Publications

Abstract not provided.

Abstract not provided.

Abstract not provided.

This paper presents automated expert modeling for automated student evaluation, or AEMASE (pronounced "amaze"). This technique grades students by comparing their actions to a model of expert behavior. The expert model is constructed with machine learning techniques, avoiding the costly and time-consuming process of manual knowledge elicitation and expert system implementation. A brief summary of after action review (AAR) and intelligent tutoring systems (ITS) provides background for a prototype AAR application with a learning expert model. A validation experiment confirms that the prototype accurately grades student behavior on a tactical aircraft maneuver application. Finally, several topics for further research are proposed. © Springer-Verlag Berlin Heidelberg 2006.

Abstract not provided.

Many problems in science and engineering require the solution of a long sequence of slowly changing linear systems. We propose and analyze two methods that significantly reduce the total number of matrix-vector products required to solve all systems. We consider the general case where both the matrix and right-hand side change, and we make no assumptions regarding the change in the right-hand sides. Furthermore, we consider general nonsingular matrices, and we do not assume that all matrices are pairwise close or that the sequence of matrices converges to a particular matrix. Our methods work well under these general assumptions, and hence form a significant advancement with respect to related work in this area. We can reduce the cost of solving subsequent systems in the sequence by recycling selected subspaces generated for previous systems. We consider two approaches that allow for the continuous improvement of the recycled subspace at low cost. We consider both Hermitian and non-Hermitian problems, and we analyze our algorithms both theoretically and numerically to illustrate the effects of subspace recycling. We also demonstrate the effectiveness of our algorithms for a range of applications from computational mechanics, materials science, and computational physics. © 2006 Society for Industrial and Applied Mathematics.

We discuss computing first derivatives for models based on elements, such as large-scale finite-element PDE discretizations, implemented in the C++ programming language. We use a hybrid technique of automatic differentiation (AD) and manual assembly, with local element-level derivatives computed via AD and manually summed into the global derivative. C++ templating and operator overloading work well for both forward- and reverse-mode derivative computations. We found that AD derivative computations compared favorably in time to finite differencing for a scalable finite-element discretization of a convection-diffusion problem in two dimensions. © Springer-Verlag Berlin Heidelberg 2006.

Abstract not provided.

Surfpack is a general-purpose software library of multidimensional function approximation methods for applications such as data visualization, data mining, sensitivity analysis, uncertainty quantification, and numerical optimization. Surfpack is primarily intended for use on sparse, irregularly-spaced, n-dimensional data sets where classical function approximation methods are not applicable. Surfpack is under development at Sandia National Laboratories, with a public release of Surfpack version 1.0 in August 2006. This paper provides an overview of Surfpack's function approximation methods along with some of its software design attributes. In addition, this paper provides some simple examples to illustrate the utility of Surfpack for data trend analysis, data visualization, and optimization. Copyright © 2006 by the American Institute of Aeronautics and Astronautics, Inc.

In evaluating new high-speed network interfaces, the usual metrics of latency and bandwidth are commonly measured and reported. There are numerous other message passing characteristics that can have a dramatic effect on application performance that should be analyzed when evaluating a new interconnect. One such metric is overhead, which dictates the networks ability to allow the application to perform non-message passing work while a transfer is taking place. A method for measuring overhead, and hence calculating application availability, is presented. Results for several next-generation network interfaces are also presented. © Springer-Verlag Berlin Heidelberg 2006.

Surfpack is a general-purpose software library of multidimensional function approximation methods for applications such as data visualization, data mining, sensitivity analysis, uncertainty quantification, and numerical optimization. Surfpack is primarily intended for use on sparse, irregularly-spaced, n-dimensional data sets where classical function approximation methods are not applicable. Surfpack is under development at Sandia National Laboratories, with a public release of Surfpack version 1.0 in August 2006. This paper provides an overview of Surfpack's function approximation methods along with some of its software design attributes. In addition, this paper provides some simple examples to illustrate the utility of Surfpack for data trend analysis, data visualization, and optimization. Copyright © 2006 by the American Institute of Aeronautics and Astronautics, Inc.

Constitutive models for chemically reacting networks are formulated based on a generalization of the independent network hypothesis. These models account for the coupling between chemical reaction and strain histories, and have been tested by comparison with microscopic molecular dynamics simulations. An essential feature of these models is the introduction of stress transfer functions that describe the interdependence between crosslinks formed and broken at various strains. Efforts are underway to implement these constitutive models into the finite element code Adagio. Preliminary results are shown that illustrate the effects of changing crosslinking and scission rates and history.

Chemical/Biological/Radiological (CBR) contamination events pose a considerable threat to our nation's infrastructure, especially in large internal facilities, external flows, and water distribution systems. Because physical security can only be enforced to a limited degree, deployment of early warning systems is being considered. However to achieve reliable and efficient functionality, several complex questions must be answered: (1) where should sensors be placed, (2) how can sparse sensor information be efficiently used to determine the location of the original intrusion, (3) what are the model and data uncertainties, (4) how should these uncertainties be handled, and (5) how can our algorithms and forward simulations be sufficiently improved to achieve real time performance? This report presents the results of a three year algorithmic and application development to support the identification, mitigation, and risk assessment of CBR contamination events. The main thrust of this investigation was to develop (1) computationally efficient algorithms for strategically placing sensors, (2) identification process of contamination events by using sparse observations, (3) characterization of uncertainty through developing accurate demands forecasts and through investigating uncertain simulation model parameters, (4) risk assessment capabilities, and (5) reduced order modeling methods. The development effort was focused on water distribution systems, large internal facilities, and outdoor areas.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

Abstract not provided.

The physics of ballistic penetration mechanics is of great interest in penetrator and counter-measure design. The phenomenology associated with these events can be quite complex and a significant number of studies have been conducted ranging from purely experimental to 'engineering' models based on empirical and/or analytical descriptions to fully-coupled penetrator/target, thermo-mechanical numerical simulations. Until recently, however, there appears to be a paucity of numerical studies considering 'non-ideal' impacts [1]. The goal of this work is to demonstrate the SHISM algorithm implemented in the ALEGRA Multi-Material ALE (Arbitrary Lagrangian Eulerian) code [13]. The SHISM algorithm models the three-dimensional continuum solid mechanics response of the target and penetrator in a fully coupled manner. This capability allows for the study of 'non-ideal' impacts (e.g. pitch, yaw and/or obliquity of the target/penetrator pair). In this work predictions using the SHISM algorithm are compared to previously published experimental results for selected ideal and non-ideal impacts of metal penetrator-target pairs. These results show good agreement between predicted and measured maximum depth-of-penetration, DOP, for ogive-nose penetrators with striking velocities in the 0.5 to 1.5 km/s range. Ideal impact simulations demonstrate convergence in predicted DOP for the velocity range considered. A theory is advanced to explain disagreement between predicted and measured DOP at higher striking velocities. This theory postulates uncertainties in angle-of-attack for the observed discrepancies. It is noted that material models and associated parameters used here, were unmodified from those in the literature. Hence, no tuning of models was performed to match experimental data. Copyright © 2005 by ASME.

Kevlar materials make excellent body armor due to their fabric-like flexibility and ultra-high tensile strength. Carbon composites are made up from many layers of carbon AS-4 material impregnated with epoxy. Fiber orientation is bidirectional, orientated at 0° and 90°. They also have ultra-high tensile strength but can be made into relatively hard armor pieces. Once many layers are cut and assembled they can be ergonomicically shaped in a mold during the heated curing process. Kevlar and carbon composites can be used together to produce light and effective body armor. This paper will focus on computer analysis and laboratory testing of a Kevlar/carbon composite cross-section proposed for body armor development. The carbon composite is inserted between layers of Kevlar. The computer analysis was performed with a Lagrangian transversely Isotropic material model for both the Kevlar and Carbon Composite. The computer code employed is AUTODYN. Both the computer analysis and laboratory testing utilized different fragments sizes of hardened steel impacting on the armor cross-section. The steel fragments are right-circular cylinders. Laboratory testing was undertaken by firing various sizes of hardened steel fragments at square test coupons of Kevlar layers and heat cured carbon composites. The V50 velocity for the various fragment sizes was determined from the testing. This V50 data can be used to compare the body armor design with other previously designed armor systems. AUTODYN [1] computer simulations of the fragment impacts were compared to the experimental results and used to evaluate and guide the overall design process. This paper will include the detailed transversely isotropic computer simulations of the Kevlar/carbon composite cross-section as well as the experimental results and a comparison between the two. Conclusions will be drawn about the design process and the validity of current computer modeling methods for Kevlar and carbon composites. Copyright © 2005 by ASME.

Processing-in-Memory (PIM) technology encompasses a range of research leveraging a tight coupling of memory and processing. The most unique features of the technology are extremely wide paths to memory, extremely low memory latency, and wide functional units. Many PIM researchers are also exploring extremely fine-grained multi-threading capabilities. This paper explores a mechanism for leveraging these features of PIM technology to enhance commodity architectures in a seemingly mundane way: accelerating MPI. Modern network interfaces leverage simple processors to offload portions of the MPI semantics, particularly the management of posted receive and unexpected message queues. Without adding cost or increasing clock frequency, using PIMs in the network interface can enhance performance. The results are a significant decrease in latency and increase in small message bandwidth, particularly when long queues are present.

The critical Rayleigh number Racr of the Hopf bifurcation that signals the limit of steady flows in a differentially heated 8:1:1 cavity is computed. The two-dimensional analog of this problem was the subject of a comprehensive set of benchmark calculations that included the estimation of Racr [1]. In this work we begin to answer the question of whether the 2D results carry over into 3D models. For the case of the 2D model being extruded for a depth of 1, and no-slip/no-penetration and adiabatic boundary conditions placed at these walls, the steady flow and destabilizing eigenvectors qualitatively match those from the 2D model. A mesh resolution study extending to a 20-million unknown model shows that the presence of these walls delays the first critical Rayleigh number from 3.06 × 105 to 5.13 × 105. © 2005 Elsevier Ltd.