LOCA, the Library of Continuation Algorithms, is a software library for performing stability analysis of large-scale applications. LOCA enables the tracking of solution branches as a function of a system parameter, the direct tracking of bifurcation points, and, when linked with the ARPACK library, a linear stability analysis capability. It is designed to be easy to implement around codes that already use Newton's method to converge to steady-state solutions. The algorithms are chosen to work for large problems, such as those that arise from discretizations of partial differential equations, and to run on distributed memory parallel machines. This manual presents LOCA's continuation and bifurcation analysis algorithms, and instructions on how to implement LOCA with an application code. The LOCA code is being made publicly available at www.cs.sandia.gov/loca.
The final report for a Laboratory Directed Research and Development project entitled, Molecular Simulation of Reacting Systems is presented. It describes efforts to incorporate chemical reaction events into the LAMMPS massively parallel molecular dynamics code. This was accomplished using a scheme in which several classes of reactions are allowed to occur in a probabilistic fashion at specified times during the MD simulation. Three classes of reaction were implemented: addition, chain transfer and scission. A fully parallel implementation was achieved using a checkerboarding scheme, which avoids conflicts due to reactions occurring on neighboring processors. The observed chemical evolution is independent of the number of processors used. The code was applied to two test applications: irreversible linear polymerization and thermal degradation chemistry.
This report describes research and development of the large eddy simulation (LES) turbulence modeling approach conducted as part of Sandia's laboratory directed research and development (LDRD) program. The emphasis of the work described here has been toward developing the capability to perform accurate and computationally affordable LES calculations of engineering problems using unstructured-grid codes, in wall-bounded geometries and for problems with coupled physics. Specific contributions documented here include (1) the implementation and testing of LES models in Sandia codes, including tests of a new conserved scalar--laminar flamelet SGS combustion model that does not assume statistical independence between the mixture fraction and the scalar dissipation rate, (2) the development and testing of statistical analysis and visualization utility software developed for Exodus II unstructured grid LES, and (3) the development and testing of a novel new LES near-wall subgrid model based on the one-dimensional Turbulence (ODT) model.
The theory, numerical algorithm, and user documentation are provided for a new ''Centroidal Voronoi Tessellation (CVT)'' method of filling a region of space (2D or 3D) with particles at any desired particle density. ''Clumping'' is entirely avoided and the boundary is optimally resolved. This particle placement capability is needed for any so-called ''mesh-free'' method in which physical fields are discretized via arbitrary-connectivity discrete points. CVT exploits efficient statistical methods to avoid expensive generation of Voronoi diagrams. Nevertheless, if a CVT particle's Voronoi cell were to be explicitly computed, then it would have a centroid that coincides with the particle itself and a minimized rotational moment. The CVT code provides each particle's volume and centroid, and also the rotational moment matrix needed to approximate a particle by an ellipsoid (instead of a simple sphere). DIATOM region specification is supported.
The theory and algorithm for the Material Point Method (MPM) are documented, with a detailed discussion on the treatments of boundary conditions and shock wave problems. A step-by-step solution scheme is written based on direct inspection of the two-dimensional MPM code currently used at the University of Missouri-Columbia (which is, in turn, a legacy of the University of New Mexico code). To test the completeness of the solution scheme and to demonstrate certain features of the MPM, a one-dimensional MPM code is programmed to solve one-dimensional wave and impact problems, with both linear elasticity and elastoplasticity models. The advantages and disadvantages of the MPM are investigated as compared with competing mesh-free methods. Based on the current work, future research directions are discussed to better simulate complex physical problems such as impact/contact, localization, crack propagation, penetration, perforation, fragmentation, and interactions among different material phases. In particular, the potential use of a boundary layer to enforce the traction boundary conditions is discussed within the framework of the MPM.
The Computational Plant or Cplant is a commodity-based supercomputer under development at Sandia National Laboratories. This paper describes resource-allocation strategies to achieve processor locality for parallel jobs in Cplant and other supercomputers. Users of Cplant and other Sandia supercomputers submit parallel jobs to a job queue. When a job is scheduled to run, it is assigned to a set of processors. To obtain maximum throughput, jobs should be allocated to localized clusters of processors to minimize communication costs and to avoid bandwidth contention caused by overlapping jobs. This paper introduces new allocation strategies and performance metrics based on space-filling curves and one dimensional allocation strategies. These algorithms are general and simple. Preliminary simulations and Cplant experiments indicate that both space-filling curves and one-dimensional packing improve processor locality compared to the sorted free list strategy previously used on Cplant. These new allocation strategies are implemented in the new release of the Cplant System Software, Version 2.0, phased into the Cplant systems at Sandia by May 2002.
Collaboration between Sandia National Laboratories and the University of New Mexico Biology Department resulted in the capability to train students in microarray techniques and the interpretation of data from microarray experiments. These studies provide for a better understanding of the role of stationary phase and the gene regulation involved in exit from stationary phase, which may eventually have important clinical implications. Importantly, this research trained numerous students and is the basis for three new Ph.D. projects.
Simulations of a turbulent methanol pool fire are conducted using both Reynolds-Averaged Navier-Stokes (RANS) and Large Eddy Simulation (LES) modeling methodologies. Two simple conserved scalar flameletbased combustion models with assumed PDF are developed and implemented. The first model assumes statistical independence between mixture fraction and its variance and results in poor predictions of time-averaged temperature and velocity. The second combustion model makes use of the PDF transport equation for mixture fraction and does not employ the statistical independence assumption. Results using this model show good agreement with experimental data for both the 2D and 3D LES, indicating that the use of statistical independence between mixture fraction and its dissipation is not valid for pool fire simulations. Lastly, "finger-like" flow structures near the base of the plume, generated from stream-wise vorticity, are shown to be important mixing mechanisms for accurate prediction of time-averaged temperature and velocity.
The concept of genetic divisors can be given a quantitative measure with a non-Archimedean p-adic metric that is both computationally convenient and physically motivated. For two particles possessing distinct mass parameters x and y, the metric distance D(x, y) is expressed on the field of rational numbers Q as the inverse of the greatest common divisor [gcd (x , y)]. As a measure of genetic similarity, this metric can be applied to (1) the mass numbers of particle states and (2) the corresponding subgroup orders of these systems. The use of the Bezout identity in the form of a congruence for the expression of the gcd (x , y) corresponding to the v{sub e} and {sub {mu}} neutrinos (a) connects the genetic divisor concept to the cosmic seesaw congruence, (b) provides support for the {delta}-conjecture concerning the subgroup structure of particle states, and (c) quantitatively strengthens the interlocking relationships joining the values of the prospectively derived (i) electron neutrino (v{sub e}) mass (0.808 meV), (ii) muon neutrino (v{sub {mu}}) mass (27.68 meV), and (iii) unified strong-electroweak coupling constant ({alpha}*{sup -1} = 34.26).
In exploring the question of how humans reason in ambiguous situations or in the absence of complete information, we stumbled onto a body of knowledge that addresses issues beyond the original scope of our effort. We have begun to understand the importance that philosophy, in particular the work of C. S. Peirce, plays in developing models of human cognition and of information theory in general. We have a foundation that can serve as a basis for further studies in cognition and decision making. Peircean philosophy provides a foundation for understanding human reasoning and capturing behavioral characteristics of decision makers due to cultural, physiological, and psychological effects. The present paper describes this philosophical approach to understanding the underpinnings of human reasoning. We present the work of C. S. Peirce, and define sets of fundamental reasoning behavior that would be captured in the mathematical constructs of these newer technologies and would be able to interact in an agent type framework. Further, we propose the adoption of a hybrid reasoning model based on his work for future computational representations or emulations of human cognition.
We demonstrate two specific examples that show how our exiting capabilities in solving large systems of partial differential equations associated with transport/reaction systems can be easily applied to outstanding problems in computational biology. First, we examine a three-dimensional model for calcium wave propagation in a Xenopus Laevis frog egg and verify that a proposed model for the distribution of calcium release sites agrees with experimental results as a function of both space and time. Next, we create a model of the neuron's terminus based on experimental observations and show that the sodium-calcium exchanger is not the route of sodium's modulation of neurotransmitter release. These state-of-the-art simulations were performed on massively parallel platforms and required almost no modification of existing Sandia codes.
Icarus is a 2D Direct Simulation Monte Carlo (DSMC) code which has been optimized for the parallel computing environment. The code is based on the DSMC method of Bird[11.1] and models from free-molecular to continuum flowfields in either cartesian (x, y) or axisymmetric (z, r) coordinates. Computational particles, representing a given number of molecules or atoms, are tracked as they have collisions with other particles or surfaces. Multiple species, internal energy modes (rotation and vibration), chemistry, and ion transport are modeled. A new trace species methodology for collisions and chemistry is used to obtain statistics for small species concentrations. Gas phase chemistry is modeled using steric factors derived from Arrhenius reaction rates or in a manner similar to continuum modeling. Surface chemistry is modeled with surface reaction probabilities; an optional site density, energy dependent, coverage model is included. Electrons are modeled by either a local charge neutrality assumption or as discrete simulational particles. Ion chemistry is modeled with electron impact chemistry rates and charge exchange reactions. Coulomb collision cross-sections are used instead of Variable Hard Sphere values for ion-ion interactions. The electro-static fields can either be: externally input, a Langmuir-Tonks model or from a Green's Function (Boundary Element) based Poison Solver. Icarus has been used for subsonic to hypersonic, chemically reacting, and plasma flows. The Icarus software package includes the grid generation, parallel processor decomposition, post-processing, and restart software. The commercial graphics package, Tecplot, is used for graphics display. All of the software packages are written in standard Fortran.
An effort is underway at Sandia National Laboratories to develop a library of algorithms to search for potential interactions between surfaces represented by analytic and discretized topological entities. This effort is also developing algorithms to determine forces due to these interactions for transient dynamics applications. This document describes the Application Programming Interface (API) for the ACME (Algorithms for Contact in a Multiphysics Environment) library.
This report documents an exploratory FY 00 LDRD project that sought to demonstrate the first steps toward a realistic computational representation of the variability encountered in individual human behavior. Realism, as conceptualized in this project, required that the human representation address the underlying psychological, cultural, physiological, and environmental stressors. The present report outlines the researchers' approach to representing cognitive, cultural, and physiological variability of an individual in an ambiguous situation while faced with a high-consequence decision that would greatly impact subsequent events. The present project was framed around a sensor-shooter scenario as a soldier interacts with an unexpected target (two young Iraqi girls). A software model of the ''Sensor Shooter'' scenario from Desert Storm was developed in which the framework consisted of a computational instantiation of Recognition Primed Decision Making in the context of a Naturalistic Decision Making model [1]. Recognition Primed Decision Making was augmented with an underlying foundation based on our current understanding of human neurophysiology and its relationship to human cognitive processes. While the Gulf War scenario that constitutes the framework for the Sensor Shooter prototype is highly specific, the human decision architecture and the subsequent simulation are applicable to other problems similar in concept, intensity, and degree of uncertainty. The goal was to provide initial steps toward a computational representation of human variability in cultural, cognitive, and physiological state in order to attain a better understanding of the full depth of human decision-making processes in the context of ambiguity, novelty, and heightened arousal.
In order to achieve higher rendering performance, the use of parallel sort-last architecture on a PC cluster is presented. The sort-last library (libpglc) can be linked to an existing parallel application to achieve high rendering rates. The efficient use of 64 commodity graphics cards enables to establish pace-setting rendering performance of 300 million triangles per second on extremely large data.
As a participating national lab in the inter-institutional effort to resolve performance issues of the non-elutable ion exchange technology for Cs extraction, they have carried out a series of characterization studies of UOP IONSIV{reg_sign} IE-911 and its component parts. IE-911 is a bound form (zirconium hydroxide-binder) of crystalline silicotitanate (CST) ion exchanger. The crystalline silicotitanate removes Cs from solutions by selective ion exchange. The performance issues of primary concern are: (1) excessive Nb leaching and subsequent precipitation of column-plugging Nb-oxide material, and (2) precipitation of aluminosilicate on IE-911 pellet surfaces, which may be initiated by dissolution of Si from the IE-911, thus creating a supersaturated solution with respect to silica. In this work, they have identified and characterized Si- and Nb-oxide based impurity phases in IE-911, which are the most likely sources of leachable Si and Nb, respectively. Furthermore, they have determined the criteria and mechanism for removal from IE-911 of the Nb-based impurity phase that is responsible for the Nb-oxide column plugging incidents.
The construction of inverse states in a finite field F{sub P{sub P{alpha}}} enables the organization of the mass scale by associating particle states with residue class designations. With the assumption of perfect flatness ({Omega}total = 1.0), this approach leads to the derivation of a cosmic seesaw congruence which unifies the concepts of space and mass. The law of quadratic reciprocity profoundly constrains the subgroup structure of the multiplicative group of units F{sub P{sub {alpha}}}* defined by the field. Four specific outcomes of this organization are (1) a reduction in the computational complexity of the mass state distribution by a factor of {approximately}10{sup 30}, (2) the extension of the genetic divisor concept to the classification of subgroup orders, (3) the derivation of a simple numerical test for any prospective mass number based on the order of the integer, and (4) the identification of direct biological analogies to taxonomy and regulatory networks characteristic of cellular metabolism, tumor suppression, immunology, and evolution. It is generally concluded that the organizing principle legislated by the alliance of quadratic reciprocity with the cosmic seesaw creates a universal optimized structure that functions in the regulation of a broad range of complex phenomena.
Arithmetic conditions relating particle masses can be defined on the basis of (1) the supersymmetric conservation of congruence and (2) the observed characteristics of particle reactions and stabilities. Stated in the form of common divisors, these relations can be interpreted as expressions of genetic elements that represent specific particle characteristics. In order to illustrate this concept, it is shown that the pion triplet ({pi}{sup {+-}}, {pi}{sup 0}) can be associated with the existence of a greatest common divisor d{sub 0{+-}} in a way that can account for both the highly similar physical properties of these particles and the observed {pi}{sup {+-}}/{pi}{sup 0} mass splitting. These results support the conclusion that a corresponding statement holds generally for all particle multiplets. Classification of the respective physical states is achieved by assignment of the common divisors to residue classes in a finite field F{sub P{sub {alpha}}} and the existence of the multiplicative group of units F{sub P{sub {alpha}}} enables the corresponding mass parameters to be associated with a rich subgroup structure. The existence of inverse states in F{sub P{sub {alpha}}} allows relationships connecting particle mass values to be conveniently expressed in a form in which the genetic divisor structure is prominent. An example is given in which the masses of two neutral mesons (K{degree} {r_arrow} {pi}{degree}) are related to the properties of the electron (e), a charged lepton. Physically, since this relationship reflects the cascade decay K{degree} {r_arrow} {pi}{degree} + {pi}{degree}/{pi}{degree} {r_arrow} e{sup +} + e{sup {minus}}, in which a neutral kaon is converted into four charged leptons, it enables the genetic divisor concept, through the intrinsic algebraic structure of the field, to provide a theoretical basis for the conservation of both electric charge and lepton number. It is further shown that the fundamental source of supersymmetry can be expressed in terms of hierarchical relationships between odd and even order subgroups of F{sub P{sub {alpha}}}, an outcome that automatically reflects itself in the phenomenon of fermion/boson pairing of individual particle systems. Accordingly, supersymmetry is best represented as a group rather than a particle property. The status of the Higgs subgroup of order 4 is singular; it is isolated from the hierarchical pattern and communicates globally to the mass scale through the seesaw congruence by (1) fusing the concepts of mass and space and (2) specifying the generators of the physical masses.
This technical report presents the initial proposal and renewable proposals for an LDRD project whose intended goal was to enable applications to take full advantage of the hardware available on Sandia's current and future massively parallel supercomputers by analyzing various ways of combining distributed-memory and shared-memory programming models. Despite Sandia's enormous success with distributed-memory parallel machines and the message-passing programming model, clusters of shared-memory processors appeared to be the massively parallel architecture of the future at the time this project was proposed. They had hoped to analyze various hybrid programming models for their effectiveness and characterize the types of application to which each model was well-suited. The report presents the initial research proposal and subsequent continuation proposals that highlight the proposed work and summarize the accomplishments.
