Linear algebra is a powerful and proven tool in web search. Techniques, such as the PageRank algorithm of Brin and Page and the HITS algorithm of Kleinberg, score web pages based on the principal eigenvector (or singular vector) of a particular non-negative matrix that captures the hyperlink structure of the web graph. We propose and test a new methodology that uses multilinear algebra to elicit more information from a higher-order representation of the hyperlink graph. We start by labeling the edges in our graph with the anchor text of the hyperlinks so that the associated linear algebra representation is a sparse, three-way tensor. The first two dimensions of the tensor represent the web pages while the third dimension adds the anchor text. We then use the rank-1 factors of a multilinear PARAFAC tensor decomposition, which are akin to singular vectors of the SVD, to automatically identify topics in the collection along with the associated authoritative web pages.
Controlling the distribution of chemical constituents within complex, structurally heterogeneous systems represents one of the fundamental challenges of alloy design. We demonstrate how the combination of recent developments in sophisticated experimental high resolution characterization techniques and ab initio theoretical methods provide the basis for a detailed level of understanding of the microscopic factors governing compositional distributions in metallic alloys. In a study of the partitioning of Mg in two-phase ternary Al-Sc-Mg alloys by atom-probe tomography, we identify a large Mg concentration enhancement at the coherent {alpha}-Al/Al{sub 3}Sc heterophase interface with a relative Gibbsian interfacial excess of Mg with respect to Al and Sc, {Lambda}{sub Mg}{sup rel}, equal to 1.9 {+-} 0.5 atom nm{sup -2}. The corresponding calculated value of {Lambda}{sub Mg}{sup rel} is -1.2 atom nm{sup -2}. Theoretical ab initio investigations establish an equilibrium driving force for Mg interfacial segregation that is primarily chemical in nature and reflects the strength of the Mg-Sc interactions in an Al-rich alloy.
Effective monitoring of large computational clusters demands the analysis of a vast amount of raw data from a large number of machines. The fundamental interactions of the system are not, however, well-defined, making it difficult to draw meaningful conclusions from this data, even if one were able to efficiently handle and process it. In this paper we show that computational clusters, because they are comprised of a large number of identical machines, behave in a statistically meaningful fashion. We therefore can employ normal statistical methods to derive information about individual systems and their environment and to detect problems sooner than with traditional mechanisms. We discuss design details necessary to use these methods on a large system in a timely and low-impact fashion.
Electrochromic (EC) materials are used in 'smart' windows that can be darkened by applying a voltage across an EC stack on the window. The associated change in refractive index (n) in the EC materials might allow their use in tunable or temperature-insensitive Fabry-Perot filters and transmissive-spatial-light-modulators (SLMs). The authors are conducting a preliminary evaluation of these materials in many applications, including target-in-the-loop systems. Data on tungsten oxide, WO{sub 3}, the workhorse EC material, indicate that it's possible to achieve modest changes in n with only slight increases in absorption between the visible and {approx}10 {micro}m. This might enable construction of a tunable Fabry-Perot filter consisting of an active EC layer (e.g. WO{sub 3}) and a proton conductor (e.g.Ta{sub 2}O{sub 5}) sandwiched between two gold electrodes. A SLM might be produced by replacing the gold with a transparent conductor (e.g. ITO). This SLM would allow broad-band operation like a micromirror array. Since it's a transmission element, simple optical designs like those in liquid-crystal systems would be possible. Our team has fabricated EC stacks and characterized their switching speed and optical properties (n, k). We plan to study the interplay between process parameters, film properties, and performance characteristics associated with the FP-filter and then extend what we learn to SLMs. Our goals are to understand whether the changes in absorption associated with changes in n are acceptable, and whether it's possible to design an EC-stack that's fast enough to be interesting. We'll present our preliminary findings regarding the potential viability of EC materials for target-in-the-loop applications.
There are many important biological processes involving lipid bilayers on times scales beyond that accessible by atomistic simulations. We have developed coarse-grained, bead-spring models of lipid molecules to treat membrane fusion, domain formation and the general physical characteristics of lipid bilayers. A key aspect of these coarse-grained models is that the liquid nature of a bilayer is explicitly present in the simulations; the lipids diffuse far beyond their neighbors in contrast to atomistic simulations. With these models self-assembly into a bilayer starting from a random configuration of lipids and solvent is readily simulated. We have performed extensive simulations to characterize these lipid models in single component lipid bilayers. For a variety of tail lengths, the area per lipid as a function of temperature has been calculated; the liquid-gel transition has been characterized. Models have been developed for a variety of lipids including double bonds in the lipid tails. Simulation results will be presented for fusion and domain formation.
The latency and throughput of MPI messages are critically important to a range of parallel scientific applications. In many modern networks, both of these performance characteristics are largely driven by the performance of a processor on the network interface. Because of the semantics of MPI, this embedded processor is forced to traverse a linked list of posted receives each time a message is received. As this list grows long, the latency of message reception grows and the throughput of MPI messages decreases. This paper presents a novel hardware feature to handle list management functions on a network interface. By moving functions such as list insertion, list traversal, and list deletion to the hardware unit, latencies are decreased by up to 20% in the zero length queue case with dramatic improvements in the presence of long queues. Similarly, the throughput is increased by up to 10% in the zero length queue case and by nearly 100% in the presence queues of 30 messages.
The conclusions of this report are: (1) 1D and 2D RMHD simulations indicate feasibility of producing high thermonuclear neutron yields in deuterium and DT gas-puff Z-pinches -- (a) Z 1.7 x 10{sup 13} DD neutrons at 70 kV, 13 MA (Z1384); (b) (3 to 6) x 10{sup 14} at 90 kV, 17 MA (Z1422); (c) Predicted for ZR 2 x 10{sup 15} DD and 6 x 10{sup 16} DT neutrons; (2) Theory and modeling issues -- collisionless ions, nonthermal ions; (3) Experimental data on the origin of the neutrons not yet conclusive, need more shots; and (4) Applications -- (a) Fusion 2.5 and 14 MeV neutron source; (b) Pulsed subcritical neutron source with uranium blanket for {approx}10x neutron and {approx}1000x energy multiplication (Smirnov, Feoktistov and Klimov); and (c) Fusion-assisted keV x-ray plasma radiation source.
New experimental diagnostics and computational modeling provide an unprecedented means for improving the understanding of energetic material behavior at the mesoscale (grain or crystal ensemble levels). This study focuses on the determination of appropriate constitutive and EOS property data of the constituents of an energetic composite at high stress and moderate strain-rate states. The Sandia Z accelerator is used to determine the mechanical response of energetic composites via isentropic ramp wave compression loading. In this paper we describe an energy source method in CTH that models ramp loading for the analysis of ICE experiments. This approach is applied to design experimental configurations to probe the constituent response of PBX 9501 subjected to {approx}40 Kbar ramp load over 300 ns duration. Multiple VISAR are used to determine the averaged response of the composite material in comparison to the individual constituents including the effects of anisotropy of HMX crystals and the interactions of fine crystallites with binder material.
Mechanical dynamics can be a determining factor for the switching speed of radio-frequency microelectromechanical systems (RF MEMS) switches. This paper presents the simulation of the mechanical motion of a microswitch under actuation. The switch has a plate suspended by springs. When an electrostatic actuation is applied, the plate moves toward the substrate and closes the switch. Simulations are calculated via a high-fidelity finite element model that couples solid dynamics with electrostatic actuation. It incorporates non-linear coupled dynamics and accommodates fabrication variations. Experimental modal analysis gives results in the frequency domain that verifies the natural frequencies and mode shapes predicted by the model. An effective 1D model is created and used to calculate an actuation voltage waveform that minimizes switch velocity at closure. In the experiment, the switch is actuated with this actuation voltage, and the displacements of the switch at various points are measured using a laser Doppler velocimeter through a microscope. The experiments are repeated on several switches from different batches. The experimental results verify the model.