Search NASA⌕ Search

SEARCH · Search NASA

Results for “forecast error”

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 1,063 records · Page 59

MO-FG-303-03: Demonstration of Universal Knowledge-Based 3D Dose Prediction

Purpose: To demonstrate a knowledge-based 3D dose prediction methodology that can accurately predict achievable radiotherapy distributions. Methods: Using previously treated plans as input, an artificial neural network (ANN) was trained to predict 3D dose distributions based on 14 patient-specific anatomical parameters including the distance (r) to planning target volume (PTV) boundary, organ-at-risk (OAR) boundary distances, and angular position ( θ,φ). 23 prostate and 49 stereotactic radiosurgery (SRS) cases with ≥1 nearby OARs were studied. All were planned with volumetric-modulated arc therapy (VMAT) to prescription doses of 81Gy for prostate and 12–30Gy for SRS. Site-specific ANNs were trained using all prostate 23 plans and using a 24 randomly-selected subset for the SRS model. The remaining 25 SRS plans were used to validate the model. To quantify predictive accuracy, the dose difference between the clinical plan and prediction were calculated on a voxel-by-voxel basis δD(r,θ,φ)=Dclin(r,θ,φ)-Dpred(r, θ,φ). Grouping voxels by boundary distance, the mean <δ Dr>=(1/N)Σ -θ,φ D(r,θ,φ) and inter-quartile range (IQR) quantified the accuracy of this method for deriving DVH estimations. The standard deviation (σ) of δ D quantified the 3D dose prediction error on a voxel-by-voxel basis. Results: The ANNs were highly accurate in predictive ability for both prostate and SRS plans. For prostate, <δDr> ranged from −0.8% to +0.6% (max IQR=3.8%) over r=0–32mm, while 3D dose prediction accuracy averaged from σ=5–8% across the same range. For SRS, from r=0–34mm the training set <δDr> ranged from −3.7% to +1.5% (max IQR=4.4%) while the validation set <δDr> ranged from −2.2% to +5.8% (max IQR=5.3%). 3D dose prediction accuracy averaged σ=2.5% for the training set and σ=4.0% over the same interval. Conclusion: The study demonstrates this technique’s ability to predict achievable 3D dose distributions for VMAT SRS and prostate. Future investigations will attempt to separate and quantify modeling error and plan quality variations in the statistical noise. Research funding support from Varian Medical Systems.

60 APPLIED LIFE SCIENCES↗

Error localization in RHIC by fitting difference orbits

The presence of realistic errors in an accelerator or in the model used to describe the accelerator are such that a measurement of the beam trajectory may deviate from prediction. Comparison of measurements to model can be used to detect such errors. To do so the initial conditions (phase space parameters at any point) must be determined which can be achieved by fitting the difference orbit compared to model prediction using only a few beam position measurements. Using these initial conditions, the fitted orbit can be propagated along the beam line based on the optics model. Measurement and model will agree up to the point of an error. The error source can be better localized by additionally fitting the difference orbit using downstream BPMs and back-propagating the solution. If one dominating error source exist in the machine, the fitted orbit will deviate from the difference orbit at the same point.

43 PARTICLE ACCELERATORS↗

Effect of GOES-R Image Navigation and Registration Errors on Atmospheric Motion Vectors

High temporal frequency imagery from geostationary satellites allows for the continuous monitoring of rapidly changing atmospheric constituents such as smoke, dust, water vapor and clouds. The image sequences are often used to quantify the displacement of image features such as water vapor and clouds to produce atmospheric motion vectors (AMVs) which are used as diagnostic tools and also assimilated into numerical weather forecast models. The basic principle behind the determination of AMVs is the calculation of the physical displacement of features from one image (time) to the next. This process assumes that the features being tracked do not change as a function of time, usually requiring the use of short time interval imagery to minimize substantial change in size and shape of the features being tracked. High spatial resolution imagery also is required for reliable feature identification. While these image resolution and temporal sampling requirements often provide major drivers for space-based instrument design requirements, accurate image navigation and registration, INn (between a sequence of images), is also critical to the derivation of useful AMVs. In this paper and poster to be presented at the conference, the image navigation and registration (INR) accuracy expected for the Advanced Baseline Imager (ABI) on the GOES-R series of satellites will be discussed in light of its impact on AMV accuracy. Significant satellite platform and modeling enhancements are planned which should significantly improve INn performance of the GOES-R instruments. Some of these improvements have been demonstrated for the GOES-13 satellite which was launched in summer of 2006. An analysis of GOES-13 INR data, from the special satellite check out period, will be used in the assessment.

Jedlovec, Gary↗

Predicting oxidation damage of ultra high-temperature carbide ceramics in extreme environments using machine learning

Determining the oxidation resistance of UHTC carbides in extreme environments is challenging theoretically and experimentally due to the high dimensional complexity of influencing variables and intricate testing setups. Herein we demonstrate the use of machine learning (ML) models trained with experimental literature data to predict the oxide thickness of UHTC carbides exposed to air based on composition, mean grain size, relative densification, holding time, and temperature. A multi-dimensional database with 76 occurrences is created containing experimental results of Hf, Zr, and Ta carbides plus additives. In this study, the preprocessed database is then used to train ML models to predict their oxidation behavior. The trained model predicts the oxidation damage in the form of an average oxide thickness in UHTC carbides with a Mean Absolute Error (MAE) of ±65.45 μm for samples in the testing set that developed thicknesses up to 1000 μm. The model successfully predicted oxidation damage for a recession rate lower than 60 μm/min. It is noticed that the ensemble method MAE is increased to ±134.34 μm while forecasting the oxidation of samples with a recession rate higher than the threshold. The unprecedented approach is a novel way to predict the damage through the oxidation of carbide compounds before processing for a smarter design with room for improvement.

36 MATERIALS SCIENCE↗

A Global Data Assimilation System for Atmospheric Aerosol

We will give an overview of an aerosol data assimilation system which combines advances in remote sensing of atmospheric aerosols, aerosol modeling and data assimilation methodology to produce high spatial and temporal resolution 3D aerosol fields. Initially, the Goddard Aerosol Assimilation System (GAAS) will assimilate TOMS, AVHRR and AERONET observations; later we will include MODIS and MISR. This data assimilation capability will allows us to integrate complementing aerosol observations from these platforms, enabling the development of an assimilated aerosol climatology as well as a global aerosol forecasting system in support of field campaigns. Furthermore, this system provides an interactive retrieval framework for each aerosol observing satellites, in particular TOMS and AVHRR. The Goddard Aerosol Assimilation System (GAAS) takes advantage of recent advances in constituent data assimilation at DAO, including flow dependent parameterizations of error covariances and the proper consideration of model bias. For its prognostic transport model, GAAS will utilize the Goddard Ozone, Chemistry, Aerosol, Radiation and Transport (GOCART) model developed at NASA/GSFC Codes 916 and 910.3. GOCART includes the Lin-Rood flux-form, semi-Langrangian transport model with parameterized aerosol chemistry and physical processes for absorbing (dust and black carbon) and non-absorbing aerosols (sulfate and organic carbon). Observations and model fields are combined using a constituent version of DAO's Physical-space Statistical Analysis System (PSAS), including its adaptive quality control system. In this talk we describe the main components of this assimilation system and present preliminary results obtained by assimilating TOMS data.

daSilva, Arlindo↗

Accelerated Aging in Electrolytic Capacitors for Prognostics

The focus of this work is the analysis of different degradation phenomena based on thermal overstress and electrical overstress accelerated aging systems and the use of accelerated aging techniques for prognostics algorithm development. Results on thermal overstress and electrical overstress experiments are presented. In addition, preliminary results toward the development of physics-based degradation models are presented focusing on the electrolyte evaporation failure mechanism. An empirical degradation model based on percentage capacitance loss under electrical overstress is presented and used in: (i) a Bayesian-based implementation of model-based prognostics using a discrete Kalman filter for health state estimation, and (ii) a dynamic system representation of the degradation model for forecasting and remaining useful life (RUL) estimation. A leave-one-out validation methodology is used to assess the validity of the methodology under the small sample size constrain. The results observed on the RUL estimation are consistent through the validation tests comparing relative accuracy and prediction error. It has been observed that the inaccuracy of the model to represent the change in degradation behavior observed at the end of the test data is consistent throughout the validation tests, indicating the need of a more detailed degradation model or the use of an algorithm that could estimate model parameters on-line. Based on the observed degradation process under different stress intensity with rest periods, the need for more sophisticated degradation models is further supported. The current degradation model does not represent the capacitance recovery over rest periods following an accelerated aging stress period.

Celaya, Jose R.↗

Evaluation of a data fusion approach to estimate daily PM{sub 2.5} levels in North China

PM{sub 2.5} air pollution has been a growing concern worldwide. Previous studies have conducted several techniques to estimate PM{sub 2.5} exposure spatiotemporally in China, but all these have limitations. This study was to develop a data fusion approach and compare it with kriging and Chemistry Module. Two techniques were applied to create daily spatial cover of PM{sub 2.5} in grid cells with a resolution of 10 km in North China in 2013, respectively, which was kriging with an external drift (KED) and Weather Research and Forecast Model with Chemistry Module (WRF-Chem). A data fusion technique was developed by fusing PM{sub 2.5} concentration predicted by KED and WRF-Chem, accounting for the distance from the central of grid cell to the nearest ground observations and daily spatial correlations between WRF-Chem and observations. Model performances were evaluated by comparing them with ground observations and the spatial prediction errors. KED and data fusion performed better at monitoring sites with a daily model R{sup 2} of 0.95 and 0.94, respectively and PM{sub 2.5} was overestimated by WRF-Chem (R{sup 2}=0.51). KED and data fusion performed better around the ground monitors, WRF-Chem performed relative worse with high prediction errors in the central of study domain. In our study, both KED and data fusion technique provided highly accurate PM{sub 2.5}. Current monitoring network in North China was dense enough to provide a reliable PM{sub 2.5} prediction by interpolation technique. - Highlights: • KED and data fusion model predicted daily PM{sub 2.5} with high accuracy. • WRF-Chem performed worse in PM{sub 2.5} prediction compared with KED and data fusion. • The PM{sub 2.5} monitoring network in North China was able to support reliable PM{sub 2.5} interpolation.

54 ENVIRONMENTAL SCIENCES↗

A Method for Assessing the Quality of Model-Based Estimates of Ground Temperature and Atmospheric Moisture Using Satellite Data

A method is developed for validating model-based estimates of atmospheric moisture and ground temperature using satellite data. The approach relates errors in estimates of clear-sky longwave fluxes at the top of the Earth-atmosphere system to errors in geophysical parameters. The fluxes include clear-sky outgoing longwave radiation (CLR) and radiative flux in the window region between 8 and 12 microns (RadWn). The approach capitalizes on the availability of satellite estimates of CLR and RadWn and other auxiliary satellite data, and multiple global four-dimensional data assimilation (4-DDA) products. The basic methodology employs off-line forward radiative transfer calculations to generate synthetic clear-sky longwave fluxes from two different 4-DDA data sets. Simple linear regression is used to relate the clear-sky longwave flux discrepancies to discrepancies in ground temperature ((delta)T(sub g)) and broad-layer integrated atmospheric precipitable water ((delta)pw). The slopes of the regression lines define sensitivity parameters which can be exploited to help interpret mismatches between satellite observations and model-based estimates of clear-sky longwave fluxes. For illustration we analyze the discrepancies in the clear-sky longwave fluxes between an early implementation of the Goddard Earth Observing System Data Assimilation System (GEOS2) and a recent operational version of the European Centre for Medium-Range Weather Forecasts data assimilation system. The analysis of the synthetic clear-sky flux data shows that simple linear regression employing (delta)T(sub g)) and broad layer (delta)pw provides a good approximation to the full radiative transfer calculations, typically explaining more thin 90% of the 6 hourly variance in the flux differences. These simple regression relations can be inverted to "retrieve" the errors in the geophysical parameters, Uncertainties (normalized by standard deviation) in the monthly mean retrieved parameters range from 7% for (delta)T(sub g) to approx. 20% for the lower tropospheric moisture between 500 hPa and surface. The regression relationships developed from the synthetic flux data, together with CLR and RadWn observed with the Clouds and Earth Radiant Energy System instrument, ire used to assess the quality of the GEOS2 T(sub g) and pw. Results showed that the GEOS2 T(sub g) is too cold over land, and pw in upper layers is too high over the tropical oceans and too low in the lower atmosphere.

Wu, Man Li C.↗

Streamlining Ocean Dynamics Modeling with Fourier Neural Operators: A Multiobjective Hyperparameter and Architecture Optimization Approach

Training an effective deep learning model to learn ocean processes involves careful choices of various hyperparameters. We leverage DeepHyper’s advanced search algorithms for multiobjective optimization, streamlining the development of neural networks tailored for ocean modeling. The focus is on optimizing Fourier neural operators (FNOs), a data-driven model capable of simulating complex ocean behaviors. Selecting the correct model and tuning the hyperparameters are challenging tasks, requiring much effort to ensure model accuracy. DeepHyper allows efficient exploration of hyperparameters associated with data preprocessing, FNO architecture-related hyperparameters, and various model training strategies. We aim to obtain an optimal set of hyperparameters leading to the most performant model. Moreover, on top of the commonly used mean squared error for model training, we propose adopting the negative anomaly correlation coefficient as the additional loss term to improve model performance and investigate the potential trade-off between the two terms. The numerical experiments show that the optimal set of hyperparameters enhanced model performance in single timestepping forecasting and greatly exceeded the baseline configuration in the autoregressive rollout for long-horizon forecasting up to 30 days. Utilizing DeepHyper, we demonstrate an approach to enhance the use of FNO in ocean dynamics forecasting, offering a scalable solution with improved precision.

97 MATHEMATICS AND COMPUTING↗

Examples of data assimilation in mesoscale models

The keynote address was the problem of physical initialization of mesoscale models. The classic purpose of physical or diabatic initialization is to reduce or eliminate the spin-up error caused by the lack, at the initial time, of the fully developed vertical circulations required to support regions of large rainfall rates. However, even if a model has no spin-up problem, imposition of observed moisture and heating rate information during assimilation can improve quantitative precipitation forecasts, especially early in the forecast. The two key issues in physical initialization are the choice of assimilating technique and sources of hydrologic/hydrometeor data. Another example of data assimilation in mesoscale models was presented in a series of meso-beta scale model experiments with and 11 km version of the MASS model designed to investigate the sensitivity of convective initiation forced by thermally direct circulations resulting from differential surface heating to four dimensional assimilation of surface and radar data. The results of these simulations underscore the need to accurately initialize and simulate grid and sub-grid scale clouds in meso- beta scale models. The status of the application of the CSU-RAMS mesoscale model by the NOAA Forecast Systems Lab for producing real-time forecasts with 10-60 km mesh resolutions over (4000 km)(exp 2) domains for use by the aviation community was reported. Either MAPS or LAPS model data are used to initialize the RAMS model on a 12-h cycle. The use of MAPS (Mesoscale Analysis and Prediction System) model was discussed. Also discussed was the mesobeta-scale data assimilation using a triply-nested nonhydrostatic version of the MM5 model.

Carr, Fred↗

Which strategy for a protein crystallization project?

The three-dimensional, atomic-resolution protein structures produced by X-ray crystallography over the past 50+ years have led to tremendous chemical understanding of fundamental biochemical processes. The pace of discovery in protein crystallography has increased greatly with advances in molecular biology, crystallization techniques, cryocrystallography, area detectors, synchrotrons and computing. While the methods used to produce single, well-ordered crystals have also evolved over the years in response to increased understanding and advancing technology, crystallization strategies continue to be rooted in trial-and-error approaches. This review summarizes the current approaches in protein crystallization and surveys the first results to emerge from the structural genomics efforts.

Review↗

Instrumentation and First Results of the Reflected Solar Demonstration System for the Climate Absolute Radiance and Refractivity Observatory

The Climate Absolute Radiance and Refractivity Observatory (CLARREO) mission key goals include enabling observation of high accuracy long-term climate change trends, use of these observations to test and improve climate forecasts, and calibration of operational and research sensors. The spaceborne instrument suites include a reflected solar spectroradiometer, emitted infrared spectroradiometer, and radio occultation receivers. The requirement for the RS instrument is that derived reflectance must be traceable to Sl standards with an absolute uncertainty of <0.3% and the error budget that achieves this requirement is described in previo1L5 work. This work describes the Solar/Lunar Absolute Reflectance Imaging Spectroradiometer (SOLARIS), a calibration demonstration system for RS instrument, and presents initial calibration and characterization methods and results. SOLARIS is an Offner spectrometer with two separate focal planes each with its own entrance aperture and grating covering spectral ranges of 320-640, 600-2300 nm over a full field-of-view of 10 degrees with 0.27 milliradian sampling. Results from laboratory measurements including use of integrating spheres, transfer radiometers and spectral standards combined with field-based solar and lunar acquisitions are presented. These results will be used to assess the accuracy and repeatability of the radiometric and spectral characteristics of SOLARIS, which will be presented against the sensor-level requirements addressed in the CLARREO RS instrument error budget.

McCorkel, Joel↗

Combining Galaxy-Galaxy Lensing and Galaxy Clustering

Combining galaxy-galaxy lensing and galaxy clustering is a promising method for inferring the growth rate of large scale structure, a quantity that will shed light on the mechanism driving the acceleration of the Universe. The Dark Energy Survey (DES) is a prime candidate for such an analysis, with its measurements of both the distribution of galaxies on the sky and the tangential shears of background galaxies induced by these foreground lenses. By constructing an end-to-end analysis that combines large-scale galaxy clustering and small-scale galaxy-galaxy lensing, we also forecast the potential of a combined probes analysis on DES datasets. In particular, we develop a practical approach to a DES combined probes analysis by jointly modeling the assumptions and systematics affecting the different components of the data vector, employing a shared halo model, HOD parametrization, photometric redshift errors, and shear measurement errors. Furthermore, we study the effect of external priors on different subsets of these parameters. We conclude that DES data will provide powerful constraints on the evolution of structure growth in the universe, conservatively/ optimistically constraining the growth function to 8%/4.9% with its first-year data covering 1000 square degrees, and to 4%/2.3% with its full five-year data covering 5000 square degrees.

Park, Youngsoo↗

Investigation of boundary-layer wind predictions during nocturnal low-level jet events using the Weather Research and Forecasting model

Simulations of two periods featuring three consecutive low level jet (LLJ) events in the US Upper Great Plains during the autumn of 2011 were conducted to explore the impacts of various setup configurations and physical process models on simulated flow parameters within the lowest 200 m above the surface, using the Weather Research and Forecasting (WRF) model. Sensitivities of simulated flow parameters to the horizontal and vertical grid spacing, planetary boundary layer (PBL) and land surface model (LSM) physics options, were assessed. Data from a Light Detection and Ranging (lidar) system, deployed to the Weather Forecast Improvement Project (WFIP; Finley et al. 2013) were used to evaluate the accuracy of simulated wind speed and direction at 80 m above the surface, as well as their vertical distributions between 120 and 40 m, covering the typical span of contemporary tall wind turbines. All of the simulations qualitatively captured the overall diurnal cycle of wind speed and stratification, producing LLJs during each overnight period, however large discrepancies occurred at certain times for each simulation in relation to the observations. 54-member ensembles encompassing changes of the above discussed configuration parameters displayed a wide range of simulated vertical distributions of wind speed and direction, and potential temperature, reflecting highly variable representations of stratification during the weakly stable overnight conditions. Root mean square error (RMSE) statistics show that different ensemble members performed better and worse in various simulated parameters at different times, with no clearly superior configuration . Simulations using a PBL parameterization designed specifically for the stable conditions investigated herein provided superior overall simulations of wind speed at 80 m, demonstrating the efficacy of targeting improvements of physical process models in areas of known deficiencies. However, the considerable magnitudes of the RMSE values of even the best performing simulations indicate ample opportunities for further improvements.

Mirocha, Jeff D.↗

Tension between SNeIa and BAO: current status and future forecasts

Using real and synthetic Type Ia SNe (SNeIa) and baryon acoustic oscillations (BAO) data representing current observations forecasts, this paper investigates the tension between those probes in the dark energy equation of state (EoS) reconstruction considering the well known CPL model and Wang's low correlation reformulation. In particular, here we present simulations of BAO data from both the the radial and transverse directions. We also explore the influence of priors on Ω{sub m} and Ω{sub b} on the tension issue, by considering 1σ deviations in either one or both of them. Our results indicate that for some priors there is no tension between a single dataset (either SNeIa or BAO) and their combination (SNeIa+BAO). Our criterion to discern the existence of tension (σ-distance) is also useful to establish which is the dataset with most constraining power; in this respect SNeIa and BAO data switch roles when current and future data are considered, as forecasts predict and spectacular quality improvement on BAO data. We also find that the results on the tension are blind to the way the CPL model is addressed: there is a perfect match between the original formulation and that by the correlation optimized proposed in Wang (2008), but the errors on the parameters are much narrower in all cases of our exhaustive exploration, thus serving the purpose of stressing the convenience of this reparametrization.

79 ASTRONOMY AND ASTROPHYSICS↗

Improving Assimilated Global Climate Data Using TRMM and SSM/I Rainfall and Moisture Data

Current global analyses contain significant errors in primary hydrological fields such as precipitation, evaporation, and related cloud and moisture in the tropics. Work has been underway at NASA's Data Assimilation Office to explore the use of TRMM and SSM/I-derived rainfall and total precipitable water (TPW) data in global data assimilation to directly constrain these hydrological parameters. We found that assimilating these data types improves not only the precipitation and moisture estimates but also key climate parameters directly linked to convection such as the outgoing longwave radiation, clouds, and the large-scale circulation in the tropics. We will present results showing that assimilating TRMM and SSM/I 6-hour averaged rain rates and TPW estimates significantly reduces the state-dependent systematic errors in assimilated products. Specifically, rainfall assimilation improves cloud and latent heating distributions, which, in turn, improves the cloudy-sky radiation and the large-scale circulation, while TPW assimilation reduces moisture biases to improve radiation in clear-sky regions. Rainfall and TPW assimilation also improves tropical forecasts beyond 1 day.

Hou, Arthur Y.↗

TRMM Data Assimilation at NASA

Current global analyses contain significant errors in primary hydrological fields such as precipitation, evaporation, and related cloud and moisture in the tropics. Work has been underway at NASA's Data Assimilation Office to explore the use of Tropical Rainfall Measuring Mission (TRMM) and Special Sensor Microwave Imager (SSM/I)-derived rainfall and total precipitable water (TPW) data in global data assimilation to directly constrain these hydrological parameters. We found that assimilating these data types improves not only the precipitation and moisture estimates but also key climate parameters directly linked to convection such as the outgoing longwave radiation, clouds, and the large-scale circulation in the tropics. We will present results showing that assimilating TRMM and SSM/I 6-hour averaged rain rates and TPW estimates significantly reduces the state-dependent systematic errors in assimilated products. Specifically, rainfall assimilation improves cloud and latent heating distributions, which, in turn, improves the cloudy-sky radiation and the large-scale circulation, while TPW assimilation reduces moisture biases to improve radiation in clear-sky regions. Rainfall and TPW assimilation also improves tropical forecasts beyond I day.

Hou, Arthur Y.↗

Resiliency in numerical algorithm design for extreme scale simulations

Here this work is based on the seminar titled ‘Resiliency in Numerical Algorithm Design for Extreme Scale Simulations’ held March 1–6, 2020, at Schloss Dagstuhl, that was attended by all the authors. Advanced supercomputing is characterized by very high computation speeds at the cost of involving an enormous amount of resources and costs. A typical large-scale computation running for 48 h on a system consuming 20 MW, as predicted for exascale systems, would consume a million kWh, corresponding to about 100k Euro in energy cost for executing 10 23 floating-point operations. It is clearly unacceptable to lose the whole computation if any of the several million parallel processes fails during the execution. Moreover, if a single operation suffers from a bit-flip error, should the whole computation be declared invalid? What about the notion of reproducibility itself: should this core paradigm of science be revised and refined for results that are obtained by large-scale simulation? Naive versions of conventional resilience techniques will not scale to the exascale regime: with a main memory footprint of tens of Petabytes, synchronously writing checkpoint data all the way to background storage at frequent intervals will create intolerable overheads in runtime and energy consumption. Forecasts show that the mean time between failures could be lower than the time to recover from such a checkpoint, so that large calculations at scale might not make any progress if robust alternatives are not investigated. More advanced resilience techniques must be devised. The key may lie in exploiting both advanced system features as well as specific application knowledge. Research will face two essential questions: (1) what are the reliability requirements for a particular computation and (2) how do we best design the algorithms and software to meet these requirements? While the analysis of use cases can help understand the particular reliability requirements, the construction of remedies is currently wide open. One avenue would be to refine and improve on system- or application-level checkpointing and rollback strategies in the case an error is detected. Developers might use fault notification interfaces and flexible runtime systems to respond to node failures in an application-dependent fashion. Novel numerical algorithms or more stochastic computational approaches may be required to meet accuracy requirements in the face of undetectable soft errors. These ideas constituted an essential topic of the seminar. The goal of this Dagstuhl Seminar was to bring together a diverse group of scientists with expertise in exascale computing to discuss novel ways to make applications resilient against detected and undetected faults. In particular, participants explored the role that algorithms and applications play in the holistic approach needed to tackle this challenge. This article gathers a broad range of perspectives on the role of algorithms, applications and systems in achieving resilience for extreme scale simulations. The ultimate goal is to spark novel ideas and encourage the development of concrete solutions for achieving such resilience holistically.

79 ASTRONOMY AND ASTROPHYSICS↗