Search NASA⌕ Search

SEARCH · Search NASA

Results for “Data Variables”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 541 records · Page 30

Material World: Forecasting Household Appliance Ownership in a Growing Global Economy

Over the past years the Lawrence Berkeley National Laboratory (LBNL) has developed an econometric model that predicts appliance ownership at the household level based on macroeconomic variables such as household income (corrected for purchase power parity), electrification, urbanization and climate variables. Hundreds of data points from around the world were collected in order to understand trends in acquisition of new appliances by households, especially in developing countries. The appliances covered by this model are refrigerators, lighting fixtures, air conditioners, washing machines and televisions. The approach followed allows the modeler to construct a bottom-up analysis based at the end use and the household level. It captures the appliance uptake and the saturation effect which will affect the energy demand growth in the residential sector. With this approach, the modeler can also account for stock changes in technology and efficiency as a function of time. This serves two important functions with regard to evaluation of the impact of energy efficiency policies. First, it provides insight into which end uses will be responsible for the largest share of demand growth, and therefore should be policy priorities. Second, it provides a characterization of the rate at which policies affecting new equipment penetrate the appliance stock. Over the past 3 years, this method has been used to support the development of energy demand forecasts at the country, region or global level.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Generation and validation of comprehensive synthetic weather histories using auto-regressive moving-average models

As energy system design moves to more complex methods of optimization including machine learning there is a significant need for more weather data than is available. One method to solve this is using synthetic data models such as the auto-regressive moving-average (ARMA) model which has been frequently utilized to create such data. This paper looks at extending the ARMA algorithm to generate solar components through the use of clearsky detrending, maintaining vector relationships and by leveraging physical relationships. The method for the creation of entirely synthetic weather data files including key weather variables for energy system analysis is presented. Furthermore, a detailed comparison of energy system simulations utilizing both real and synthetic data is made using NREL’s System Advisor Model. Whilst good agreement is made for the solar variables, and other weather variables, ARMA methods often fail to capture the standard deviation and skew of annual weather distributions. Vector-ARMA is shown to maintain correlations between variables and thus generate data sets that perform similarly in energy system design. Here, it is finally shown that the ARMA method fails to preserve day-today correlations in weather variables and thus over-predicts optimal energy storage by 21% for a residential solar application.

42 ENGINEERING↗

Regional Feedstock Partnership Biomass Quality Assessment Final Report

The United States (U.S.) Department of Energy (DOE) developed the Billion-Ton Vision to enable production of one-billion tons of sustainable, reliable biomass for the bioenergy industry by 2030 (Perlack et al., 2005). The Sun Grant Regional Feedstock Partnership (RFP) was organized to fill information gaps and validate biomass yield assumptions related to the Billion-Ton Study (Owens, 2018; Owens, Karlen, and Lacey, 2016). Along with the more than 130 scientific publications generated from these studies, yield and sustainability data from the RFP field trials not only validated the Billion-Ton estimates, but were critical in developing both the U.S. Billion-Ton Update report in 2011 and the 2016 Billion-Ton Report (DOE, 2011; 2016). The intention of this biomass quality assessment report is to build on these initial successes from the RFP field trials by focusing on variability in biomass quality data necessary to evaluate conversion performance. This report contains a summary of chemical quality results from samples collected as part of the RFP field trials. This report focuses on assessment of the impact of experimental agronomic designs on biomass properties followed by analyses of the impact of environmental and production variables on biomass properties. Datasets include species and other genetic variables, fertilizer treatments, harvest information, and yield, as well as other publicly available data such as precipitation, temperature, soil properties, and drought. The key outcomes from this chemical quality focused assessment have included: • Complete evaluation of the impacts of agronomic designs, genetics, and environmental conditions on chemical properties for Miscanthus, switchgrass, sorghum, energycane, mixed perennial grasses, and shrub willow short-rotation feedstocks • Over 30 peer review publications and technical reports focused on variability in quality data • Development of spatial and temporal environmental quality prediction maps for Miscanthus and switchgrass feedstocks allowing for comprehensive evaluation of variability in feedstock chemical quality across U.S. regions and over multiple harvest years

09 BIOMASS FUELS↗

Flexible and scalable methods for quantifying stochastic variability in the era of massive time-domain astronomical data sets

We present the use of continuous-time autoregressive moving average (CARMA) models as a method for estimating the variability features of a light curve, and in particular its power spectral density (PSD). CARMA models fully account for irregular sampling and measurement errors, making them valuable for quantifying variability, forecasting and interpolating light curves, and variability-based classification. We show that the PSD of a CARMA model can be expressed as a sum of Lorentzian functions, which makes them extremely flexible and able to model a broad range of PSDs. We present the likelihood function for light curves sampled from CARMA processes, placing them on a statistically rigorous foundation, and we present a Bayesian method to infer the probability distribution of the PSD given the measured light curve. Because calculation of the likelihood function scales linearly with the number of data points, CARMA modeling scales to current and future massive time-domain data sets. We conclude by applying our CARMA modeling approach to light curves for an X-ray binary, two active galactic nuclei, a long-period variable star, and an RR Lyrae star in order to illustrate their use, applicability, and interpretation.

79 ASTRONOMY AND ASTROPHYSICS↗

Systems and methods for control of polymer reactions and processing using automatic continuous online monitoring

Manual and automatic methods and devices using a ACOMP system for active control of polymerization reaction processes. An ideal desired trajectory of one or more reaction and polymer characteristics can be established to produce a desired final polymer product with specified characteristics from a polymerization reaction process. A current reaction trajectory of a polymerization reaction process can be driven to an ideal or desired reaction trajectory. In a manual embodiment an operator can use ACOMP data to adjust process variables in order to drive the current reaction trajectory toward the ideal or desired reaction trajectory. In an automated mode a control program can use ACOMP data to make adjustments to process variables to drive the polymerization reaction process toward the desired trajectory as closely as possible either empirically or by solving the governing equations for the polymerization reaction process.

Reed, Wayne Frederick↗

Distributionally Robust Decision Making Leveraging Conditional Distributions

Distributionally robust optimization (DRO) is a powerful tool for decision making under uncertainty. It is particularly appealing because of its ability to leverage existing data. However, many practical problems call for decision-making with some auxiliary information, and DRO in the context of conditional distributions is not straightforward. We propose a conditional kernel distributionally robust optimization (CKDRO) method that enables robust decision making under conditional distributions through kernel DRO and the conditional mean operator in the reproducing kernel Hilbert space (RKHS). In particular, we consider problems where there is a correlation between the unknown variable y and an auxiliary observable variable x. Given past data of the two variables and a queried auxiliary variable, CKDRO represents the conditional distribution P(y|x) as the conditional mean operator in the RKHS space and quantifies the ambiguity set in the RKHS as well, which depends on the size of the dataset as well as the query point. To justify the use of RKHS, we demonstrate that the ambiguity set defined in RKHS can be viewed as a ball under a metric that is similar to the Wasserstein metric. The DRO is then dualized and solved via a finite dimensional convex program. The proposed CKDRO approach is applied to a generation scheduling problem and shows that the result of CKDRO is superior to common benchmarks in terms of quality and robustness.

Chen, Yuxiao↗

Capturing provenance as a diagnostic tool for workflow performance evaluation and optimization

In extreme-scale computing environments such as the DOE Leadership Computing Facilities scientific workflows are routinely used to coordinate software processes for the execution of complex, computational applications that perform in-silico experiments. Monitoring the performance of workflows without also simultaneously tracking provenance is not sufficient to understand variations between runs, configurations, versions of a code, and between changes in an implemented stack, and systems, i.e. the variability of performance metrics data in their historical context. We take a provenance-based approach and demonstrate that provenance is useful as a tool for evaluating and optimizing workflow performance in extreme-scale HPC environments. We present Chimbuko, a framework for the analysis and visualization of the provenance of performance. Chimbuko implements a method for the evaluation of workflow performance from multiple components that enables the exploration of performance metrics data at scale.

97 MATHEMATICS AND COMPUTING↗

Multivariate Pointwise Information-Driven Data Sampling and Visualization

With increasing computing capabilities of modern supercomputers, the size of the data generated from the scientific simulations is growing rapidly. As a result, application scientists need effective data summarization techniques that can reduce large-scale multivariate spatiotemporal data sets while preserving the important data properties so that the reduced data can answer domain-specific queries involving multiple variables with sufficient accuracy. While analyzing complex scientific events, domain experts often analyze and visualize two or more variables together to obtain a better understanding of the characteristics of the data features. Therefore, data summarization techniques are required to analyze multi-variable relationships in detail and then perform data reduction such that the important features involving multiple variables are preserved in the reduced data. To achieve this, in this work, we propose a data sub-sampling algorithm for performing statistical data summarization that leverages pointwise information theoretic measures to quantify the statistical association of data points considering multiple variables and generates a sub-sampled data that preserves the statistical association among multi-variables. Using such reduced sampled data, we show that multivariate feature query and analysis can be done effectively. The efficacy of the proposed multivariate association driven sampling algorithm is presented by applying it on several scientific data sets.

97 MATHEMATICS AND COMPUTING↗

Reactive particle-tracking solutions to a benchmark problem on heavy metal cycling in lake sediments

Geochemical systems are known to exhibit highly variable spatiotemporal behavior. This may be observed both in non-smooth concentration curves in space for a single sampling time and also in variability between samples taken from the same location at different times. However, most models that are designed to simulate these systems provide only single-solution smooth curves and fail to capture the noise and variability seen in the data. We apply a recently developed reactive particle-tracking method to a system that displays highly complex geochemical behavior. When the method is made to most closely resemble a corresponding Eulerian method, in its unperturbed form, we see near-exact match between solutions of the two models. More importantly, we consider two approaches for perturbing the model and find that the spatially-perturbed condition is able to capture a greater degree of the variability present in the data. This method of perturbation is a task to which particle methods are uniquely suited and Eulerian models are not well-suited. Additionally, because of the nature of the algorithm, noisy spatial gradients can be highly resolved by a large number of mobile particles, and this incurs negligible computational cost, as compared to expensive chemistry calculations.

54 ENVIRONMENTAL SCIENCES↗

MFNets: Multifidelity data-driven networks for Bayesian learning and prediction

This paper presents a multifidelity uncertainty quantification framework called MFNets. We seek to address three existing challenges that arise when experimental and simulation data from different sources are used to enhance statistical estimation and prediction with quantified uncertainty. Specifically, we demonstrate that MFNets can (1) fuse heterogeneous data sources arising from simulations with different parameterizations, e.g simulation models with different uncertain parameters or data sets collected under different environmental conditions; (2) encode known relationships among data sources to reduce data requirements; and (3) improve the robustness of existing multi-fidelity approaches to corrupted data. MFNets construct a network of latent variables (LVs) to facilitate the fusion of data from an ensemble of sources of varying credibility and cost. These LVs are posited as explanatory variables that provide the source of correlation in the observed data. Furthermore, MFNets provide a way to encode prior physical knowledge to enable efficient estimation of statistics and/or construction of surrogates via conditional independence relations on the LVs. We highlight the utility of our framework with a number of theoretical results which assess the quality of the posterior mean as a frequentist estimator and compare it to standard sampling approaches that use single fidelity, multilevel, and control variate Monte Carlo estimators. We also use the proposed framework to derive the Monte Carlo-based control variate estimator entirely from the use of Bayes rule and linear-Gaussian models -- to our knowledge the first such derivation. Finally, we demonstrate the ability to work with different uncertain parameters across different models.

97 MATHEMATICS AND COMPUTING↗

A method for predicting failure statistics for steady state elevated temperature structural components

This paper presents the initial development of a high temperature life prediction method that accounts for the variability in the material properties of Grade 91 steel. The method accounts for material variability by fitting a variable 3-parameter Weibull distribution to experimental rupture data and accounts for the variability of creep deformation on the steady-state stresses via a Monte Carlo approach. To ensure reasonable computational times, the model represents the material as an extremely viscous Stokes fluid with a non-Newtonian viscosity, therefore solving the stress relaxation problem with a steady, static, instead of transient, analysis. Furthermore, the complete statistical analysis combines this model for creep deformation with a probabilistic model for creep rupture to evaluate the probability of premature failure for a set of sample problems, comparing the predicted failure statistics to the design life predicted by the ASME Boiler and Pressure Vessel Code rules.

42 ENGINEERING↗

Crawling the Control System

Information about accelerator operations and the control system resides in various formats in a variety of places on the lab network. There are operating procedures, technical notes, engineering drawings, and other formal controlled documents. There are programmer references and API documentation generated by tools such as doxygen and javadoc. There are the thousands of electronic records generated by and stored in databases and applications such as electronic logbooks, training materials, wikis, and bulletin boards and the contents of text-based configuration files and log files that can also be valuable sources of information. The obvious way to aggregate all these sources is to index them with a search engine that users can then query from a web browser. Toward this end, the Google "mini" search appliance was selected and implemented because of its low cost and its simple web-based configuration and management. In addition to crawling and indexing electronic documents, the appliance provides an API that has been used to supplement search results with live control system data such as current values of EPICS process variables and graphs of recent data from the archiver.

43 PARTICLE ACCELERATORS↗

Climate data, analysis and models for the study of natural variability and anthropogenic change

Gridded Temperature Under prior/current support, we completed and published (Jones et al., 2012) the fourth major update to our global land dataset of near-surface air temperatures, CRUTEM4. This is one of the most widely used records of the climate system, having been updated, maintained and further developed with DoE support since the 1980s. We have continued to update the CRUTEM4 (Jones et al., 2012) database that is combined with marine data to produce HadCRUT4 (Morice et al., 2012). The emphasis in our use of station temperature data is to access as many land series that have been homogenized by National Meteorological Services (NMSs, including NCDC/NOAA, Asheville, NC). Unlike the three US groups monitoring surface temperatures in a similar way, we do not infill areas that have no or missing data. We can only infill such regions in CRUTEM4 by accessing more station temperature series. During early 2014, we have begun the extensive task of updating as many of these series as possible using data provided by some NMSs and also through a number of research projects and programs around the world. All the station data used in CRUTEM4 have been available since 2009, but in Osborn and Jones (2014) we have made this more usable using a Google Earth interface (http://www.cru.uea.ac.uk/cru/data/crutem/ge/ ). We have recently completed the update of our infilled land multi-variable dataset (CRU TS 3.10, Harris et al., 2014). This additionally produces complete land fields (except for the Antarctic) for temperature, precipitation, diurnal temperature range, vapour pressure and sunshine/cloud. Using this dataset we have calculated sc-PDSI (self-calibrating Palmer Drought Severity Index) data and compared with other PDSI datasets (Trenberth et al., 2014). Also using CRU TS 3.10 and Reanalysis datasets, we showed no overall increase in global temperature variability despite changing regional patterns (Huntingford et al., 2013). Harris et al. (2014) is an update of an earlier dataset (Mitchell and Jones, 2005) which also had earlier DoE support. The earlier dataset has been cited over 1700 times according to ResearcherID on 31/July/2014 and the recent paper has already been cited 22 times. Analyses of Temperature Data Using the ERA-Interim estimate of the absolute surface air temperature of the Earth (instead of in the more normal form of anomalies) we compared the result against estimates we produced in 1999 with earlier DoE support. The two estimates are surprisingly close (differing by a couple of tenths of a degree Celsius), with the average temperature of the world (for 1981-2010) being very close to 14°C (Jones and Harpham, 2013). We have assessed ERA-Interim against station temperatures from manned and automatic weather station measurements across the Antarctic (Jones and Lister, 2014). Agreement is generally excellent across the Antarctic Peninsula and the sparsely sampled western parts of Antarctica. Differences tend to occur over eastern Antarctica where ERA-Interim is biased warm (up to 6°C) in the interior of the continent and biased cool (up to 6°C) for some of the coastal locations. Opportunities presented themselves during 2012 for collaborative work with a couple of Chinese groups. Three papers develop new temperature series for China as a whole and also for the eastern third of China (Wang et al., 2014, Cao et al., 2013 and Zhao et al., 2014). A dataset of ~400 daily Chinese temperature stations has been added to the CRU datasets. The latter paper finds that urban effects are generally about 10% of the long-term warming trend across eastern China. A fourth paper (Wang et al., 2013) illustrates issues with comparisons between reanalyses and surface temperatures across China, a method that has been widely used by some to suggest urban heating effects are much larger in the region. ERA-Interim can be used but NCEP/NCAR comparisons are very dependent on the period analysed. Earlier a new temperature dataset of homogenized records was developed for China (Li et al., 2009). Urbanization has also been addressed for London (Jones and Lister, 2009) where two rural sites have not warmed more than a city centre site since 1900. Additionally, in Ethymiadis and Jones (2010) we show that land air temperatures agree with marine data around coastal areas, further illustrating that urbanization is not a major component of large-scale surface air temperature change. Early instrumental data (before the development of modern thermometer screens) have always been suspected of being biased warm in summer, due to possible direct exposure to the sun. Two studies (Böhm et al., 2010 and Brunet et al., 2010) show this for the Greater Alpine Region (GAR) and for mainland Spain respectively. The issue is important before about 1870 in the GAR and before about 1900 in Spain. After correction for the problems, summer temperature estimates before these dates are cooler by about 0.4°C. In Jones and Wigley (2010), we discussed the importance of the biases in global temperature estimation. Exposure and to a lesser extent urbanization are the most important biases for the land areas, but both are dwarfed by the necessary adjustments for bucket SST measurements before about 1950. Individual station homogeneity is only important at the local scale. This was additionally illustrated by Hawkins and Jones (2013) where we replicated the temperature record developed by Guy Stewart Callendar in papers in 1938 and 1961. Analyses of Daily Climate Data Work here indicates that ERA-Interim (at least in Europe, Cornes and Jones, 2013, discussed in more detail in this proposal) can be used to monitor extremes (using the ETCCDI software – see Zhang et al., 2011). Additionally, also as a result of Chinese collaboration, a new method of daily temperature homogenization has been developed (Li et al., 2014). In Cornes and Jones (2011) we assessed storm activity in the northeast Atlantic region using daily gridded data. Even though the grid resolution is coarse (5° by 5° lat/long) the changes in storm activity are similar to those developed from the pressure triangle approach with station data. Analyses of humidity and pressure data In Simmons et al. (2010) we showed a reduction in relative humidity over low-latitude and mid-latitude land areas for the 10 years to 2008, based on monthly anomalies of surface air temperature and humidity from ECMWF reanalyses (ERA-40 and ERA-Interim) and our earlier land-only dataset (CRUTEM3) and synoptic humidity observations (HadCRUH). Updates of this station-based humidity dataset (now called HadISDH) extend the record, showing continued reductions (Willett et al., 2013). Analyses of Proxy Temperature Data In Vinther et al. (2010), relationships between the seasonal stable isotope data from Greenland Ice Cores and Greenland and Icelandic instrumental temperatures were investigated for the past 150-200 years. The winter season stable isotope data are found to be influenced by the North Atlantic Oscillation (NAO) and very closely related to SW Greenland temperatures. The summer season stable isotope data display higher correlations with Icelandic summer temperatures and North Atlantic SST conditions than with local SW Greenland temperatures. In Jones et al. (2014) we use these winter isotope reconstructions to show the expected inverse correlation (due to the NAO) with winter-season documentary reconstructions from the Netherlands and Sweden over the last 800 years. Finally, in this section Jones et al. (2013) shows the agreement between tree-ring width measurements from Northern Sweden and Finland and an assessment of the link to explosive volcanic eruptions. An instrumental record for the region in the early 19th century indicates that the summer of 1816 was only slightly below normal, explaining why this year has normal growth for both ring width and density. GCM/RCM/Reanalysis Evaluation In this section we have intercompared daily temperature extremes across Europe in Cornes and Jones (2013) using station data, E-OBS and ERA-Interim. We have additionally considered the impact of the urban issue on the global scale using the results of the Compo et al. (2011) Reanalyses, 20CR. These only make use of SST and station pressure data. Across the world’s land areas, they indicate similar warming since 1900 to that which has occurred (Compo et al., 2013), again illustrating that urbanization is not the cause of the long-term warming. Changes in HadCRUH global land surface specific humidity and CRUTEM3 surface temperatures from 1973 to 1999 were compared to the CMIP3 archive of climate model simulations with 20th Century forcings (Willett et al., 2010). The models reproduce the magnitude of observed interannual variance over all large regions. Observed and modelled trends and temperature-humidity relationships are comparable with the exception of the extra-tropical Southern Hemisphere where observations exhibit no trend but models exhibit moistening.

54 ENVIRONMENTAL SCIENCES↗

Radial velocity curves of ellipsoidal red giant binaries in the Large Magellanic Cloud

Ellipsoidal red giant binaries are close binary systems where an unseen, relatively close companion distorts the red giant, leading to light variations as the red giant moves around its orbit. These binaries are likely to be the immediate evolutionary precursors of close binary planetary nebula and post-asymptotic giant branch and post-red giant branch stars. Due to the MACHO and OGLE photometric monitoring projects, the light variability nature of these ellipsoidal variables has been well studied. However, due to the lack of radial velocity curves, the nature of their masses, separations, and other orbital details has so far remained largely unknown. In order to improve this situation, we have carried out spectral monitoring observations of a large sample of 80 ellipsoidal variables in the Large Magellanic Cloud and we have derived radial velocity curves. At least 12 radial velocity points with good quality were obtained for most of the ellipsoidal variables. The radial velocity data are provided with this paper. Combining the photometric and radial velocity data, we present some statistical results related to the binary properties of these ellipsoidal variables.

79 ASTRONOMY AND ASTROPHYSICS↗

Assessment of the Idaho National Laboratory Remote Handled Low Level Waste Disposal Facility Hydraulic Performance

This document provides a summary of the expected hydraulic performance of the vault drainage system installed at Idaho National Laboratory’s Remote Handled Low-Level Waste disposal facility. Hydraulic performance was assessed through analysis of the mechanical properties of the drainage system and backfill materials, including material gradation and proctor data; laboratory data, including bulk density, porosity, unsaturated hydraulic conductivity, moisture-retention data, and van Genuchten parameters; and field-data collected using in-situ instrumentation with data collected during a vault-scale infiltration characterization test. Hydraulic performance of the vault drainage system was designed to promote a greater than 500-year vault concrete longevity. The predicted hydraulic performance is, in part, a function of the materials used in construction, sequencing of the hydraulic drainage materials, and the as-built condition of the materials following compaction necessary to ensure vault stability. Specifying specific as-built hydraulic properties is exceedingly difficult at the facility specification/design phase. Instead, material mechanical properties were specified and confirmed early during the construction process. As the drainage materials were installed, samples were sent to a laboratory for hydraulic property testing and, after drainage system installation was complete around the performance assessment vaults, a field-scale infiltration test was conducted with data collection to verify the as-built hydraulic performance. This report provides summary data for the measured mechanical properties, laboratory data, and field infiltration characterization test data. Mechanical data have relatively little variability because of the sieve system used when the materials were crushed onsite. Laboratory data were obtained for a relatively small number of material samples, but the data obtained for each sample are extensive. However, laboratory data provide the moisture retention relationships and relative hydraulic conductivity-moisture content relationships for uninstalled materials. Conducting field-scale infiltration tests with collection of moisture retention data and wetting front propagation data allows extension of laboratory data to the as-installed conditions of the vault drainage system. These combined data types are integrated using a numerical simulation of the infiltration characterization test to refine model parameters. The infiltration characterization model will be used to support the hydraulic and concrete performance of the vault system provided in the facility performance assessment.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

The public health exposome and pregnancy-related mortality in the United States: a high-dimensional computational analysis

Racial inequities in maternal mortality in the U.S. continue to be stark. The 2015–2018, 4-year total population, county-level, pregnancy-related mortality ratio (PRM; deaths per 100,000 live births; National Center for Health Statistics (NCHS), restricted use mortality file) was linked with the Public Health Exposome (PHE). Using data reduction techniques, 1591 variables were extracted from over 62,000 variables for use in this analysis, providing information on the relationships between PRM and the social, health and health care, natural, and built environments. Graph theoretical algorithms and Bayesian analysis were applied to PHE/PRM linked data to identify latent networks. PHE variables most strongly correlated with total population PRM were years of potential life lost and overall life expectancy. Population-level indicators of PRM were overall poverty, smoking, lack of exercise, heat, and lack of adequate access to food. In this high-dimensional analysis, overall life expectancy, poverty indicators, and health behaviors were found to be the strongest predictors of pregnancy-related mortality. This provides strong evidence that maternal death is part of a broader constellation of both similar and unique health behaviors, social determinants and environmental exposures as other causes of death.

60 APPLIED LIFE SCIENCES↗

LADR: low-cost application-level detector for reducing silent output corruptions

Applications running on future high performance computing (HPC) systems are more likely to experience transient faults due to technology scaling trends with respect to higher circuit density, smaller transistor size and near-threshold voltage (NTV) operations. A transient fault could corrupt application state without warning, possibly leading to incorrect application output. Such errors are called silent data corruptions (SDCs). In this paper, we present LADR, a low-cost application-level SDC detector for scientific applications. LADR protects scientific applications from SDCs by watching for data anomalies in their state variables (those of scientific interest). It employs compile-time data-flow analysis to minimize the number of monitored variables, thereby reducing runtime and memory overheads while maintaining a high level of fault coverage with low false positive rates. We evaluated LADR with 4 scientific workloads and results show that LADR achieved < 80% fault coverage with only ∼ 3% runtime overheads and ∼ 1% memory overheads. As compared to prior state-of-the-art anomaly-based detection methods, SDC achieved comparable or improved fault coverage, but reduced runtime overheads by 21% ∼ 75%, and memory overheads by 35% ∼ 55% for the evaluated workloads. We believe that such an approach with low memory and runtime overheads coupled with attractive detection precision makes LADR a viable approach for assuring the correct output from large-scale high performance simulations.

Chen, Chao↗