Sandia Report

Improving performance via mini-applications.

Application performance is determined by a combination of many choices: hardware platform, runtime environment, languages and compilers used, algorithm choice and implementation, and more. In this complicated environment, we find that the use of mini-applications - small self-contained proxies for real applications - is an excellent approach for rapidly exploring the parameter space of all these choices. Furthermore, use of mini-applications enriches the interaction between application, library and computer system developers by providing explicit functioning software and concrete performance results that lead to detailed, focused discussions of design trade-offs, algorithm choices and runtime performance issues. In this paper we discuss a collection of mini-applications and demonstrate how we use them to analyze and improve application performance on new and future computer platforms.

Toward a new metric for ranking high performance computing systems.

The High Performance Linpack (HPL), or Top 500, benchmark [1] is the most widely recognized and discussed metric for ranking high performance computing systems. However, HPL is increasingly unreliable as a true measure of system performance for a growing collection of important science and engineering applications. In this paper we describe a new high performance conjugate gradient (HPCG) benchmark. HPCG is composed of computations and data access patterns more commonly found in applications. Using HPCG we strive for a better correlation to real scientific application performance and expect to drive computer system design and implementation in directions that will better impact performance improvement.

Poblano v1.0: A Matlab Toolbox for Gradient-Based Optimization

We present Poblano v1.0, a Matlab toolbox for solving gradient-based unconstrained optimization problems. Poblano implements three optimization methods (nonlinear conjugate gradients, limited-memory BFGS, and truncated Newton) that require only first order derivative information. In this paper, we describe the Poblano methods, provide numerous examples on how to use Poblano, and present results of Poblano used in solving problems from a standard test collection of unconstrained optimization problems.

An optimization approach for fitting canonical tensor decompositions.

Tensor decompositions are higher-order analogues of matrix decompositions and have proven to be powerful tools for data analysis. In particular, we are interested in the canonical tensor decomposition, otherwise known as the CANDECOMP/PARAFAC decomposition (CPD), which expresses a tensor as the sum of component rank-one tensors and is used in a multitude of applications such as chemometrics, signal processing, neuroscience, and web analysis. The task of computing the CPD, however, can be difficult. The typical approach is based on alternating least squares (ALS) optimization, which can be remarkably fast but is not very accurate. Previously, nonlinear least squares (NLS) methods have also been recommended; existing NLS methods are accurate but slow. In this paper, we propose the use of gradient-based optimization methods. We discuss the mathematical calculation of the derivatives and further show that they can be computed efficiently, at the same cost as one iteration of ALS. Computational experiments demonstrate that the gradient-based optimization methods are much more accurate than ALS and orders of magnitude faster than NLS.

Condence Region Estimation Techniques for Nonlinear Regression: Three Case Studies

This work focuses on different methods to generate confidence regions for nonlinear parameter identification problems. Three methods for confidence region estimation are considered: a linear approximation method, an F-test method, and a Log-Likelihood method. Each of these methods are applied to three case studies. One case study is a problem with synthetic data, and the other two case studies identify hydraulic parameters in groundwater flow problems based on experimental well-test results. The confidence regions for each case study are analyzed and compared. Although the F-test and Log-Likelihood methods result in similar regions, there are differences between these regions and the regions generated by the linear approximation method for nonlinear problems. The differing results, capabilities, and drawbacks of all three methods are discussed.

APPSPACK 4.0 : asynchronous parallel pattern search for derivative-free optimization.

APPSPACK is software for solving unconstrained and bound constrained optimization problems. It implements an asynchronous parallel pattern search method that has been specifically designed for problems characterized by expensive function evaluations. Using APPSPACK to solve optimization problems has several advantages: No derivative information is needed; the procedure for evaluating the objective function can be executed via a separate program or script; the code can be run in serial or parallel, regardless of whether or not the function evaluation itself is parallel; and the software is freely available. We describe the underlying algorithm, data structures, and features of APPSPACK version 4.0 as well as how to use and customize the software.

Peridigm Users' Guide v1.0.0

Peridigm is Sandia’s primary open-source computational peridynamics code. It is a component software project, built largely upon Sandia’s Trilinos project and Sandia’s agile software components efforts. It is massively parallel, utilizes peridynamic state-based material models, Exodus/Genesis-format mesh input, Exodus-format output, and multiple material blocks. It performs explicit dynamic, implicit dynamic, and quasistatic analyses utilizing powerful nonlinear and linear solvers.

Summary of the CSRI Workshop on Combinatorial Algebraic Topology (CAT): Software, Applications, & Algorithms

This report summarizes the Combinatorial Algebraic Topology: software, applications & algorithms workshop (CAT Workshop). The workshop was sponsored by the Computer Science Research Institute of Sandia National Laboratories. It was organized by CSRI staff members Scott Mitchell and Shawn Martin. It was held in Santa Fe, New Mexico, August 2930. The CAT Workshop website has links to some of the talk slides and other information, http://www.cs.sandia.gov/CSRI/Workshops/2009/CAT/index.html. The purpose of the report is to summarize the discussions and recap the sessions. There is a special emphasis on technical areas that are ripe for further exploration, and the plans for follow-up amongst the workshop participants. The intended audiences are the workshop participants, other researchers in the area, and the workshop sponsors.

Sensitivity of Precipitation to Parameter Values in the Community Atmosphere Model Version 5

One objective of the Climate Science for a Sustainable Energy Future (CSSEF) program is to develop the capability to thoroughly test and understand the uncertainties in the overall climate model and its components as they are being developed. The focus on uncertainties involves sensitivity analysis: the capability to determine which input parameters have a major influence on the output responses of interest. This report presents some initial sensitivity analysis results performed by Lawrence Livermore National Laboratory (LNNL), Sandia National Laboratories (SNL), and Pacific Northwest National Laboratory (PNNL). In the 2011-2012 timeframe, these laboratories worked in collaboration to perform sensitivity analyses of a set of CAM5, 2◦ runs, where the response metrics of interest were precipitation metrics. The three labs performed their sensitivity analysis (SA) studies separately and then compared results. Overall, the results were quite consistent with each other although the methods used were different. This exercise provided a robustness check of the global sensitivity analysis metrics and identified some strongly influential parameters.

Analysis of Global Horizontal Irradiance in Version 3 of the National Solar Radiation Database

Abstract We report an analysis that compares global horizontal irradiance (GHI) estimates from version 3 of the National Solar Radiation Database (NSRDB v3) with surface measurements of GHI at a wide variety of locations over the period spanning from 2005 to 2012The NSRDB v3 es. timate of GHI are derived from the Physical Solar Model (PSM) which employs physics-based models to estimate GHI from measurements of reflected visible and infrared irradiance collected by Geostationary Operational Envonment Satellites (GOES) and several other data sourcesir . Because the ground measurements themselves are uncertain our analysis does not establish the absolute accuracy for GHI. However by examining the comparison for trends and for PSMconsistency across a large number of sites, we may establish a level of confidence in GHI PSMand identify conditions which indicate opportunities to improve PSM. We focus our evaluation on annual and monthly insolation because these quantities directly relate to prediction of energy production from solar power systems. We find that generally, PSMGHI exhibits a bias towards overestimating insolation, on the order of 5% when all sky conditions are considered, and somewhat less (~3%) when only clear sky conditions are considered. The biases persist across multiple years and are evident at many locations. In our opinion the bias originates with and we view as less credible that the bias stems from PSMcalibration drift or soiling of ground instruments. We observe that GHI may significantly underestimate monthly insolationPSM in locations subject to broad snow cover. We found examples of days where PSM GHI apparently misidentified snow cover as clouds, resulting in significant underestimates of GHI during these days and hence leading to substantial understatement of monthly insolation.Analysis of PSM GHI in adjacent pixels shows that the level of agreement between PSM GHI and ground data can vary substantially over distances on the order of 2 km. We conclude that the variance most likely originates from dramatic contrasts in the ground’s appearance over these distances.