am. Phys. Soc. 46, 321 共2001兲.
25
R. V. Budny, J. P. Christiansen, and J. G. Cordey, Phys. Plasmas 7, 5038
共2000兲; T. L. Rhodes, Bull. Am. Phys. Soc. 46, 323 共2001兲.
26
C. Petty, M. R. Wade, J. E. Kinsey, D. R. Baker, and T. C. Luce, Phys.
Plasmas 9, 128 共2002兲.
27
M. N. Rosenbluth and F. L. Hinton, Phys. Rev. Lett. 80, 724 共1998兲.
28
L. Chen, Z. Lin, and R. White, Phys. Plasmas 7, 3129 共2000兲.
29
P. H. Diamond, S. Champeaux, and M. Malkov, Nucl. Fusion 41, 1067
共2001兲.
30
M. A. Malkov, P. H. Diamond, and M. N. Rosenbluth, Phys. Plasmas 8,
5073 共2001兲.
31
Z. Lin and L. Chen, Phys. Plasmas 8, 4435 共2001兲; W. W. Lee, J. L. W.
Lewandowski, T. S. Hahm, and Z. Lin, ibid. 8, 1435 共2001兲.
32
F. Jenko, W. Dorland, M. Kotschenreuther, and B. N. Rogers, Phys. Plas-
mas 7, 1904 共2000兲; W. Dorland, F. Jenko, M. Kotschenreuther, and B. N.
Rogers, Phys. Rev. Lett. 85, 5579 共2000兲.
33
A. M. Dimits, G. Bateman, M. A. Beer et al., Phys. Plasmas 7, 969
共2000兲.
34
M. S. Pindzola, Principal Investigator, Auburn University, D. R. Schultz,
F. J. Robicheaux, DOE Fusion SciDAC Atomic Physics Project, 2001.
35
K. Hallatschek and A. Zeiler, Phys. Plasmas 7, 2554 共2000兲; B. Rogers, J.
F. Drake, and A. Zeiler, Phys. Rev. Lett. 81, 4396 共1998兲; X. Q. Xu, R. H.
Cohen, T. D. Rognlien, and J. R. Myra, Phys. Plasmas 7, 1951 共2000兲; F.
Jenko and B. D. Scott, ibid. 6, 2705 共1999兲.
S. Zweben, J. L. Terry, and M. Greenwald, Bull. Am. Phys. Soc. 46, 322
共2001兲.
D. Schissel, Principal Investigator, General Atomics, D. McCune, M.
Greenwald, DOE Fusion SciDAC Collaboratory Project 共Department of
Energy, Washington, DC, 2001兲.
38
V. M. Vasiliunuas, Rev. Geophys. Space Phys. 13, 303 共1975兲; E. N.
Parker, Cosmical Magnetic Fields 共Clarendon, Oxford, 1979兲; Astrophys.
J. 180, 247 共1973兲; T. Sato and T. Hayashi, Phys. Fluids 22, 1189 共1979兲;
D. Biskamp, ibid. 29, 1520 共1986兲; R. M. Kulsrud, Phys. Plasmas 5, 1599
共1998兲; W. M. Tang, Science 279, 1488 共1998兲.
1872 Phys. Plasmas, Vol. 9, No. 5, May 2002 W. M. Tang

39
J. Birn, J. F. Drake, and M. A. Shay, J. Geophys. Res. 106, 3715 共2001兲; A. 43
M. Swisdak, J. F. Drake, and B. N. Rogers, Bull. Am. Phys. Soc. 46, 143
Y. Aydemir, Phys. Fluids B 4, 3469 共1992兲; B. N. Rogers, W. Dorland, and 共2001兲.
F. Jenko, ‘‘The role of dispersive waves in collisionless magnetic recon- 44
A. Reiman, L. Ku, D. Monticello et al., Phys. Plasmas 8, 2083 共2001兲.
45
nection,’’ submitted to Phys. Rev. Lett. R. Nazikian, E. Valeo, and K. Shinohara, Bull. Am. Phys. Soc. 46, 155
40
M. Yamada, H. Ji, T. Carter, R. Kulsrud, and F. Trintchouk, Phys. Plasmas 共2001兲.
7, 1781 共2000兲; M. Yamada, Earth Planets Space 53, 509 共2001兲. 46
H. Qin, R. Davidson, and W. W. Lee, Phys. Rev. ST Accel. Beams 3,
41
A. Bhattacharjee, Principal Investigator, University of Iowa, R. Rosner, R. 08440 共2000兲; 3, 109901 共2000兲.
47
Fitzpatrick, DOE Fusion SciDAC Magnetic Reconnection Project, 2001. R. O. Bangerter, ‘‘Ion beam fusion,’’ Proceedings of the Royal Society,
42
R. Rosner, Principal Investigator, University of Chicago, A. Calder, T. London, Meeting of 15–16 July 1998; preprints available as LBNL Report
Linde, University of Chicago Flash Code Project, 2000. HIFAN #943, LBNL #41564, UC-419.

Downloaded 23 Apr 2002 to 198.35.4.133. Redistribution subject to AIP license or copyright, see http://ojps.aip.org/pop/popcr.jsp