Sandia National Laboratories

About: Sandia National Laboratories is a facility organization based out in Livermore, California, United States. It is known for research contribution in the topics: Laser & Thin film. The organization has 21501 authors who have published 46724 publications receiving 1484388 citations. The organization is also known as: SNL & Sandia National Labs.

What Supercomputers Say: A Study of Five System Logs

Abstract: If we hope to automatically detect and diagnose failures in large-scale computer systems, we must study real deployed systems and the data they generate. Progress has been hampered by the inaccessibility of empirical data. This paper addresses that dearth by examining system logs from five supercomputers, with the aim of providing useful insight and direction for future research into the use of such logs. We present details about the systems, methods of log collection, and how alerts were identified; propose a simpler and more effective filtering algorithm; and define operational context to encompass the crucial information that we found to be currently missing from most logs. The machines we consider (and the number of processors) are: Blue Gene/L (131072), Red Storm (10880), Thunderbird (9024), Spirit (1028), and Liberty (512). This is the first study of raw system logs from multiple supercomputers.

Two-dimensional phase unwrapping with minimum weighted discontinuity

Abstract: Given an interferometric phase image of a surface profile, the task of two-dimensional phase unwrapping is to reconstruct the profile by adding multiples of 2π to the image. Discontinuities in the unwrapped phase must be restricted to areas of noise and true discontinuity in the profile. Such areas can often be identified by their low quality. This suggests that the unwrapped phase should be chosen to minimize a weighted sum of discontinuity magnitudes. An algorithm is presented that computes such an unwrapped phase from any initial guess. The elementary operation of the algorithm is to partition the image into two connected regions, then raise the unwrapped phase by 2π in one of the regions, reducing the weighted sum; this is done repeatedly until no suitable partitions exist. The operations are found by creating paths that follow discontinuity curves and extending them to form complete partitions. The algorithm terminates when no path can be extended. The behavior of the algorithm and the benefits of weighting are illustrated with an example.

Origin of the Invar effect in iron–nickel alloys

Abstract: In 1897 Guillaume1 discovered that face-centred cubic alloys of iron and nickel with a nickel concentration of around 35 atomic per cent exhibit anomalously low (almost zero) thermal expansion over a wide temperature range. This effect, known as the Invar effect, has since been found in various ordered and random alloys and even in amorphous materials2. Other physical properties of Invar systems, such as atomic volume, elastic modulus, heat capacity, magnetization and Curie (or Neel) temperature, also show anomalous behaviour. Invar alloys are used in instrumentation, for example as hair springs in watches. It has long been realized that the effect is related to magnetism2,3; but a full understanding is still lacking. Here we present ab initio calculations of the volume dependences of magnetic and thermodynamic properties for the most typical Invar system, a random face-centred cubic iron–nickel alloy, in which we allow for non-collinear spin alignments—that is, spins that may be canted with respect to the average magnetization direction. We find that the magnetic structure is characterized, even at zero temperature, by a continuous transition from the ferromagnetic state at high volumes to a disordered non-collinear configuration at low volumes. There is an additional, comparable contribution to the net magnetization from the changes in the amplitudes of the local magnetic moments. The non-collinearity gives rise to an anomalous volume dependence of the binding energy, and explains other peculiarities of Invar systems.

Operation of terahertz quantum-cascade lasers at 164 K in pulsed mode and at 117 K in continuous-wave mode.

Abstract: We report the demonstration of a terahertz quantum-cascade laser that operates up to 164 K in pulsed mode and 117 K in continuous-wave mode at approximately 3.0 THz. The active region was based on a resonant-phonon depopulation scheme and a metal-metal waveguide was used for modal confinement. Copper to copper thermocompression wafer bonding was used to fabricate the waveguide, which displayed improved thermal properties compared to a previous indium-gold bonding method.

Graph partitioning models for parallel computing

Abstract: Calculations can naturally be described as graphs in which vertices represent computation and edges reflect data dependencies. By partitioning the vertices of a graph, the calculation can be divided among processors of a parallel computer. However, the standard methodology for graph partitioning minimizes the wrong metric and lacks expressibility. We survey several recently proposed alternatives and discuss their relative merits.