Gaussian processes are used as emulators for expensive computer simulations. Recently, Gaussian processes have also been used to model the "error field" or "code discrepancy" between a computer simulation code and experimental data, and the delta term between two levels of computer simulation (multi-fidelity codes). This work presents the use of Gaussian process models to approximate error or delta fields, and examines how one calculates the parameters governing the process. In multi-fidelity modeling, the delta term is used to correct a lower fidelity model to match or approximate a higher fidelity model. The terms governing the Gaussian process (e.g., the parameters of the covariance matrix) are updated using a Bayesian approach. We have found that use of Gaussian process models requires a good understanding of the method itself and an understanding of the problem in enough detail to identify reasonable covariance parameters. The methods are not "black-box" methods that can be used without some statistical understanding. However, Gaussian processes offer the ability to account for uncertainties in prediction. This approach can help reduce the number of high-fidelity function evaluations necessary in multi-fidelity optimization.
We define a new diagnostic method where computationally-intensive numerical solutions are used as an integral part of making difficult, non-contact, nanometer-scale measurements. The limited scope of this report comprises most of a due diligence investigation into implementing the new diagnostic for measuring dynamic operation of Sandia's RF Ohmic Switch. Our results are all positive, providing insight into how this switch deforms during normal operation. Future work should contribute important measurements on a variety of operating MEMS devices, with insights that are complimentary to those from measurements made using interferometry and laser Doppler methods. More generally, the work opens up a broad front of possibility where exploiting massive high-performance computers enable new measurements.
The purpose of this project is to develop tools to model and simulate the processes of self-assembly and growth in biological systems from the molecular to the continuum length scales. The model biological system chosen for the study is the tendon fiber which is composed mainly of Type I collagen fibrils. The macroscopic processes of self-assembly and growth at the fiber scale arise from microscopic processes at the fibrillar and molecular length scales. At these nano-scopic length scales, we employed molecular modeling and simulation method to characterize the mechanical behavior and stability of the collagen triple helix and the collagen fibril. To obtain the physical parameters governing mass transport in the tendon fiber we performed direct numerical simulations of fluid flow and solute transport through an idealized fibrillar microstructure. At the continuum scale, we developed a mixture theory approach for modeling the coupled processes of mechanical deformation, transport, and species inter-conversion involved in growth. In the mixture theory approach, the microstructure of the tissue is represented by the species concentration and transport and material parameters, obtained from fibril and molecular scale calculations, while the mechanical deformation, transport, and growth processes are governed by balance laws and constitutive relations developed within a thermodynamically consistent framework.
Field programmable gate arrays (FPGAs) have been used as alternative computational de-vices for over a decade; however, they have not been used for traditional scientific com-puting due to their perceived lack of floating-point performance. In recent years, there hasbeen a surge of interest in alternatives to traditional microprocessors for high performancecomputing. Sandia National Labs began two projects to determine whether FPGAs wouldbe a suitable alternative to microprocessors for high performance scientific computing and,if so, how they should be integrated into the system. We present results that indicate thatFPGAs could have a significant impact on future systems. FPGAs have thepotentialtohave order of magnitude levels of performance wins on several key algorithms; however,there are serious questions as to whether the system integration challenge can be met. Fur-thermore, there remain challenges in FPGA programming and system level reliability whenusing FPGA devices.4 AcknowledgmentArun Rodrigues provided valuable support and assistance in the use of the Structural Sim-ulation Toolkit within an FPGA context. Curtis Janssen and Steve Plimpton provided valu-able insights into the workings of two Sandia applications (MPQC and LAMMPS, respec-tively).5
As George W. Bush recognized in November 2001, "Infectious diseases make no distinctions among people and recognize no borders." By their very nature, infectious diseases of natural or intentional (bioterrorist) origins are capable of threatening regional health systems and economies. The best mechanism for minimizing the spread and impact of infectious disease is rapid disease detection and diagnosis. For rapid diagnosis to occur, infectious substances (IS) must be transported very quickly to appropriate laboratories, sometimes located across the world. Shipment of IS is problematic since many carriers, concerned about leaking packages, refuse to ship this material. The current packaging does not have any ability to neutralize or kill leaking IS. The technology described here was developed by Sandia National Laboratories to provide a fail-safe packaging system for shipment of IS that will increase the likelihood that critical material can be shipped to appropriate laboratories following a bioterrorism event or the outbreak of an infectious disease. This safe and secure packaging method contains a novel decontaminating material that will kill or neutralize any leaking infectious organisms; this feature will decrease the risk associated with shipping IS, making transport more efficient. 3 DRAFT4
We have enhanced our parallel molecular dynamics (MD) simulation software LAMMPS (Large-scale Atomic/Molecular Massively Parallel Simulator, lammps.sandia.gov) to include many new features for accelerated simulation including articulated rigid body dynamics via coupling to the Rensselaer Polytechnic Institute code POEMS (Parallelizable Open-source Efficient Multibody Software). We use new features of the LAMMPS software package to investigate rhodopsin photoisomerization, and water model surface tension and capillary waves at the vapor-liquid interface. Finally, we motivate the recipes of MD for practitioners and researchers in numerical analysis and computational mechanics.