Search NASA⌕ Search

SEARCH · Search NASA

Results for “extreme-value statistics”

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.

11 records
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

Extreme-value statistics in nonlinear optics

We show that, although nonlinear optics may give rise to a vast multitude of statistics, all these statistics converge, in their extreme-value limit, to one of a few universal extreme-value statistics. Specifically, in the class of polynomial nonlinearities, such as those found in the Kerr effect, weak-field harmonic generation, and multiphoton ionization, the statistics of the nonlinear-optical output converges, in the extreme-value limit, to the exponentially tailed, Gumbel distribution. Exponentially growing nonlinear signals, on the other hand, such as those induced by parametric instabilities and stimulated scattering, are shown to reach their extreme-value limits in the class of the Fréchet statistics, giving rise to extreme-value distributions (EVDs) with heavy, manifestly nonexponential tails, thus favoring extreme-event outcomes and rogue-wave buildup.

Zheltikov, Aleksei M. (ORCID:0000000291380576)↗
DOE OSTI2017

Extreme-value statistics reveal rare failure-critical defects in additive manufacturing

Additive manufacturing enables the rapid, cost effective production of large populations of material test coupons such as tensile bars. By adopting streamlined test methods including ‘drop-in’ grips and non-contact extensometry, testing these large populations becomes more efficient. Unlike hardness tests, the tensile test provides a direct measure of yield strength, flow properties, and ductility, which can be directly incorporated into solid mechanics simulations. In the present work, over 1000 nominally identical tensile tests were used to explore the effect of process variability on the mechanical property distributions of a precipitation hardened stainless steel, 17-4PH, produced by a laser powder bed fusion process, also known as direct metal laser sintering. With this large dataset, rare defects are revealed that affect only ~2% of the population, stemming from a single build lot of material. Lastly, the rare defects caused a substantial loss in ductility and were associated with an interconnected network of porosity.

36 MATERIALS SCIENCE↗
DOE OSTI2018

Nanotube/Nanowire as Effective Defect Sinks in Metals—Atomistic Simulations and In Situ Ion Radiation Transmission Electron Microscopy

The accumulation of defects during irradiation leads to material property degradation modes such as embrittlement and swelling, eventually causing material failure. Effective and efficient removal of defects is of crucial importance to design radiation damage-tolerant materials. Here, by biasing defect migration pathways via carbon nanotube (CNT) infiltration, we present a greatly enhanced damage-tolerant Al-CNT composite with defect storage measured to be one order of magnitude lower than that in pure, irradiated Al. Furthermore, extreme-value statistics (largest size) of defect clusters are significantly changed in the presence of CNT. In situ ion irradiation transmission electron microscopy (TEM) experiments and atomistic simulations together reveal the dynamic evolution and convergent diffusion of radiation-induced defects to CNTs, facilitating defect recombination and enhancing radiation tolerance. The occurrence of CNT-biased defect convergent migration is tuned by the thermodynamic driving force of stress gradient in Al matrix due to the CNT phase transformation. This approach to controlling defect migration using 1D interface engineering creates a new opportunity to enhance the properties of nuclear materials.

77 NANOSCIENCE AND NANOTECHNOLOGY↗
DOE OSTI2015

Scaling exponents for ordered maxima

We study extreme value statistics of multiple sequences of random variables. For each sequence with N variables, independently drawn from the same distribution, the running maximum is defined as the largest variable to date. We compare the running maxima of m independent sequences and investigate the probability S N that the maxima are perfectly ordered, that is, the running maximum of the first sequence is always larger than that of the second sequence, which is always larger than the running maximum of the third sequence, and so on. The probability S N is universal: it does not depend on the distribution from which the random variables are drawn. For two sequences, S N ~N –1/2 , and in general, the decay is algebraic, S N ~N –σm , for large N. We analytically obtain the exponent σ 3 ≅1.302931 as root of a transcendental equation. Moreover, the exponents σ m grow with m, and we show that σ m ~m for large m.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2026

Multitask graph neural networks for elastoplastic response prediction in dual-phase polycrystals

Microstructure-sensitive prediction of elastoplastic response remains a recurring bottleneck in multiscale damage and fatigue modeling, where large ensembles of statistically distinct polycrystals are required to quantify variability and extreme-value behavior. In this work, we develop a multitask graph neural network (GNN) surrogate that maps dual-phase ferrite–martensite polycrystal microstructures to Statistical Volume Element (SVE)-level elastoplastic Quantities of Interest (QoIs). Each SVE is represented as a grain-adjacency graph, with node features encoding phase, geometry, and crystallographic orientation, and edge features encoding relative misorientation. A message-passing graph convolution generates node embeddings, which are pooled into a graph representation and passed to a multitask regression head that jointly predicts 10 scalar QoIs and vector-valued stress–strain responses in orthogonal loading directions across multiple martensite volume fractions and SVE sizes. Results show high accuracy for scalar QoIs and strong agreement for full stress–strain trajectories, with population envelopes reproducing both median behavior and finite-SVE variability across compositions and partition scales. A unified model trained on pooled volume-fraction data preserves most within-regime accuracy relative to regime-specific models while also capturing the broader cross-regime variation reflected in the pooled test set. Distributional comparisons further demonstrate that the surrogate preserves heterogeneity under SVE partitioning, enabling statistically consistent block-wise random-field construction for mesoscale analyses. Overall, the proposed grain-graph surrogate provides a practical pathway to accelerate ensemble-based studies of SVE-level constitutive variability in dual-phase polycrystals.

Crystal plasticity↗
DOE OSTI2017

Characterization of Inclusion Populations in Mn-Si Deoxidized Steel

Four plant heats of Mn-Si deoxidized steel were conducted to follow the evolution of the inclusion population through ladle furnace (LF) treatment and subsequent vacuum treatment (VT). The liquid steel was sampled, and the chemical composition and size distribution of the inclusion populations were characterized. The Gumbel generalized extreme-value (GEV) and generalized Pareto (GP) distributions were used for the statistical analysis of the inclusion size distributions. The inclusions found at the beginning of the LF treatment were mostly fully liquid SiO{sub 2}-Al{sub 2}O{sub 3}-MnO inclusions, which then evolved into fully liquid SiO{sub 2}-Al{sub 2}O{sub 3}-CaO-MgO and partly liquid SiO{sub 2}-CaO-MgO-(Al{sub 2}O{sub 3}-MgO) inclusions detected at the end of the VT. The final fully liquid inclusions had a desirable chemical composition for plastic behavior in subsequent metallurgical operations. The GP distribution was found to be undesirable for statistical analysis. The GEV distribution approach led to shape parameter values different from the zero value hypothesized from the Gumbel distribution. According to the GEV approach, some of the final inclusion size distributions had statistically significant differences, whereas the Gumbel approach predicted no statistically significant differences. The heats were organized according to indicators of inclusion cleanliness and a statistical comparison of the size distributions.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Uniformly Ordered Binary Decision Algorithm for Benchmark Experiment Correlations in Whisper Validation

When performing a validation exercise for determining the upper subcritical limit of a nuclear criticality safety application, an analyst should select and perform a statistical analysis on a population of benchmark experiments that are neutronically similar to the application. The size of this population should be sufficiently large such that the statistical analysis has a high degree of confidence that the bias plus bias uncertainty (calculational margin) has been accurately quantified. A complication arises because many benchmark experiments share common components, leading to correlations in their measured effective multiplication factors. Correlations between benchmark experiments within the population reduces its predictive power. This motivates the need for methods that consider benchmark experiment correlations and ensure adequate statistical significance of results. The Whisper code is a statistical analysis pack- age that incorporates nuclear data sensitivity coefficients from MCNP to assess benchmark experiment similarity and then performs an extreme-value analysis to estimate the bias plus bias uncertainty. The original methodology in Whisper does not consider the effect of benchmark experiment correlations when making this estimation, and this summary proposes the uniformly ordered binary decision algorithm to address this shortcoming. The original methodology in Whisper computes similarity coefficients ck for an application compared to all benchmark experiments in its library and develops weighting factors for a selected population proportional to the ck values. The methodology can be interpreted as statistically emulating a validation exercise for a particular application where the weighting factors may be viewed as the likelihood that an analyst would include a particular benchmark experiment within the population. The effective sample size of the population is the expected or mean number of benchmark experiments in the population. The uniformly ordered binary decision algorithm identifies clusters of correlated benchmark experiments within the population and then computes adjusted weighting factors based on the magnitude of the correlation coefficients within the cluster to compute a reduced effective sample size accounting for the lower information content because of correlations. Benchmark experiments within the cluster are ordered randomly with equal probability and probabilistic decisions are made as to whether a benchmark. experiment within the cluster should treated as redundant with a previous one; if two redundant benchmark experiments are included, then the conservative worst case bias plus bias uncertainty is used and the pair is counted as a single benchmark experiment in the population. Results are provided for HEU solutions in a research version of the Whisper software using benchmark experiment correlations provided by DICE, the Database for the International Criticality Safety Benchmark Evaluation Project (ICSBEP). These show that there can be a significant increase in the bias plus bias uncertainty because the effective sample size is reduced, and therefore the algorithm, needing to meet sample size requirements, expands the benchmark experiment population by accepting less similar benchmark experiments that would have otherwise not been included.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
DOE OSTI2016

Methane Leaks from Natural Gas Systems Follow Extreme Distributions

Future energy systems may rely on natural gas as a low-cost fuel to support variable renewable power. However, leaking natural gas causes climate damage because methane (CH 4 ) has a high global warming potential. In this study, we use extreme-value theory to explore the distribution of natural gas leak sizes. By analyzing ~15,000 measurements from 18 prior studies, we show that all available natural gas leakage datasets are statistically heavy-tailed, and that gas leaks are more extremely distributed than other natural and social phenomena. A unifying result is that the largest 5% of leaks typically contribute over 50% of the total leakage volume. While prior studies used lognormal model distributions, we show that lognormal functions poorly represent tail behavior. Our results suggest that published uncertainty ranges of CH 4 emissions are too narrow, and that larger sample sizes are required in future studies to achieve targeted confidence intervals. Additionally, we find that cross-study aggregation of datasets to increase sample size is not recommended due to apparent deviation between sampled populations. Finally, understanding the nature of leak distributions can improve emission estimates, better illustrate their uncertainty, allow prioritization of source categories, and improve sampling design. Also, these data can be used for more effective design of leak detection technologies.

03 NATURAL GAS↗
DOE OSTI2017

A practical method to test the validity of the standard Gumbel distribution in logit-based multinomial choice models of travel behavior

Most multinomial choice models (e.g., the multinomial logit model) adopted in practice assume an extreme-value Gumbel distribution for the random components (error terms) of utility functions. This distributional assumption offers a closed-form likelihood expression when the utility maximization principle is applied to model choice behaviors. As a result, model coefficients can be easily estimated using the standard maximum likelihood estimation method. However, maximum likelihood estimators are consistent and efficient only if distributional assumptions on the random error terms are valid. It is therefore critical to test the validity of underlying distributional assumptions on the error terms that form the basis of parameter estimation and policy evaluation. In this paper, a practical yet statistically rigorous method is proposed to test the validity of the distributional assumption on the random components of utility functions in both the multinomial logit (MNL) model and multiple discrete-continuous extreme value (MDCEV) model. Based on a semi-nonparametric approach, a closed-form likelihood function that nests the MNL or MDCEV model being tested is derived. The proposed method allows traditional likelihood ratio tests to be used to test violations of the standard Gumbel distribution assumption. Simulation experiments are conducted to demonstrate that the proposed test yields acceptable Type-I and Type-II error probabilities at commonly available sample sizes. The test is then applied to three real-world discrete and discrete-continuous choice models. For all three models, the proposed test rejects the validity of the standard Gumbel distribution in most utility functions, calling for the development of robust choice models that overcome adverse effects of violations of distributional assumptions on the error terms in random utility functions.

30 DIRECT ENERGY CONVERSION↗
DOE OSTI2021

Quantitative Precipitation Estimation of Extremes in CONUS With Radar Data

When constructing an accurate, continental, in-situ-based, kilometer-scale, long-term record of the precipitation field and its spatiotemporal changes remains a significant challenge. Here, we determine the extreme-value behavior of the NEXRAD Stage IV radar-based quantitative precipitation estimate. We find that the climatology of 5-year daily return values in the contiguous United States East of the Rocky Mountains shows only slight variability on spatial scales smaller than (Formula presented.) ~100 km. In light of this finding, we test whether rain-gauge-only daily precipitation data sets can produce accurate extreme-value behavior at spatial scales finer than the spacing between gauges. We find that the 5-year daily return values are accurate at locations far from rain gauges only if the interpolation between gauges is carried out appropriately for extremes. Precipitation statistics derived from in-situ rain gauge data are therefore of sufficient spatial resolution to faithfully capture daily extremes over much of the eastern United States.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2025

Probabilistic Hazard Assessment for Tornadoes, Straight-Line Wind, and Extreme Precipitation at the Savannah River Site

Recent data sets for three meteorological phenomena with the potential to inflict damage on SRS facilities – tornadoes, straight-line winds, and heavy precipitation – are analyzed using appropriate statistical techniques to estimate the occurrence probabilities for these events in the future. Summaries of the results for DOE-mandated return periods and comparisons to similar calculations performed in 2013 by Werth et al. (W2013) are given. Using tornado statistics for i) the combined states of Georgia and South Carolina, and ii) a 2⁰ square area surrounding SRS, we calculated the probability per year of any location at SRS being struck by a tornado (the ‘strike’ probability) and the probability that any point will experience winds above set thresholds. The strike probability was calculated to be 7.04E-4 (1 chance in 1420) per year and tornadic wind speeds for DOE mandated return periods of 50,000 years (corresponding to wind design category 3 (WDC-3), and 125,000 years (meeting WDC-4) (USDOE, 2016) were estimated to be 132 mph and 147 mph, respectively. By contrast, default tornado wind speeds taken from ANSI/ANS-2.3-2011 are somewhat higher: 161 mph for return periods of 50,000 years and 173 mph every 125,000 years (ANS, 2011). Although the ANS and the SRS evaluation used the same basic model (Ramsdell and Rishel, 2007), the region defined in ANS 2.3 that encompasses the SRS also includes areas of the Great Plains and lower Midwest, regions with much higher occurrence frequencies of strong tornadoes. The SRS straight-line wind values associated with various return periods were calculated by fitting existing wind data to a GEV1 distribution and extrapolating the values for any return period from the tail of that function. For the DOE mandated return periods, we expect straight-line winds of 117 mph every 2500 years (the required WDC-3 standard) and 125 mph every 6250 years (WDC-4) at any point within the SRS. These values are similar to those from the ANS-2.3-2011 report, which has wind speeds of 125mph and 133 mph for return periods of 2500 years and 6250 years, respectively. For extreme precipitation, we compared the fits of two different theoretical extreme-value distributions and applied the one that fit the data best for each of several accumulation periods. The DOE mandated 6-hr accumulated rainfall for return periods of 10,000 years (corresponding to precipitation design category 3 (PDC-3) and 25,000 years (PDC-4) were estimated as 9.1 inches and 10.1 inches, respectively. For the 24-hr rainfall return periods of 10,000 years and 25,000 years, total rainfall estimates were 12.02 inches and 13.17 inches, respectively, higher than comparable values provided in the W2013 report.

54 ENVIRONMENTAL SCIENCES↗