Session L - Poster Session.
INVITED session, Tuesday evening, August 26
, Cowell Dining Hall

[L.01] Spectral Reduction for Two-Dimensional Turbulence

A new method for predicting the statistical properties of fluid turbulence, called Spectral Reduction, is described. Collections of Fourier wavenumbers are represented by certain nonuniformly spaced sample modes that interact via enhanced coupling coefficients. The approximation reduces to the exact Navier--Stokes equation as the number of fundamental wavenumbers associated with each sample mode tends to one. Even at large values of this parameter, the statistics of the exact dynamics may be recovered from the time-averaged predictions of the theory to high accuracy. The agreement with full pseudospectral simulations appears to be remarkably good, even in flows containing long-lived coherent structures. The method is used to illustrate a modification to Kraichnan's logarithmically corrected enstrophy cascade.(J.~C. Bowman, J. Fluid Mech.) 306, 167 (1996). Non-Gaussian features of the higher-order statistics, such as the kurtosis, are also correctly captured. In the inviscid limit, Spectral Reduction satisfies a Liouville theorem. The method could be used to assess the effect of various dissipation mechanisms in large-eddy simulations, as a subgrid model, or as a substitute for full simulation of high-Reynolds turbulence.

[L.02] Cubic-spline technique for solving the compressible Navier-Stokes equations for turbulent convection

A cubic spline is used to solve the conservative form of the fully compressible Navier-Stokes equations to model high-Rayleigh-number convection. The spline function is used to determine both the gradients and the interpolated values of the state variables, while the fourth-order Runge-Kutta method is used to integrate the system in time.

The method is used to model high-Rayleigh-number convection in a rectangular cell with periodic side boundaries and rigid upper and lower boundaries. The well known power-law relation linking the Nusselt and Rayleigh numbers, NU-1 \sim Ra^2 øver 7 is recovered, as are the asymmetric (non-Boussinesq) convection plumes observed in experiments.

[L.03] A Large Eddy Simulation of Rayleigh-Taylor Instability with ALE

This research will address the application of the Arbitrary Lagrangian Eulerian (ALE) method to Rayleigh-Taylor instability problems. Finite volume methods will be used on a logically hexahedral grid. Direct numerical simulation will be compared with various large eddy simulations. Several subgrid scale turbulence models will be used, including the Smagorinski model, Smagorinski with Leith's buoyancy extension, and a single equation K model. Numerical results will be compared with experimental results.

This work was performed under the auspices of the United States Department of Energy at the Lawrence Livermore National Laboratory under contract number W-7405-Eng-48.

[L.04] The Motion of the Vortex-Current Filaments.

The motion of the vortex-current filaments in ideal MHD is examined via the cut-off Biot-Savart numerical simulations. The vortex-current filament consists of the electric current and vorticity which are parallel, or anti-parallel, along the axis of the filament. We have analytically obtained the explicit formula of the macroscopic force balance equation of the vortex-current filaments in our previous work. The obtained equation is correct up to the order of \rho^-2 where \rho is the local radius of curvature of the filament. Furthermore, our force balance equation coincides with the cut-off Biot-Savart integral up to the O(\rho^-1). It deserves attention to clarify the difference between the vortex filament and vortex-current filament. At present we analyze the motion of the vortex-current filaments numerically using the cut-off Biot-Savart integral. In the elliptic ring filament simulation, we have observed the Widnall's instability which is well-known phenomena of the ``pure" vortex filament. It is remarkable that the period of the Widnall's instability becomes infinite at a critical electric current inside the filament. We are now carring on the analysis of the collision of two filaments. In this simulation, the chaotic configuration of filaments is caused by the interaction between the two filaments. However, we find that the space-averaged distribution of the vorticity and electric current exhibits a very simple reconnected trajectory. We consider the correct understanding of the result needs further examination.

[L.05] Macroscopic Surface Tension in a Lattice Boltzmann BGK Model of Two Immiscible Fluids.

We present a method by which an interface generating algorithm, similar to that of earlier lattice Boltzmann models of immisible fluids, may be extended to a two component, two-speed D2Q9 lattice Bhatnagar Gross Krook fluid. For two-dimensional, microcurrent-free planar interfaces between the two immiscible fluids we derive expressions for static interfacial tensions and interfacial distributions of the two fluids. Extending our analysis to curved interfaces we propose a scheme for incorporating the influence of interfacial microcurrents which is based upon general symmetry arguments and is correct to second order in lattice velocity. The analysis demonstrates that the interfacial microcurrents have only second order influence upon the macroscopic behaviour of the model. We find good agreement between our calculations and simulation results based on the microcurrent stream function and surface tension results from the pressure tensor or Laplace law.

[L.06] Lattice Boltzmann Simulation of Fluid Flow For Industrial Applications.

Recent modifications to the two-speed, two-dimensional lattice Bhatnagar--Gross--Krook method eliminate compressibility errors from the simulation of macroscopic fluids governed by the steady--state Navier--Stokes equation. We describe the means by which this modified scheme makes it possible to compute both pressure drops and flow fields, bringing internal, pressure driven, isothermal flows within the scope of the method. The scheme is applied to the simulation of flow of a viscous incompressible fluid past a sudden expansion in an infinite aspect ratio duct (back--facing step geometry). Results are presented and compared with experiment over a range of Reynolds number.

[L.07] A Computational Environment for Marine Seismic Tomography

We are building a computational environment for seismologists who study sub-oceanic volcanic regions through 3D tomographic image analysis. Tomographic images of the Earth's interior are obtained by solving an inverse problem that includes over 10^4 seismic wave travel time observations and over 10^6 basis functions; hundreds of calculations are required to search the solution space for a set of geologically plausible models. Our environment facilitates this search by providing high-performance computing, on-line model visualization, and the ability to interact with and control the computation at runtime. Runtime control liberates seismologists from the confinements of the traditional iterative analysis cycle of: compute, analyze results, modify parameters, recompute, etc. The increase in speed and the ability to interact with the computation reduces the time between changes to inversion parameters and the observation of the effects of those changes, and has already led to fundamental geologic discoveries beneath the East Pacific Rise.

[L.08] Parallel Computatinal Technology for Atmospheric Turbulence Modeling

Desktop Atmospheric Turbulence Diffussion Modeling System (DATDMS) is used by analysts with varied backgrounds for performing air quality assessment and emergency response activities. This modeling system must be robust, well documented, have minimal and well controlled user inputs, and have clear outputs. Existing coarse-grained parallel computers can provide significant increases in computation speed in desktop atmospheric dispersion modeling without considerable increases in hardware cost. This increased speed will allow for significant improvements to be made in the scientific foundations of these applied models, in the form of more advanced diffusion schemes and better representation of the wind and turbulence fields. This is especially attractive for emergency response applications where speed and accuracy are of utmost importance. This presentation describes one particular application of coarse-grained parallel computer technology to a desktop complex terrain atmospheric dispersion modeling system. By comparing performance characteristics of the coarse-grained parallel version of the model with the single-processor version, we will demonstrate that applying coarse-grained parallel computer technology to desktop atmospheric dispersion modeling systems will allow us to address critical issues facing future requirements of this class of dispersion models.

[L.09] An Advanced Complex Terrain Atmospheric Dispersion Model for Regional Air Quality Applications

Pacific Northwest National Laboratory has developed an atmospheric transport and diffusion model for Regional Air Quality Applications in non-uniform terrain area. A three-dimensional diagnostic wind module is used to specify the time- and space-varying winds over the modeling domain. A special feature of the wind module is that it accounts for flow channeling and blocking from major terrain features during stable atmospheric conditions. This model predicts ground-level concentrations and deposition fields of air contaminants released from point sources. Both wet and dry deposition and radioactive decay of the released material can be treated. It is applicable at source-to-receptor transport distances from a few hundred meters to a few hundred kilometers. The model can be launched at any time and uses the available terrain and meteorological files. Given release information by the user, the model produces concentrations and deposition amounts as a function of time at the receptor locations. A new Graphic User Interface (GUI) for running model on desktop PCs has recently been developed and implemented.

[L.10] Application of Multi-Quadric Interpolation Method to Meteorological Mapping in Pacific Northwest Region

The problem of mapping the scattered meteorological observations in the Pacific Northwest region to produce values on regular uniform grids for ecosystem modeling has presented meteorologists with a significant challenge because topography and vegetation are a prominent features in this region. An accurate mapping cannot be achieved without properly considering the effects of topography as well as vegetation type. Recently, one interpolation method, referred to as Multi-Quadric Interpolation Method (MQIM), has been developed to produce more accurate mapping results on very complex terrain. We have applied MQIM to map the scattered meteorological observations in the Pacific Northwest region and compared the results with those obtained from other more traditional methods methods and other models. In this paper, we will present some of our results, and will discusses the advantages and potential problems in applying MQIM for mapping meteorological observations over areas of complex terrain.

[L.11] Hybrid simulation of neutral gas interacting with a plasma

Plasma in contact with a material surface is neutralized and recycled as a gas that in turn interacts with the plasma. The neutral gas may be in a kinetic regime (long mean free path) in some regions and in a fluid regime (short mean free path) elsewhere. In order to model this situation, we imagine the neutral gas as the superposition of two populations, one fluid and one kinetic, with transfer terms coupling the two populations. We model these populations by coupling together a plasma fluid code, B2.5,(B.~J. Braams, Radiative Divertor Modelling for ITER and TPX, Contrib. Plasma Phys. 36), 276--281 (1996) and a neutral Monte Carlo code, Degas 2.\footnote D.~P. Stotler and C.~F.~F. Karney, Neutral Gas Transport Modeling with Degas 2, Contrib. Plasma Phys. 34, 392--397 (1994). The coupling terms conserve mass, momentum, and energy, and are chosen so that fluid neutrals are converted to kinetic neutrals where their mean free path is long and vice versa. In this scheme, self collisions are never a dominant term in the Monte Carlo code may be treated by a simplified BGK model.

[L.12] ITG Driven Mode Near Threshold

The ion temperature gradient (ITG) driven modes along with the trapped electron mode and pressure gradient ballooning modes are dominating instabilities for the most realistic tokamak parameters (low \beta and weak collisionality) and in many cases are responsible for ion anomalous transport in tokamak plasmas. Many experimental evidences and kinetic simulations indicate that the tokamak plasma profiles are in line with the assumption that plasma is near the boundary of marginal stability for \eta _i-mode. These results could be considered as support of the idea of self-organization of tokamak plasmas. The purpose of the present paper is to investigate nonlinear explosive instabilities of \eta _i-modes near the boundary of marginal stability and their possible saturation mechanisms. We demonstrate analytically and numerically that boundedness of the wave interaction area may affect the explosive instability dynamic. In this case the instability is saturated at sufficiently lower wave amplitudes if unstable perturbations escape from the interaction region in a smaller time comparing to the ``explosion'' time. Bounded wave interaction area is created in the self-organized process. This process may lead also to drift wave envelope nonlinear structures (solitons and vortices) formation. We find particular analytical solutions in the form of such structures and confirm numerically an existence of the solutions in cases of homogeneous as well as inhomogeneous background parameters.

[L.13] Modelling of low-pressure gas discharges using electron clouds

A selfconsistent, time-dependent, hybrid fluid-particle model of a dc low-pressure gas discharge is presented. The description includes all parts of the discharge, the cathode fall, the negative glow and the positive column. Emphasis is put on the emitted radiation of the discharge. Electrons are combined in gaussian-shaped clouds as a local distribution. These clouds are emitted by the cathode, gaining energy in the cathode fall and generating new electrons. Because of thermal movement clouds of the plasma electrons spread slowly. For each cloud there is a maxwell energy distribution below the first inelastic collision energy. Above this energy the distribution function is represented at discrete points in an irregular mesh. If clouds are too large, the whole electron-distribution in the discharge is divided into new clouds. Time-dependent fluid equations (continuity equations) are used to describe the ions. The electric field is calculated by the Poisson equation. With a given cold spot temperature, fill gas pressure, current, radius and length of the lamp, the electric field, the electron and ion density, the electron temperature and the radiation can be calculated.

[L.14] Diffusion in Self-Consistent Beam Dynamics

In this paper we analyze particle diffusion and emittance growth induced by discrete-particle effects in two-dimensional self-consistent numerical simulation studies of beam dynamics. In particular, an analytical model is presented which describes the slow time-scale variation of single particle emittance for a perfectly matched beam in a periodic solenoidal magnetic focusing field. It is found that the diffusion process is independent of the explicit form of the focusing magnetic field and has a quadratic dependence on the propagation distance s. Furthermore, the diffusion coefficient D is shown analytically to scale according to D\propto K^2/N_p, where K is the normalized perveance of the beam and N_p is the total number of macroparticles in the numerical simulation. Numerical results are presented in agreement with the theoretical predictions.

[L.15]

This abstract was not submitted electronically.

[L.16] End-to-End Modeling of a Heavy-Ion Fusion Acceler- ator

One of the main approaches to inertial fusion energy (IFE) employs a high-energy (E\sim 5\,GV), heavy-ion accelerator to produce an intense beam which, when focused onto the target capsule containing DT fuel, leads to implosion and thermonuclear ignition. We have begun an ``end-to-end'' computational modeling effort to examine design tradeoffs for the next test accelerator. This effort includes both 2- and 3-D electrostatic and electromagnetic PIC codes to study ion beam transport physics starting at the emission surface of multiple-beam injector and ending at the target pellet in the reactor vessel. The particular codes run on a variety of platforms ranging from single processor UNIX workstations to the CRAY T3E at NERSC. Areas of concern include beam quality preservation (both longitudinal and transverse), permissible tolerances on the multitude of focusing elements, and multiple beam interaction effects both in the accelerator and reactor vessel. This poster present some initial results on these topics.

[L.17] Simulation of Ion/Electron Beams in Electrostatic and Magnetostatic Fields with Embedded Curved Boundaries as Applied to Accelerators and Beam Injectors

The simulation code GYMNOS-ECB models the interaction of charged particles with their self-consistent and applied electrostatic and magnetostatic fields.(D.W. Hewett, submitted to J. Comput. Phys. 1997.) The core physics involves kinetic particle effects and the solution to the noninductive Maxwell's equations.(D.W. Hewett, Comp. Phys. Comm. 84 (1994) 243-277.) Boundaries for emission, collection, and applied fields can have an arbitrary shape. The Embedded Curved Boundary (ECB) method allows curved surfaces to be converted to piecewise linear segments on an orthogonal grid. Thus emission normal to surfaces is modeled with little evidence of the coarse underlying stairstep structure. Use of the orthogonal grid with the ECB method allows modeling of realistic structures while using a coarser grid and larger timestep than traditional methods. GYMNOS-ECB is being used to aid in engineering designs and physics studies on high energy ion beams and relativistic electron beams such as for beam size, current limits, and beam instabilities.

[L.18] 3-D Parallel, Object-Oriented, Hybrid, PIC Code for Ion Ring Studies

The 3-D hybrid, Particle-in-Cell (PIC) code, FLAME has been developed to study low-frequency, large orbit plasmas in realistic cylindrical configurations. FLAME assumes plasma quasineutrality and solves the Maxwell equations with displacement current neglected. The electron component is modeled as a massless fluid and all ion components are represented by discrete macro-particles. The poloidal discretization is done by a finite-difference staggered grid method. FFT is applied in the azimuthal direction. A substantial reduction of CPU time is achieved by enabling separate time advances of background and beam particle species in the time-averaged fields. The FLAME structure follows the guidelines of object-oriented programming. Its C++ class hierarchy comprises the Utility, Geometry, Particle, Grid and Distributed base class packages. The latter encapsulates implementation of concurrent grid and particle algorithms. The particle and grid data interprocessor communications are unified and designed to be independent of both the underlying message-passing library and the actual poloidal domain decomposition technique (FFT's are local). Load balancing concerns are addressed by using adaptive domain partitions to account for nonuniform spatial distributions of particle objects. The results of 2-D and 3-D FLAME simulations in support of the FIREX program at Cornell are presented.

[L.19] Parallel Simulation of Self-focusing of a Laser Pulse from Nonlinear Polarization

Due to the recent advance of short pulse lasers extremely intense laser beams have become available making it feasible to generate wake fields necessary for the implementation of a laser wake field accelerator. A key issue for the practical implementation of such an accelerator is the length over which the laser beam can self-focus. This determines the length over which electrons can be accelerated and, therefore, the final energy of the electrons. We examine numerically the effects of nonlinear polarization on the focusing of a short pulse laser beam. Maxwell's equations are implemented on a rectangular grid using a simple finite difference scheme with the polarization response modeled as a nonlinear harmonic oscillator. Although the method is computationally intensive, it is simple to implement on a parallel computer with various beam profiles for realistic situations. We present results from these simulations and compare them with recent experiments showing laser pulse self-focusing in various gases.

[L.20] Parallelization of an Edge Plasma Transport Code via Domain Decomposition.

Results from the parallelization of the 2-D finite-volume plasma fluid code UEDGE are described with applications to fusion-device edge plasmas. UEDGE solves the Braginskii plasma equations with classical transport along the magnetic-field, B, and anomalous transport across B, together with a fluid model for neutrals, and impurity transport determined by force balance. Because the transport along B is very rapid, the magnetic flux surfaces are retained as one of the coordinates, while the second coordinate can be at an arbitrary angle to the flux surfaces to fit material surface boundaries. This stiff, highly nonlinear system is solved using either of two fully-implicit, parallelized Newton-Krylov solvers: PVODE for time-dependent problems and PNKSOL for steady-state problems. An important part of the parallel solution algorithm is the user-specified preconditioner. Results of the performance of the domain-decomposition algorithm for the preconditioner are presented for the edge plasma problem where coupling between domains is neglected. The preconditioning Jacobian is calculated by finite-difference quotients.

[L.21] A Nonlinear 3D MHD Code NFTC for Numerical Simulations of Plasma Instabilities in Tokamaks

In this report a new nonlinear 3D MHD code NFTC is presented for the numerical simulations of magnetohydrodynamic (MHD) stability of plasmas. The nonlinear 3D evolution of a tokamak plasma is described by the full (nonreduced, compressible) MHD system of equations in general toroidal geometry. The equations include a viscosity, resistivity and sources. Arbitrary plasma rotation is included in terms of new equilibrium functions. The neoclassical effects such as the bootstrap current are considered in the MHD model.

A straight field line flux, nonorthogonal coordinate system is used corresponding to a given equilibrium described by the Grad-Shafranov equation. The solution is represented as finite Fourier series in both poloidal and toroidal angles. A fully implicit finite difference scheme is presently used in the radial direction. To resolve the quasilinear operators a Newton iterational method is applied. For the nonlinear terms Gauss-Zidel elemination is considered. The peculiarity of fully implicit scheme is studied. A specific regularizing algorithm is obtained to improve the numerical stability.

The developed NFTC code is ultilized for simulating of nonlinear MHD stability of plasma in experiments of DIII-D tokamak.

[L.22] Numerical Simulation and Spectral Analysis of Intense Laser-Plasma Interactions

A new numerical scheme is presented for simulating Raman Backscatter (RBS) of laser pulses in underdense plasma, based on operator-splitting in the dynamical evolution, fast trigonometric function evaluation, flexible models of plasma noise, and modern statistical estimation of power spectra. The model equations, developed earlier(G. Shvets, J.S. Wurtele, B.A. Shadwick, Phys. Plasmas) 5, 1 (1997). offer a simple alternative to Particle-in-Cell algorithms while allowing for careful comparison between simulation, theory, and experiment. RBS dynamics are of interest in many intense laser-plasma interactions and have close connections to free electron laser physics, providing a case study for driven growth of plasma instabilities from noise. Efficiency in integration is achieved by an operator-splitting method which can generate a stable finite-difference approximation of any order of accuracy. Evaluation of complex exponentials to a precision higher than the per-step error of the integration scheme proves wasteful. We employ a fast custom algorithm which calculates trigonometric functions only to the precision needed. We also assess alternative methods for estimating power spectra that are more efficient than traditional FFT-based techniques.

[L.23] Stability limits of spiral defects under restraint conditions

We show that the patterns in the solution of complex Ginzburg-Landau equation (CGLE) under restraint conditions are not equal to the patterns simulated normally. We gave the upper limit to the absolute value as restraint condition. Although the parameters which make the solution spatio-temporal chaos were used, we got spiral defects which can be usually observed in systems with less nonlinearity. The number of the spiral defects is minimum under a critical restraint condition.

[L.24] How Close Can One Approach a Sine-Gordon Homoclinic Orbit?

Ablowitz, Herbst and Schober have recently raised great concern about numerically induced spatial and temporal chaos in solving the Sine-Gordon equation (J. Comput. Phys. 126, 299 (1996)). They have shown that some previous symplectic integrable discretization schemes encounter difficulties, including numerically induced spatiotemporal chaos, when the initial values are ``exponentially close'' to a Sine-Gordon homoclinic orbit. Homoclinic-orbit-crossing seems to be unavoidable for numerical integrators. An interesting question is how close can one approach a Sine-Gordon homoclinic orbit without crossing it. In this work we introduce a recently developed Lagrange distributed approximating functional (LDAF) integrator (Phys. Rev. Lett., in press) to answer this question. The present approach is chaos free for the same numerical conditions suggested in the previous work which are exponentially close to the phase space homoclinic manifolds. Moreover, due to high accuracy in the LDAF discretization of the partial differential equation, the present approach is stable and homoclinic-orbit-crossing free for the phase space separatrix value \pi truncated up to 3.1415926. It is found that present approach is accurate up to 7 significant figures for computing the Sine-Gordon flow describing a breather-kink and antikink transition.

[L.25] Chaos and Energy Redistribution in the Spatio-Temporal Dynamics of Two Nonlinearly Coupled Wave Triplets

In this paper we examine the spatio-temporal dynamics of two nonlinearly coupled wave triplets(Phys. D, accepted (1997).). When spatial dependence is suppressed, the homogeneous manifold so obtained can be chaotic or regular. If chaotic, it drives energy diffusion from long to small wavelengths as soon as inhomogeneous perturbations are added to the system(Phys. Rev. E 54) 3239 (1996).. If regular, one has two possibilities: (i) energy diffusion is again present if the inhomogeneous modes are linearly unstable, or (ii) energy diffusion is totally absent if the inhomogeneous modes are linearly stable. We also discuss related topics as turbulence, spatio-temporal chaos, and thermalization.

[L.26] Simulation of Flame Patterns on a Circular Burner Using the K-S Equation

The stability of the flame front on a circular burner is a crucial criterion for the industrial design of burners. The Kuramoto-Sivashinsky (KS) equation has been widely regarded as the simplest description for the thermodiffusive instability. To preserve the O invariance of the system, the integration of the KS equation has to be carried out using lattice method in polar coordinates. Previous attempts to integrate the KS equation in the circular domain were hindered by the presence of the \nabla^4 u term, which when expanded in polar coordinates, gives terms with cross derivatives such as \partial_rr\theta\theta u. The standard implicit techniques of numerical integration fail in this situation. In the present study, we employ the distributed approximating functional (DAF) method for the spatial discretization of the KS equation in polar cooerdinates. A second order implicit scheme is utilized for the time discretization. Fourier-Bessel series expansion is used to identify the basic modes. Bifurcation theory is employed to analyze the structure and dynamics of patterns. A direct modal analysis of analogous experimental patterns using the Karhunen-Loéve analysis is given for comparison. Our results agree qualitatively with experimental observations.

[L.27] Liquid Crystals Displays: A Monte Carlo Study

Monte Carlo simulations have been widely used in investigating liquid crystals mainly for basic research studies. From this point of view the Lebwohl-Lasher (LL) model and its generalization has proven successful in providing an accurate microscopic-scale analysis which faithfully reproduces orientational LC behaviour. Recently we have applied the Monte Carlo (MC) technique to simulate models of liquid crystal displays starting from a microscopic interaction, trying, in such a way, to render more immediately applicable the MC metodologies developed for basic research.

Our models are based on a LL lattice spin system with suitable boundary conditions that mimic the aligned cell surfaces and with appropriate electric fields applied to a pixel array. We visualize the simulated display calculating optical textures from the Monte Carlo generated microscopic equilibrium configurations using a Müller matrix approach. Suitable orientational order parameters are calculated in the different regions of the sample. In particular here results for a display based on the in-plane switching effect are presented.

The simulations have been performed until now on relatively small lattices containing at most 10^5 particles. The present results can then be considered as a first attempt to study LC displays using Monte Carlo techniques. The use of more powerful computers should allow large-scale simulations with complete (pixel by pixel) control of a number of technologically relevant parameters such as display surface, material purity, etc...

[L.28] Monte Carlo Simulation of Spin density wave and ferromagnetism in a quasi-one-dimensional organic polymer

The spin density wave(SDW) -- charge density wave(CDW) phase transition and the magnetic properties in a half-filled quasi-one-dimensional organic polymer are investigated by the world line Monte Carlo simulations. The itinerant \pi electrons moving along the polymer chain are coupled radically to localized unpaired d electrons, which are are situated at every other site of the polymer chain. The results show that both ferromagnetic and anti-ferromagnetic radical couplings enhance the SDW phase and the ferromagnet order of the radical spins, but suppress the CDW phase. By finite size scaling, we are able to obtain the phase transition line in the parameter space. The ferromagnetic order of the radical spins are observed to co-exist with the SDW phase. As compared to the system being free of the radical coupling, the phase transition line is shifted upward in the U-V parameter space in favor of larger V, where U is the on-site repulsion and V is the nearest-neighbor interaction between the \pi electrons. All of these findings can be understood qualitatively by a second-order perturbation theory starting from the classical state at zero temperature in the strong coupling limit. We also address the consequences of the radical coupling for the persistent current if the polymer chain is fabricated as a mesoscopic ring.

[L.29] Radiation Background Study for Beam Monitoring Detector of Hall C Beam Line at Jefferson Lab

Beams of high energy electrons are used to conduct nuclear physics experiments at Thomas Jefferson National Accelerator Facility (TJNAF). Sufficient information about the beam profile and momentum distribution is necessary for most experiments. For on-line measurement of the beam profile and energy, a non-destructive beam profile monitor detector, based on residual gas ionization and utilizing microchannel plate technology, is being prototyped to be used in the Hall C beam line. A GEANT Monte Carlo simulation used to study the radiation background on the detector and the beam line will be discussed.

[L.30] Electrophoretic Deposition of Polymer Chains.

A Monte Carlo simulation is performed to study the evolution of the density profile of the chains and their size near an adsorbing wall. We consider a two dimensional lattice of size L_x \times L_y. Polymer chains are positively charged and released from one end of the sample (the source end) at a constant rate. A nearest neighbor polymer-polymer repulsive interaction is considered and an external field is applied to drive the chains along the X-direction where a wall is placed opposite to the source end. We consider attractive, neutral, and repulsive interactions beween the wall and the chain nodes. Metropolis algorithm is used to reptate the chains. We investigate the evolution of the density profile of the chains as they deposit on the wall as a function of chain length, field strength, and temperature. The profile of the size of the chains is also examined by studying the evolution of the radius of gyration (R_g) and end-to-end distance (R_e) of the chains. We observe several interesting results such as the onset of oscillation in density profile, variation of the polymer coverage on the wall with chain length, field, temperature, etc.

[L.31] Electrophoretic Flow of Polymer Chains Through a Porous Medium.

A computer simulation model is used to study the flow of polymer chains through a porous medium. We consider a discrete lattice of size L^d with dimension d = 3. A coarse-grained model of polymer chains with SAW constraints is used. Initially, polymer chains are randomly distributed in the lattice and the porous medium is generated by randomly distributing point-like quenched barriers; the empty lattice sites act as solvent background. A nearest neighbour polymer-solvent attractive and polymer-polymer repulsive interaction is considered with a flow field which drives the polymer chains along the field direction. We use kink-jump and reptation dynamics to move the chain nodes with a Metropolis algorithm. We study the conformational and transport properties of polymer chains as a function of the barrier concentration (p_b), temperature (T), chain length (L_c), polymer concentration (p) and bias (B). Many interesting response properties are observed from elongation of chains to their collapse as a function of these parameters. We have examined the variation of the rms displacement of the chains with time and find unusual power-law dependence in certain range of these parameters. Using a linear response theory attempts are made to evaluate an effective permeability and observe its dependence on p_b, T, and L_c. Strong deviation from the linear response flow is noted in a wide range of parameter space. Some of these findings will be reported.

[L.32] Rotation Model of Diffusion Mobility of Macromolecules Flexible-Chain Crystalline Polymer

The model of the rotation mobility and self-diffusion of linear macromolecules in the crystallits of flexible-chain polymers (type of polyetylene) has been proposed. The interconnection of molecular degrees of freedom has been taken into account as rotation of molecular groups and chain deformation in the intermolecular field. Varying the summary potential energy, the expressions corresponding to the condition of chains equilibrium have been found for the considered model. Thus we have got complex nonhomogeneous equations with mixed terms. However, if one accepts the solution changes slowly enough, one can restrict oneself only to terms including derivative of the lowest order and their products. In this case the solution is analogous to one of Frenkel-Kontorova's equation.

[L.33] Molecular Models of Slippage for Strained Passage Nacromolecules in Oriented Crystalline Polymer

An elementary act of polymer plastic deformation and its creep is slippage of molecular chains. The oriented crystalline polymer with homogeneous chemical structure is studied. Models of the thermoactivated slippages of stressed macromolecules and the relaxation of local loads on amorphous parts of these molecules are suggested. The crystalline polymer is considered to consist of two phases, one with interchanging amorphous and crystalline regions in a microfibrille. Frenkel-Kontorova's soliton model is used for calculations. Polymer crystallites are assumed to be long enough for the continuous matter approximation being applicable. Depending on external load and amorphous region length, two cases are realized. The first case takes place when the load is moderate. In this case the load on the amorphous section of a slipped out chain is completely relaxed and this section can change its conformation state. In the second case slipped out amorphous part of a macromolecule is in strained state but its strain is less than one of the macromolecule before its slipping out. The energy activation dependencies on molecular parameters and the local load are different for the two cases.

[L.34] Computer Simulation of Tension and Slippage of Passage Macromolecules in Crystalline Regions of the Oriented Linear Polyethylene

This report presents the result of computer simulation of the local load distribution along crystalline section of the stretched passage macromolecules. The biphase model of linear oriented crystalline polymer as linear polyethylene with the real sizes of crystalline and amorphous regions was used. This model considers passage macromolecules with stressed section in one amorphous region possessing the free conformational coiled sections in adjacent amorphous region and the molecular chain with cross-link or strained loops (entanglement) on crystallite surfaces. The dislocation (kink) formation is accompanied by mechanical stimulation of molecular chains slippage and local loads relaxation in their amorphous regions, as well as by conformation reorganizations of these sections at some conditions. It is shown that the nature of distribution of the local loads, kink form and it's evolution in different degree depends on longitudinal size of crystallits, load on amorphous sections of macromolecules besides of molecular parameters. The slippage of fixed passage molecular chains of cross-links or strained loops (entanglements) on or near polymer crystallite surface (crystallite boundary) is accompanied by increase of the dislocation energy. For such chains with growth their tension at an increase of external load or at redistribution of local loads in chains slippage processes are created condition for thermofluctation rupture their stressed amorphous sections.

[L.35] Motion of a Coarse-Grained Polymer Chain in a Porous Medium: Effect of Porosity and External Field.

Monte Carlo simulation based on a coarse-grained off-lattice bead-spring model is used to reveal the dynamic of a polymer chain driven by an external biased field in a quenched porous medium. In absence of field, the mean square displacement (msd) of an inner chain bead shows a subdiffusive motion unlike the motion of an ideal polymer chain in melt in short time regime, i.e., different from Rouse and Reptation. We find that the power-law exponent for the msd depends on the barrier concentration C. In presence of field, important information on the way in which chains move between and overcome obstacles is gained from the msd vs time analysis in directions parallel and perpendicular to flow. Thus, instead of a steady approach to uniform drift like motion at low C, at early times we observe oscillations in the effective exponent describing the time dependence of the longitudinal msd, when C and the field intensity are increased. The transversal msd component of the center of mass shows a subdiffusive behavior at short time-regimes, which approaches a drift-like motion at intermediate time scales and slows down again to normal diffusion at longer times. We suggest a model which provides a possible interpretation of our simulational results.

[L.36] Impact Fragmentation of Brittle Materials

A deterministic model of fragmentation of a brittle crystal is proposed. In the model, dynamics of viscoelastic material is taken into account, though inhomogeneity in material and interactions between fragments are totally neglected to reduce computational costs. By numerical simulations for impulsive loadings, we analyze the dynamical processes during fragmentation, and find that the distribution function n(m) of resultant fragment mass m has a power law regime as n(m) \sim m^-\beta, in which \beta is insensitive to material specific parameters and initial conditions. Assuming a finite size scaling form of the mass distribution function, we derive a theoretical expression for the exponent \beta as a function of the spatial dimension and a geometrical parameter. We also find that the expression gives good agreement with our numerical simulations and other experimental observations.

(Figures are linked here.)

[L.37] A New Semiclassical Approach to the Nonlinear Optical Response of Inhomogeneous Fluids

I describe a semiclassical approach to modeling the optical response of atomic clusters and inhomogeneous atomic fluids. This method entails mapping the physical cluster onto a lattice, approximating the lattice by a recursive tree graph, and using the graph's recursion relations to obtain an approximate solution for the system's time evolution on the original lattice. This approach allows for precise representation of the cluster's connectivity and of structural constraints, and can be applied to systems not easily amenable to other approaches. Results obtained from this model for the nonlinear optical response of clusters of sodium and of argon are presented, and generalization of the model to the fully quantal case is discussed.

[L.38] Relaxation on Critical Percolation Clusters and Smart Self Avoiding Random Walks

Large contour lines in a random landscape constitute a continuum percolation problem. We consider directed walks on these lines at the percolation threshold -- self avoiding by construction -- and calculate the density correlation function using a Monte Carlo simulation. It has a scaling structure where all exponents are related to the fractal dimension d_h=7/4 of extended contour lines. The corresponding scaling function however vanishes in the limit ømega\rightarrow 0 with a power law giving rise to another a priory independent exponent \zeta. Our data indicate a Cole-Cole structure for the quantity \chi''(\vec ømega) averaged over extended lines only, which implies \zeta=2/7, \eta=0 and an anomaly in the small frequency asymptotics of the diffusion coefficient D(\vec q,ømega)\proptoømega^-\zeta/q^2. This anomaly is reminiscent to a similar one in the Quantum Hall Effect however with negative \zeta_qhe. We argue that the difference is due to the decay of phase resonances in the latter which come with a broad distribution of decay times.

[L.39] \ Quantum Monte Carlo Simulation of Helium with Random Potential \

We study a two-dimensional lattice model of helium with random potential. We performed grand canonical quatum Monte Carlo simulation for this model and analyzed the super-normal phase transition. We compared our results with the one-dimensional calculations of Hatano. We found that this two-dimensional system is less likely to become Bose glass than the one-dimensional case. The anisotropic case will also be discussed.

[L.40] Path Integral Monte Carlo Simulations of ^3He-^4He Mixtures

The Path Integral Monte Carlo (PIMC) method is an accurate way to calculate finite temperature properties of quantum systems such as superfluid ^4He, normal ^3He and isotopic helium mixtures. In the case of the later fermion systems, a restricted form of PIMC is used in which only paths that have a positive density matrix element are considered. We are investigating isotopic helium mixtures in the spinodal region, in an attempt to calculate the coarse-grained free energy from the microscopic configurations generated by PIMC. This calculated free energy will then be used in a cell dynamic systems (CDS) model of spinodal decomposition of ^3He-^4He mixtures.

[L.41] Calculation of Stark Resonances in Hydrogen

The complex scaling method is used to compute resonance energies for the Stark effect in Hydrogen. We use the fact that this Hamiltonian is separable in semi-parabolic coordinates (\eta,\xi,\varphi) to write the trial wavefunction as a product of functions, \Psi(\eta,\xi)=\left(\sum_i=1^Na_i\phi_i(\eta)\right) \left(\sum_j=1^Nb_i\phi_j(\xi)\right). Such a trial state is equivalent to using a N^2 basis of the form \phi_i(\eta)\phi_j(\xi), which has been used in previous variational calculations of this model. The method in this report optimizes the linear coefficients in the trial wavefunction by solving for the eigenvalues of a N\times N matrix at each step of an iterative procedure. Consequently, considerable computational time is saved by avoiding the need to solve a N^2\times N^2 matrix. Very accurate results are obtained with small matrix diagonalizaions (N=25).

[L.42] Algebraic Methods in Computational Quantum Mechanics

Liouville's equation for the density matrix, \rho, of an~N-level quantum system interacting with radiation has a non-trivial kinematic structure; the quantities~Tr\,\rho^n,~n=1,\dots,N remain constant in time, independent\/ of the Hamiltonian. These invariants are physically significant; the qualitative character of the solution depends on their existence. The N^2-dimensional vector space \cal V on which H and \rho act is spanned by a representation of the Lie algebra u(N). Dynamical symmetries are manifest in a Hamiltonian living in a subspace of \cal V which is spanned by a subalgebra of u(N). This gives rise to additional constants of motion that are likewise physically significant. A generic numerical method, by its very nature, will not in general preserve these invariants (either kinematic or dynamic). We present a numerical technique, ``Unitary Integration,'' that exactly preserves these invariants to all orders in the time step. Although the evolution is approximate, the structure of the Liouville equation (including that due to dynamical symmetry) is preserved exactly since the time advance map is constructed so as to yield a unitary transformation. We present algorithms for constructing such integrators and an automated procedure for detecting dynamical symmetries as well as the corresponding invariants. We provide examples, comment on applications, and discuss various numerical issues.

[L.43] An Energy Curve Family of Semi-Infinite Lattice and the Transition Point in the Two-Dimensional Potts Model.

We have studied the behavior of energy curves of semi-infinite (l\times \infty ) lattice in the two-dimensional Potts model with q states (2\le q\le 7). The transfer matrix is constructed by Kramers and Wannier's screw method, and the largest eigenvalue is calculated by iteration on a supercomputer. Let \left( K_0 \right)_l be the coupling parameter K=J \mathord\left/ \vphantom J kT \right. \kern-\nulldelimiterspace kT at the intersection point between energy versus K curves for size l and l+1 lattices. We numerically find that \left( K_0 \right)_2= \left( K_0 \right)_3=\cdots =K_C, where K_C=\ln \left( 1+\sqrt q \right) \mathord\left/ \vphantom \ln \left( 1+\sqrt q \right) 2 \right. \kern-\nulldelimiterspace 2 is the critical coupling parameter for the bulk lattice obtained from the duality argument by Potts. Our numerical figures \left( K_0 \right)_l coincide with the exact K_C to fourteen decimal places, which may obviously results from the fact that the dual transformation of the transfer matrix leaves the eigenvalues invariant even for the smallest lattice size. The duality relation which determines K_C, holds for small l lattices in two dimensions. The procedure of intersection of energy curves could be used to predict an unknown critical temperature for some statistical models which may have the duality property. This is not the case in the three-dimensional Potts model. A similar calculation in three dimensions suggests an uniformly decreasing sequence \left\ \left( K_0 \right)_l \right\ with the lower bound K_C ( =\mathop \lim \limits_l\to \infty \left( K_0 \right)_l ). We also discuss the specific heat exponent from the energy curve family.

[L.44] Nanoscale Sm_0.25Zr_0.75Fe_3 Produced By Mechanical Alloying.

The particle size distribution in mechanically produced Sm_.25Zr_.75Fe_3 is calculated from line broadening in x-ray diffraction, and magnetometry measurement. The results are compared with Mössbauer measurements.

[L.45] Modelling Solid State Laser Resonators

Recent Developments in the optics industry and in microprocessors performance enable design and modelling of complicated laser systems on desk computers. Realistic simulations of solid state lasers must take into account material parameters, resonator gemetry as well as heat sources and sinks. The optical output depends on non-linear effects,such as strong refractive index gradients, gain guiding,and facets deformation. In order to model these effects, we developed a software package for platforms that run under Windows 95/NT. The package includes three tools: 1. Cad tool with an easy-to-use graphical user interface especially designed to allow intuitive build-up of laser systems from given optical elements. This tool uses fast ABCD matrix algorithm in order to show wave front shape and optical intensity of the laser modes. 2. Finite elements analysis to compute temperature and stress distribution, as well as deformation of the laser crystal. 3. Finite difference beam propagation analysis to compute and display the resonator eigenvalues and eigenfunctions. Design and analysis of a complicated system requires to combine all three tools in an iterative analysis. In the following paper we present results of a typical analysis performed with these tools.

[L.46] Higer Order Generalized Upwinding for Simulation of Hydrodynam= ic Semiconductor Equations

A generalized upwinding method has been developed for the solution of the hydrodynamic semiconductor equations. The developed method is applied to simulation of Si submicron devices. Poisson's equation is solved together with the hydrodynamic equations to provide the space-varying electrical field. This generalization is based upon the flux vector splitting (FVS) methods developed in the area of Computational Aerodynamics. This methods essentially split the hyperbolic portion of the governing equations into two wave components. In one dimension, the compenents include right or left running waves. After splitting the equation into its components, a first order upwind difference is applied to each wave component. Upon careful examination of the results for electrons in the submicron devices, spatial first order accuracy is insufficient due to highly nonlinear source terms in the hydrodynamic equations. Therefore, a stability analysis is performed to determine the limits of this method. When the higher order upwinding is used the limits of the stability significantly increase. Third order upwinding shows no signs of instability and yield very favorable results.

[L.47] Adaptive Mesh Refinement Techniques for Plasma Based Semiconductor Processing Simulation

Many of the processing steps used in semiconductor manufacturing involve plasmas. One of the major difficulties associated with computational models of manufacturing equipment is the large scale variation between features of interest. Typical plasma ``reactors'' have dimensions of tens of centimeters while plasma features such as sheaths are hundreds of microns in size.

We have addressed this challenge by employing conservative finite difference techniques with an Adaptive Mesh Refinement (AMR) strategy. AMR is a block structured, Cartesian grid method which can be used to efficiently solve discretized partial differential equations by allowing for variable temporal and spatial resolutions within a single computational domain. In this paper we describe a time-splitting scheme for the evolution of a set of plasma fluid equations. The method uses high order Godunov advection techniques as well as multigrid methods for linear systems. The extension of the single grid algorithm to a locally refined grid is then discussed. The efficiency and accuracy of the method are examined using computational results. Finally, the algorithm is used to address issues related to reactor design. More information about this research is available on the DAS-LLNL website.

[L.48] Long Range Magnetic Order versus Singlet Formation in Correlated Electron Systems

The impurity Anderson model describes the cross--over which can occur between a state with a well--defined moment and a singlet state, when a single local spin is coupled to the itinerant electrons of a conduction band. This phenomenon also occurs in lattice models where a genuine phase transition can develop between an antiferromagnet with a set of ordered f--electron moments, and a disordered phase. We describe Quantum Monte Carlo (QMC) simulations of a number of tight--binding Hamiltonians which exhibit this effect-- coupled Hubbard planes, and the two-- and three--dimensional Anderson lattice models.

[L.49] Diffusion Monte Carlo Method on Curved Manifolds

We present a stochastic approach to solving the many-body Schrödinger equation on curved manifolds with general metric. The method is based on the Diffusion Monte Carlo (DMC) technique, modified to include `quantum corrections' into the propagator which appear due to the curvature. As an illustration of our method we apply it to the quantum Hall effect, using the spherical geometry introduced by Haldane^1. In this geometry electrons are confined to the surface of a sphere with a magnetic monopole at the center, and therefore are effectively in a curved space. To avoid the `sign problem' we follow the `fixed-phase DMC' approach of Ortiz et al.^2 and map this fermionic problem into an effective bosonic problem using a singular gauge transformation. As a result of our calculations we obtain energies for different types of quasiparticles and compare them with experiment.

^1F.D.M. Haldane, Phys. Rev. Lett. \bf51 605 (1983) ^2G. Ortiz et al., Phys. Rev. Lett. \bf17 2777 (1993)

[L.50] Antiferromagnetism in cuprates from the three--band model

The strongly correlated three--band model for high--T_c compounds at half--filling is studied using a functional integral approach. We first focus on the Antiferromagnetic Hartree--Fock approximation . A saddle--point development around this mean--field gives an effective spin model generalizing the spin 1/2 Heisenberg model. We find that both copper--oxygen covalency and spin fluctuations play a significant role in the magnetic long--range order and considerably influence the magnitude of magnetic moments on copper sites and the effective exchange constants even in the very strongly correlated limit. Using parameters from previous LDA electronic structure calculations, we apply our results and find good agreement both with other theoretical approaches such as cluster methods or Monte--Carlo and with experimental values of staggered magnetization, superexchange, and the optical gap, for a large set of actual cuprates.

[L.51] Numerical dynamics of the transient response of Josephson-coupled multilayers

Josephson-coupled multilayers are multilayers of superconducting and insulating materials with Josephson coupling between the layers. These structures have very promising applications as active switching devices in ultra high speed superconducting microelectronic circuitry, as well as high frequency oscillators. An essential key to their applied use is to understand their switching properties, or the transient response. A novel model has been developed which treats the charge on the layer interface planes of the multilayer as a dynamic variable, whose evolution is det ermined via the interlayer charge-current equations. With the time-domain current responses of both the superconducting and insulating layers included, a fully consistent set of coupled integro-differential equations has been obtained that describe the time evolution of the layer charge s and voltages, allowing a proper assessment of the switching dynamics of the multilayer as a whole. In this paper I describe the numerical treatment of this novel model and the dynamics which result.

[L.52]

This abstract was not submitted electronically.

[L.53] Polarons by Nonlocal Dynamical Coherent Potential Method

The nonlocal dynamical coherent potential approximation (NDCPA) is formulated to calculate a single\--electron(exciton) Green's function of polaron due to the interaction of an electron(exciton) with phonons with dispersion. This approximation is an extension of the well known dynamical CPA (H.Sumi, J.Phys.Soc.Jpn.) 36, 770(1974).. The NDCPA provides an efficient means of calculating of an approximate Green's function for a dynamical model of electrons(excitons) strongly coupled to dispersive phonons, in the entire ranges of the electron(exciton)\--phonon coupling strengths end electron(exciton) transfer. The electron(exciton)\--phonon coupling in the Hamiltonian may involve terms of any order with respect to the phonon operators. A set of recurrent equations is derived in the case of a system at finite temperatures, from which the coherent potential as a function of energy E and impulse vector k can be obtained. The numerical scheme provided by NDCPA has been applied to calculate the exciton absorption and emission spectra in the wide ranges of the excitonic bandwidth, the coupling constant, and the temperature.

[L.54] Macroscopic Quantum Phase-Locking Model for the Quantum Hall = Effect

A macroscopic model of nonlinear dissipative phase-locking between a Josephson-like frequency and a macroscopic electron wave frequency is proposed to explain the Quantum Hall Effect. It is well known that a r.f-biased Josephson junction displays a collective phase-locking behavior which can be described by a non-autonomous second order equation or an equivalent 2+1-dimensional dynamical system. Making a direct analogy between the QHE and the Josephson system, this report proposes a computer-solving nonlinear dynamical model for the quantization of the Hall resistance. In this model, the Hall voltage is assumed to be proportional to a Josephson-like frequency and the Hall current is assumed related to a coherent electron wave frequency. The Hall resistance is shown to be quantized in units of the fine structure constant as the ratio of these two frequencies are locked into a rational winding number. To explain the sample-width dependence of the critical current, the 2DEG under large applied current is further assumed to develop a Josephson-like junction array in which all Josephson-like frequencies are synchronized. Other remarkable features of the QHE such as the resistance fluctuation and the even-denominator states are also discussed within this picture.

[L.55] HIGH TEMPERATURE MODELING OF AGING EFFECTS IN POLYMERS OBSERVED BY THERMAL STIMULATED CURRENTS.

The temperature region comprising the glass transition temperature in polymers has been the subject of numerous studies. The problem of characterizing such temperature region is complicated by the existence of conduction phenomena together with the glass transition relaxation peak. To describe aging experiments in Poly(DTH succinate) a simple equivalent circuit model is developed. The glass transition relaxation peak is described by a four parameter mode(M. Puma. PAT 8, 39 (1997).). The conduction part of the spectra or the peak associated with it, when blocking electrodes are used, is modeled by a circuit analogy. The results of these models accurately describe the observed data and allow one to reach several important conclusions regarding the mechanisms underlying the changes in the dipolar reorientation as well as the conduction through the sample. The thermal history previous to the annealing periods is similar for all experiments and the heating and cooling rates used are also similar; therefore the glass transition temperature, T_g, should not change. The model presented agrees with the above consideration and within experimental errors shows a unique glass transition temperature. The shift of the maximum of the current observed is attributed to the charge redistribution peak that arises due to the presence of blocking electrodes.

[L.56] Defect Production and Microstructure Evolution in Irradiated Metals

The property changes that ensue when a material is subject to irradiation by energetic particles is a complex function of the interaction between the defect supersaturation induced by the irradiation and the microstructure. To describe these processes, a computational approach capable of spanning 14 orders of magnitude in time scale and 6 orders of magnitude in length scale is required. We describe an approach that couples molecular dynamics (MD) simulations of defect production and point and extended defect energetics and kinetics to a kinetic Monte Carlo description of defect diffusion and microstructure evolution. We present results for defect evolution in alpha-Fe and Au. The MD simulations describe damage production in displacement cascades. We show that in all metals a significant fraction of the intersitials produced are in small clusters and dislocation loops. However, this fraction is significantly larger in Au than in Fe. Using MD simulations, we show that small interstitial clusters form prismatic dislocation loops, and that these can glide in one dimension with a very low activation energy comparable to that for the single self interstitial. These results are then coupled to a kinetic Monte Carlo description of long length and time scale defect diffusion and evolution. We show that the fraction of defects that are able to escape recombination in their nascent cascade and migrate long distances depends on the nature of the interstitial dislocation loops produced in the cascade and changes with recoil energy and lattice temperature.

[L.57] Single particle properties of the 2-D Hubbard model

Single particle excitations of the hole-doped two dimensional repulsive Hubbard model have intensively been discussed since it is commonly accepted that the model describes essential features of the electronic states of a single CuO_2 layer of metallic cuprates. Insight into strong coupling features of the model has been obtained by quantum Monte-Carlo calculations and by exact diagonalizations. However, the shape of the Fermi surface and details of the dispersion for the metallic region are difficult to obtain in this way. Here we use standard perturbation theory in order to investigate spectral properties of the model over the entire temperature range for values of the interaction parameter U which lead to strong coupling features. Since the strong coupling regime sets in for relatively small values of U, information about strong coupling effects can be obtained in second order perturbation theory. The corresponding Feynman diagram of the irreducible self-energy has been calculated numerically. We obtain qualitative agreement with results of quantum Monte Carlo simulations.

[L.58] From nonorthogonal electron group functions to generalized pseudopotencials

The method of Pseudopotencials, which provides drastic reduction of computational work in studies of electronic structure, is well founded for one-determinantal approach. A generalization of pseudopotentials to internally correlated electron systems is shown to be possible on the basis of the Group Function (GF) method by McWeeny. However, this method itself is well developed only for strongly orthogonal GFs that sometimes presents a restriction of variational freedom of GFs. In the present work, we develop further the GF method formalism for nonorthogonal GFs using a previously proposed operator technique, which makes it possible to present results in a general compact form for any number m of electron groups in an N-electron system and for any distribution of the electrons between groups without making any further approximation. A general and compact expression for the total energy of the system is obtained in terms of reduced density matrices of electron groups. Self-consistent equations for nonorthogonal GFs are also derived. Construction of generalized pseudopotentials is discussed.

[L.59] First Principles Calculation of Intial Oxidation Processes of Si(001) Surface

Oxidation of the Si(001) surface, one of the most important processes in the modern Si technology, was studied using the first-principles calculation technique with spin-polarized gradient approximation. The spontaneous dissociative chemisorption occurs only in the exceptional case, where incident O_2 molecule attacks the center between the two surface dimers with its molecular axis parallel to a dimer row. In other cases, rather large activation energy is needed for dissociative chemisorption of an O_2 molecule. It increases as the chemisorption site becomes deeper from the surface. This is in accordance qualitatively with the recent finding by scanning reflection electron microscopy (SREM) that the oxidation occurs in a layer-by-layer reaction mode. The calculated activation energies, however, are considerably larger than the observed ones. In particular, the calculated barrier energy for backbond oxidation is more than 0.8 eV, while the first sub-surface, including backbonds, is observed to be oxidized with almost no activation energy. We have found that the barrierless backbond oxidation can be realized along a reaction path via the spontaneously chemisorbed meta-stable states on the Si top-layer mentioned earlier. Since this channel is very narrow, our model also explains why the initial sticking coefficient of O_2 molecule is small despite of the fact that the barrierless oxidation takes place along this path. This work was partly supported by the New Energy and Industrial Technology Development Organization (NEDO).

[L.60] Ab Initio Hartree-Fock Simulations of Silver Film Adhesion on Both alpha - Al2O3 (0001) and MgO (100) Surfaces

The atomic and electronic structures of Ag/\alpha -Al_2O_3 (0001) and Ag/MgO(100) interfaces are calculated using an ab~initio Hartree-Fock computer code CRYSTAL with the DFT correlation for silver monolayer atop some layers of perfect oxide lattices. The band structure, electron density distribution and densities of states are analyzed in detail for isolated and interacting slabs of silver, \alpha -Al_2O_3 and MgO. The energetically most favourable adsorption positions for Ag atoms are found to be above the centers of oxygen triangles on O-terminated \alpha -Al_2O_3(0001) surface as well as external oxygen atoms of MgO(100) surface.

The main peculiarity of silver adsorption on \alpha -Al_2O_3(0001) consists in substitution of external aluminium atoms of stable Al-terminated (0001) corundum surface by silver atoms Mulliken population analysis shows that neither appreciable charge transfer in the interfacial region, nor considerable population of bonds between silver atoms and insulating substrate takes place. The adhesion energy arises mainly due to the electrostatic interaction of substrate atoms with a complicated charge redistribution in the metal film, characterized by large quadrupole moments and electron density redistribution towards the gap position between the nearest Ag atoms.

[L.61]

This abstract was not submitted electronically.

[L.62] Cluster Calculations and Spectroscopic Study of the Electronic Structure of the Zirconolite-Type Compounds Ln_2Ti_2O_7 (Ln=Rare Earth Elements)

Theoretical and experimental investigations of the electronic structure have been carried out for the series of ternary oxides isostructural to the mineral pyrochlore. These compounds are very interesting as reference materials for the well known basic synroc mineral zirconolite CaZrTi_2O_7, which is one of the most effective means of immobilizing the high-level nuclear wastes.

Taking into account a very complex crystal structure (88 atoms per unit cell) of the investigated materials, the X\alpha(SW) cluster method has been chosen. A new approach for the pyrochlores spatial structure description, that was developed in this work, allowed to make a correct choise of the cluster for the calculation procedure. Model clusters reflect all main peculiarities of the real structure for each compound.

The atomic charge states, the charge transfer processes and the partial states distribution in the valence bands have been studied on the base of the computational results and the X-ray emission/photoelectron spectroscopic measurements. Theoretical results are in a good agreement with experimental ones.

[L.63]

This abstract was not submitted electronically.

[L.64] A Materials Science Study: Molecular-Dynamics Simulations of Crack Propagation and Martensitic Transformation

Molecular-dynamics simulations are a powerful tool in order to investigate atomistic processes in solids and liquids. With increasing computational power it has become possible to simulate large numbers of atoms which is necessary to describe defects of the solid state. In order to treat interatomic forces, many different potentials have been proposed. We have performed molecular-dynamics simulations up to 600.000 atoms on the basis of Lenard-Jones and embedded-atom potentials. In particular we have examined crack propagation of micro-crystals with notch-like defects. This allows to study the transition from brittle to ductile fracture as a function of temperature and potential. This change of fracture behavior is comparable with experimental observations of crack propagation in metals and nonmetals. Furthermore we have studied the martensitic phase transition in magnetic iron-nickel alloys. The strong concentration dependence of the structural transition temperature is very well reproduced by using embedded atom potentials for the interatomic forces. We have also investigated concomitant effects of martensitic transformations like shape-memory effects and superelasticity. The consequence of superelasticity is that austenitic crystals under strain do not show fracture behavior but undergo martensitic transformation in order to avoid the strain.

[L.65] Molecular Dynamics Study of Hydrogen in a Nb (100) slab

We have carried out a series of Molecular Dynamics simulations of the behavior of H in a Nb slab. The Nb is simulated by the well studied n-body potential of Finnis and Sinclair (M.W., Finnis, and J.E., Sinclair, Phil. Mag. A50, 45, (1984)). Some of Nb surface properties with this potential will be presented like a check for using this potential. For the Nb-H interaction, instead of using a purely repulsive potential like the ones used for bulk simulations (M.J., Gillan, Phys. Rev. Lett., 58, 313 (1987)), we have implemented a Lennard-Jones (L-J) with two parameters. The dependence of the results with respect to their values will be shown. As expected, the purely repulsive potentials, used in bulk, at low temperature tend to expel H from the BCC(100) slab (very short time H desorption), contrary to experiment. On the other hand, the L-J type potential gives rise to a stable H-Nb configurations. With these, at low temperatures, the H is trapped at tetrahedral sites in agreement with experimental findings. At higher temperatures, the H diffuses with time constants of the same order of magnitude as measured experimentally. A quantitative analysis of this behavior will be shown and the binding energy dependence will be discussed.

[L.66] Defects in hybrid nematic films: a Monte Carlo Simulation

Monte-Carlo simulations of a nematic liquid crystal film placed between two different media that set antagonistic (planar and homeotropic) orientation of the ``molecules" at the top and the bottom surfaces are presented. This system is a model of the HAND film [1] which, as observed in experimental studies, presents the formations of topological defects induced by the different alignement at the surface. Our simulation model is based on a Lebwohl-Lasher cubic lattice [2,3] where the molecules are represented by three dimensional unit vectors and interact only with their nearest neighbors. The lateral surface of the sample are left free, i.e. no boundary conditions are specified and it is found that the ground state of the system contains stable topological defects when the lateral radius of the system is larger than its thickness.

[1] O.D. Lavrentovich and V.M. Pergamnschik, Int. J. Mod. Phys. B 9, 2839 (1995).

[2] P.A. Lebwohl and G. Lasher, Phys. Rev. A 6 , 426 (1972).

[3] U. Fabbri and C. Zannoni, Mol. Phys., 58, 763 (1986).

[L.67] Application of Simulated Annealing Method to Hydrogen-bond Lattices

We have studied hydrogen-bond lattices, as a model of 5-bromo-9-hydroxyphenalenone, of which crystal shows phase transition from paraelectric phase to antiferroelectric phase through incommensurate phase for deuterated compounds. We represent the hydrogen bonds by double well potential. We assume that wave function of ground state is a linear combination of two Gaussian functions and that density operator at finite temperature consists of the ground state and the first excited state, which is a linear combination of same Gaussian functions but is orthogonalized to the ground state. In order to get most stable states, we have used simulated annealing method and minimize the ground state energy for T=0 case and the free energy at finite temperatures, which are functions of center positions, widths and coefficients of Gaussian functions (and distributions of ground and first excited states in case of the free energy). Results of numerical simulations for body centered tetragonal lattices with periodic boundary conditions show that when depth of double well potential V_0 is small, most stable state for deuteron is incommensurate phase, and when V_0 is large, most stable state is antiferroelectric phase.

[L.68] Monte Carlo Approaches to the Double-Exchange Systems on Cubic Lattices

Perovskite Mn oxides AMnO_3 is now intensively studied due to its colossal magnetoresistance phenomena. As a canonical model for these compounds, the double-exchange model (ferromagnetic Kondo lattice model) on a cubic lattice is investigated. In the infinite dimensional limit, this model reproduces many experimental results for (La,Sr)MnO_3, such as magnetoresistance curve, Curie temperature and optical conductivity spectrum [1]. In order to justify these results in three dimension, we calculate the model on a cubic lattice numerically. Monte Carlo calculation is performed by treating localized t_2g spin as classical one. Magnetic properties as well as electronic states are calculated. Through comparison we find that results on cubic lattice and in infinite-dimensional results are qualitatively similar.

\vspace3mm\par References: \relax [1] N. Furukawa: J. Phys. Soc. Jpn. 63 (1994) 3214, ibid 64 (1995) 2754, ibid 64 (1995) 2734, ibid 64 (1995) 3164.

[L.69] Perpendicular Order in Frustrated Magnetic Layers

We report results of Monte Carlo simulations of coupled two dimensional, square lattice, classical, antiferromagnetic Heisenberg systems. We find that when the two layers are offset so that frustrating interactions are present, the magnetic order in the planes can be made orthogonal, even in the absence of any explicit symmetry breaking terms on the Hamiltonian. The presence or absence of a perpendicular ordering tendency depends on the precise nature of the inter-plane offset. Connections are made to the magnetic behavior of recently sythesized pnictide--oxide materials.

[L.70] Monte Carlo Methods in Transition Metal Alloys

Giant moments of several Bohr magnetons are formed in transition metal alloys where the matrix is palladium or platinum. The interaction between these giant moments produces a phase transition from paramagnetism to ferromagnetism, when the alloy is below the Curie temperature. These giant moments can be measured mainly by neutron diffraction, although several characteristics can be determined by magnetization measurements. In this work, several magnetic properties of these alloys are presented, based on calculations made mainly by Monte Carlo simulation of these properties. A localized moment model is used to simulate the formation of magnetization clouds and thier transformation as the temperature is raised. The simulation allows the calculation of the critical temperatures of ferromagnetism, which are then compared with experimental measurements. In several of these alloys, unpolarized diffuse neutron scattering measurements show large forward peaks that would indicate giant moments larger than those obtained by magnetization measurements. We calculated, using Monte Carlo simulation, the diffuse neutron cross sections for these alloys and reproduced the neutron data. We find a significant quasielastic contribution to the scattering that can not be attributed to the magnetization cloud. The calculation methods were applied to several dilute and concentrated transition metal alloys. The results indicate that the methods and models used are valid for a large group of Pd and Pt based alloys.

[L.71] Monte Carlo study of the Widom-Rowlinson fluid using cluster methods.

The Widom-Rowlinson model of a fluid mixture is studied using a new cluster algorithm that is an adaptation of the invaded cluster method previously applied to Potts models. The algorithm overcomes the difficulties of treating continuum hard-core systems and has almost no critical slowing down at the demixing transition. Our estimates of \beta/\nu and \gamma/\nu for the two-component fluid are consistent with the Ising universality class in two and three dimensions. We also present preliminary results for the three-component fluid.

[L.72] Simulation of Optical Gradient Traps via the Vector Finite Element Method

It is well known that when light is incident upon a dielectric object there is a net force on the object due to the radiation pressure of the light. For tightly focused beams incident upon a microscopic particle the force may be greater than the competing gravitational or viscous forces, hence the particle is trapped by the beam and the particle can be manipulated by the beam. The recently developed Vector Finite Element Method, which solves Maxwell's equations on three dimensional unstructured grids, is used to compute optical gradient forces on arbitrarily shaped objects. Computed results are shown to agree with measured data for micron-sized dielectric spheres. In addition the method is used to compute the energy absorbed by the object and the stress within the object, which is useful in predicting damage when optical gradient traps are used to manipulate living biological cells.

[L.73] Hybrid N--body techniques using SCF

We present implementation of hybrid N--body schemes for self--gravitating systems using the SCF algorithm. Applications to astrophysical systems and parallelisation of the algorithm is discussed.

[L.74] Linear Scaling Solution of the Coulomb problem using wavelets

The Coulomb problem for continous charge distributions is a central problem in physics. Powerful methods, that scale linearly with system size and that allow us to use different resolutions in different regions of space are therefore highly desireable. Using wavelet based Multi Resolution Analysis we derive for the first time a method which has these properties. The power and accuracy of the method is illustrated by applying it to the calculation of the electrostatic potential of a fully three-dimensional all-electron U_2 dimer.

[L.75] Simulation of Gravitational Radiation Antennas

The large complexity associated with the design of spherical resonant mass gravitational radiation antennas, necessitates the use of reliable simulation tools, that can accurately predict consequences of specific design choices.

For this purpose, we have implemented three different methods; an analytical solution for the displacement field in spherical homogeneous objects, a method utilizing a polynomial expansion of this same displacement field, which is suitable for more complex shaped objects, and a finite element simulation. It is shown that the latter method is best suited to answer critical design issues, whereas the first two can serve as a calibration for the finite element simulation.

Since the computational complexity of the finite element kernel requires significant computing power, it has been tailored for execution on parallel computer systems.

The simulation program is used to model the influences on the frequency spectrum of the antenna, resulting from varying important design parameters, like the size, material, suspension system and read-out system. Furthermore, we simulate the antenna's response to seismic and thermal noise, cosmics and gravitational radiation.

[L.76] The Tecolote Framework for Multi-Material Hydrodynamics

Tecolote is an object-oriented framework for multi-material hydrodynamics simulations in complex geometries. This framework is providing the basis for one of the main Accelerated Strategic Computing Initiative (ASCI) code development efforts at Los Alamos National Laboratory. Tecolote has been designed to greatly simplify the task of integrating various physics packages into a single simulation code and to allow for interchange of computational algorithms as desired. Tecolote provides a straightforward mechanism for introducing new physics modules via specific features such as a fixed main routine and its DataDirectory system. Transparent parallelism and crucial optimizations such as load balancing and compressed storage for sparse fields are provided through use of the Parallel Object-Oriented Methods and Applications (POOMA) C++ class library. The object-oriented design features of Tecolote and its use of specific POOMA capabilities will be presented. In addition, comparisons will be drawn with a previous non-object-oriented hydrodynamics code written in C.

[L.77] Object Oriented Component Framework for Experimental Data Analysis in Nuclear Physics

Object-oriented data analysis system was implemented under Oberon System environment. The system is designed for efficient analysis of data from multidetector Nuclear Physics experiments. Other application areas, such as spectroscopy, are also possible. T Work supported by the U.S. DOE, grant number DE-FG02-88ER40414.

[L.78] The Embedded Curved Boundary Method for Orthogonal Simulation Meshes.

Embedded Curved Boundaries (ECB) is a new method for handling interfaces on orthogonal simulation meshes. ECB represents boundaries with piecewise linear approximations embedded in the orthogonal grid. This allows accurate modelling of complex structures even on a relatively coarse grid, using simple, parallelizable algorithms such as Dynamic ADI. The technique has been extended to provide simultaneous solutions in adjacent domains with proper treatment of the interfaces, on both uniform and non-uniform grids. A form of adaptive mesh refinement is used to resolve sub-grid structures or to provide greater detail in a given region. The algorithms and some applications are presented, along with current work on extending the method to three dimensions.

[L.79] The Generalized Balanced Ternary (GBT) High-Performance Computational ''Tool--Kit''

The Generalized Balanced Ternary (GBT) coordinate system has been proposed in the past for applications of n-dimensional scientific computations and scientific visualization. It has now been applied to high-performance calculations on the CM-5 machine at the Army's High Performance Computing Research Center with the development of ``tool kits'' in both CM FORTRAN and FORTRAN 77. GBT uses pseudo-integer calculations with an efficiency exceeding floating point methods. GBT has certain topological advantages not exhibited by the normal ``orthogonal'' coordinates, e.g. `x', `y', and `z'. The GBT Toolkit, which provides for the manipulation of GBT addresses, will be described. All of the tools will scale from a one to 30 dimensional vector space. This restriction in scalability is due to the fact that GBT addresses are limited, under CM Fortran, to INTEGER*4 values. The toolkit contains a set of primitive functions to add, subtract and multiply single address values, a set of higher level functions which call the primitive functions to add, subtract, and multiply multiple address values, and a set of functions to display GBT addresses in a form which is readable by humans. The potential applications of GBT to scientific computations (e.g. n--dimensional spaces---for Radiation Transport, Hydrodynamics, etc.) has been described in the past with the projection back into 2--D and 3--D screen displays.

[L.80] Computational/HPC Physics Education

The Physics group in NACSE (an NSF Metacenter Regional Alliance) has developed a variety of materials to be used in computational physics education and to assist working scientists and engineers. Our emphasis is to exploit Web technology to better teach about and improve the use of HPC resources in physics. We will demonstrate multimedia, interactive Web tutorials (http://nacphy.physics.orst.edu/ (Wiley, 1997). Also demonstrated will be tutorials to assist physicists with visualizations, HPC library use, PVM, and, in particular, Coping with Unix, an Interactive Survival Kit for Scientists. These latter tutorials use some special Web technology (Webterm) we developed which makes it possible to connect to a remote Unix machine and follow the lessons from any Web browser supporting Java --- even browsers on non-Unix computers such as PCs or Macs.

[L.81] Solving Bethe-Salpeter Equation Using a Separable Effective Gluon Propagator

The coupled Dyson-Schwinger Equation and Bethe-Salpeter Equation approach provides a powerful mean for low energy nuclear physics. However, solving the Bethe-Salpeter Equations for bound states of mesons is notoriously difficult. We propose an effective gluon propagator in a separable ansatz, which is determined by the quark propagators via Dyson-Schwinger Equation. The quark propagators are constructed to reproduce pion and kaon observables. With this separable gluon propagator, the Bethe-Salpeter Equation is reduced to a matrix equation and can be easily solved for meson bound state masses and their Bethe-Salpeter amplitudes. A mass spectrum of light-quark mesons is obtained with this approach and the masses of vector and axial mesons impressively agree well with the experimental data. We also apply the Bethe-Salpeter amplitudes for the vector mesons so obtained to the study of vector meson Nucleon coupling and the results agree well with the empirical values.

Part L of program listing