Results for “Statistical Testing Approach” — Page 60

DOE OSTI2019

A fast Fourier transform-based mesoscale field dislocation mechanics study of grain size effects and reversible plasticity in polycrystals

A numerical implementation of a non-local polycrystal plasticity theory based on a mesoscale version of the field dislocation mechanics theory (MFDM) of Acharya and Roy (2006) is presented using small-strain elasto-viscoplastic fast Fourier transform-based (EVPFFT) algorithm developed by Lebensohn et al. (2012). In addition to considering plastic flow and hardening only due to SSDs (statistically stored dislocations) as in the classic EVPFFT framework, the proposed method accounts for the evolution of GND (geometrically necessary dislocations) densities solving a hyperbolic-type partial differential equation, and GND effects on both plastic flow and hardening. This allows consideration of an enhanced strain-hardening law that includes the effect of the GND density tensor. The numerical implementation of a reduced version of the MFDM is presented in the framework of the FFT-based augmented Lagrangian procedure of Michel et al. (2001). A Finite Differences scheme combined with discrete Fourier transforms is applied to solve both incompatibility and equilibrium equations. The numerical procedure named MFDM-EVPFFT is used to perform full field simulations of polycrystal plasticity considering different grain sizes and their mechanical responses during monotonic tensile and reversible tension-compression tests. Using Voronoi tessellation and periodic boundary conditions, voxelized representative volume elements (RVEs) with different grain sizes are generated. With MFDM-EVPFFT, a Hall-Petch type scaling law is obtained in contrast with the conventional crystal plasticity EVPFFT. In the case of reversible plasticity, a stronger Bauschinger effect is observed with the MFDM-EVPFFT approach in comparison with conventional EVPFFT. Here, the origin of these differences is analyzed in terms of heterogeneity, GND density and stress evolutions during the compression stage.

36 MATERIALS SCIENCE↗

DOE OSTI2022

Reliability Evaluation of Bifacial and Monofacial Glass/Glass Modules with EVA and non-EVA Encapsulants

The market share for bifacial modules is projected to be doubled from 30% to 60% in the next ten years. For the monofacial crystalline silicon glass/backsheet (G/B) modules, extensive reliability data has been available for over 40 years. However, practically little/no long-term field or accelerated reliability test data is available for the new generation glass/glass (G/G) bifacial modules and glass/transparent (G/T) bifacial modules. Therefore, the primary motivation of this project was to identify the reliability strengths and weaknesses of new generation G/G modules compared to G/B modules. In this 3-year project, the goal was to objectively recommend the best construction materials for the new generation G/G (bifacial) modules through a systematic experimental approach with appropriate tasks including: Evaluation of field retrieved old-generation G/G modules; Inspection of new-generation G/G modules installed in the plants; Evaluation of new-generation G/G, G/B and G/T modules using EAST (Extended Accelerated Stress Testing), CAST (Combined Accelerated Stress Testing) and FAST (Field Accelerated Stress Testing); and, dynamic literature search and review. Major accomplishments and outcomes of this project are: Accomplishments: Evaluated more than 60 field retrieved modules and constructed and characterized more than 135 mini-modules with three substrate types (G, B, and T), two encapsulant types (EVA and POE), and two cell types (monofacial and bifacial) as well as evaluated more than 30 commercial G/G and G/B modules; Subjected all the modules to multitude characterization tests and various indoor and outdoor accelerated stress tests including FAST (Field Accelerated Stress Testing), EAST (Extended Accelerated Stress Testing), and CAST (Combined Accelerated Stress Testing) to identify and correlate the failure modes in both field and lab tests. Outcomes: Based on the accelerated test results and handling/mounting experience obtained in this project, the recommended best construction for the glass/glass modules is: “Framed GG modules with cut-cells and POE encapsulant (UVpass front; UVpass back).” However, from a statistical and manufacturing point of view (along with stakeholders surveys), the following cautionary notes are added to the above-mentioned recommendation: (i) Cut cells could introduce a higher level of manufacturing issues, including a higher number of cell interconnects; (ii) POE encapsulant is more expensive than EVA and could present a delamination risk and lower throughput during manufacturing due to lower adhesion strength. Other encapsulants, such as coextruded EPE (EVA/POE/EVA), are also recommended to be investigated. The potential public benefit of the proposed project is to present the strengths and weaknesses of glass/glass modules, so an informed procurement decision can be made.

14 SOLAR ENERGY↗

DOE OSTI2016

Investigation of heat transfer in liquid-metal flows under fusion-reactor conditions

The effect discovered in studying a downward liquid-metal flow in vertical pipe and in a channel of rectangular cross section in, respectively, a transverse and a coplanar magnetic field is analyzed. In test blanket modules (TBM), which are prototypes of a blanket for a demonstration fusion reactor (DEMO) and which are intended for experimental investigations at the International Thermonuclear Experimental Reactor (ITER), liquid metals are assumed to fulfil simultaneously the functions of (i) a tritium breeder, (ii) a coolant, and (iii) neutron moderator and multiplier. This approach to testing experimentally design solutions is motivated by plans to employ, in the majority of the currently developed DEMO blanket projects, liquid metals pumped through pipes and/or rectangular channels in a transvers magnetic field. At the present time, experiments that would directly simulate liquid-metal flows under conditions of ITER TBM and/or DEMO blanket operation (irradiation with thermonuclear neutrons, a cyclic temperature regime, and a magnetic-field strength of about 4 to 10 T) are not implementable for want of equipment that could reproduce simultaneously the aforementioned effects exerted by thermonuclear plasmas. This is the reason why use is made of an iterative approach to experimentally estimating the performance of design solutions for liquid-metal channels via simulating one or simultaneously two of the aforementioned factors. Therefore, the investigations reported in the present article are of considerable topical interest. The respective experiments were performed on the basis of the mercury magneto hydrodynamic (MHD) loop that is included in the structure of the MPEI—JIHT MHD experimental facility. Temperature fields were measured under conditions of two- and one-sided heating, and data on averaged-temperature fields, distributions of the wall temperature, and statistical fluctuation features were obtained. A substantial effect of counter thermo gravitational convection (TGC) on averaged and fluctuating quantities were found. The development of TGC in the presence of a magnetic field leads to the appearance of low-frequency fluctuations whose anomalously high intensity exceeds severalfold the level of turbulence fluctuations. This effect manifest itself over a broad region of regime parameters. It was confirmed that low-energy fluctuations penetrate readily through the wall; therefore, it is necessary to study this effect further—in particular, from the point of view of the fatigue strength of the walls of liquid-metal channels.

42 ENGINEERING↗

DOE OSTI2025

AI Model Benchmarking for Nonproliferation Applications: Steel Thread Benchmarking Task Force Technical Report (Rev. 2)

Steel Thread is a NA-22 venture that seeks to build trustworthy, reliable AI models that can be used in a wide variety of nonproliferation tasks. A key aspect of building these models is developing appropriate benchmarks and evaluation methods, which will enable the venture to identify and adapt models to provide the most value in the nonproliferation domain. Benchmarks must be relevant to key tasks in this domain, such as question answering, information retrieval, document summarization and classification, consensus analysis, and image and data analysis. This report 1) provides an overview of benchmark design, evaluation, and challenges; 2) reviews a variety of open benchmarks, with a focus on language models and tasks; and 3) identifies benchmarks that are most relevant to Steel Thread. This report is intended to serve as a basis for further efforts to classify and evaluate benchmarks and their correlation with success on nonproliferation-specific tasks. The Steel Thread venture has defined benchmarks to be a particular combination of a dataset (or datasets) and a metric (or metrics) conceptualized as representing one or more specific tasks or sets of abilities for a specific modality. It is adopted by a research community as a shared framework for comparing methods.1 It includes 1) Data: Labeled (a designated subset not used for training, which could be all the data), 2) Metric: A way to quantify performance, 3) Task/Ability: What the benchmark is testing, 4) Protocol: A structured and repeatable evaluation process, 5) Baseline/Reference Model: For comparison; could be statistical, rule-based, SME-derived, or another model, and 6) Maintenance Plan: to update with new information over time; important for long-term utility. For further clarity, the definition includes what a benchmark, in this context, is not. It is not a corpus of training data, specific to a model (it is intended to apply to a range of models), a universal evaluation of performance, a guarantee that the ‘top’ model on the leaderboard will be the best fit for every specific use case, an all-encompassing proof of a model’s universal quality, nor is it a one-size-fits-all measure of success. It does not cover every real-world constraint (like operational, ethical, or cost considerations), a systems integration test, or a unit test. This definition was inspired by and resulted from discussions within the Steel Thread Benchmarking Task Force. This group was formed to define what we would mean as a benchmark within Steel Thread but persisted as the need to develop a thorough understanding of the large and expanding existing benchmarking space. This technical report is a result of the group’s divide and conquer approach to exploring this space. The release of benchmarks might not be progressing as quickly as model development, but it is moving very fast, as many benchmarks quickly become saturated, when state-of-the-art models score so close to the benchmark’s ceiling that their results are virtually indistinguishable. At that point, the test no longer differentiates between new systems, so researchers usually stop reporting scores as the benchmark no longer informs about improvements from the next generation of models. In the OpenAI announcement of GPT-5, they reported results on six flagship public benchmarks (AIME 2025, SWE-bench Verified, Aider Polyglot, MMMU, HealthBench Hard, GPQA) but the full system-card covers roughly thirty-five separate evaluations, comprising hundreds of test task items in total. There have been some efforts to summarize benchmarks in specific fields, like for text-to-image generation, but these surveys have had a narrow methodology scope. Therefore, a comprehensive survey of all benchmarks or even all benchmarks that could be relevant to Steel Thread is outside of the scope of this report. We chose some specific benchmarks to investigate in detail.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2021

An evaluation of prognostic factors, oncologic outcomes, and management for primary and recurrent squamous cell carcinoma of the vulva

Objective: To evaluate prognostic factors, outcomes, and management patterns of patients treated for squamous cell carcinoma of the vulva. Methods: One hundred sixty-four women were retrospectively identified with primary squamous cell carcinoma of the vulva treated at our institution between 1/1996–12/2018. Descriptive statistics were performed on patient, tumor, and treatment characteristics. The χ 2 tests and t-tests were used to compare categorical variables and continuous variables, respectively. Recurrence free survival (RFS), overall survival (OS), and disease-specific survival (DSS) were analyzed with Kaplan-Meier estimates, the log-rank test, and Cox proportional hazards. Results: Median follow-up was 52.5 months. Five-year RFS was 67.9%, 60.0%, 42.1%, and 20.0% for stage I–IV, respectively. Five-year DSS was 86.2%, 81.6%, 65.0%, and 42.9% for stage I–IV, respectively. On multivariate analysis, positive margins predicted overall RFS (hazard ratio [HR]=3.55; 95% confidence interval [CI]=1.18–10.73; p=0.025), while presence of lichen sclerosus on pathology (HR=2.78; 95% CI=1.30–5.91; p=0.008) predicted local RFS. OS was predicted by nodal involvement (HR=2.51; 95% CI=1.02–6.13; p=0.043) and positive margins (HR=5.19; 95% CI=2.03–13.26; p=0.001). Adjuvant radiotherapy significantly improved RFS (p=0.016) and DSS (p=0.012) in node-positive patients. Median survival after treatment of local, groin, and pelvic/distant recurrence was 52, 8, and 5 months, respectively. Conclusion: For primary treatment, more conservative surgical approaches can be considered with escalation of treatment in patients with concurrent precursor lesions, positive margins, and/or nodal involvement. Further studies are warranted to improve risk stratification in order to optimize treatment paradigms for vulvar cancer patients.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2013

EXPLORING THE VARIABLE SKY WITH LINEAR. III. CLASSIFICATION OF PERIODIC LIGHT CURVES

We describe the construction of a highly reliable sample of ∼7000 optically faint periodic variable stars with light curves obtained by the asteroid survey LINEAR across 10,000 deg{sup 2} of the northern sky. The majority of these variables have not been cataloged yet. The sample flux limit is several magnitudes fainter than most other wide-angle surveys; the photometric errors range from ∼0.03 mag at r = 15 to ∼0.20 mag at r = 18. Light curves include on average 250 data points, collected over about a decade. Using Sloan Digital Sky Survey (SDSS) based photometric recalibration of the LINEAR data for about 25 million objects, we selected ∼200,000 most probable candidate variables with r < 17 and visually confirmed and classified ∼7000 periodic variables using phased light curves. The reliability and uniformity of visual classification across eight human classifiers was calibrated and tested using a catalog of variable stars from the SDSS Stripe 82 region and verified using an unsupervised machine learning approach. The resulting sample of periodic LINEAR variables is dominated by 3900 RR Lyrae stars and 2700 eclipsing binary stars of all subtypes and includes small fractions of relatively rare populations such as asymptotic giant branch stars and SX Phoenicis stars. We discuss the distribution of these mostly uncataloged variables in various diagrams constructed with optical-to-infrared SDSS, Two Micron All Sky Survey, and Wide-field Infrared Survey Explorer photometry, and with LINEAR light-curve features. We find that the combination of light-curve features and colors enables classification schemes much more powerful than when colors or light curves are each used separately. An interesting side result is a robust and precise quantitative description of a strong correlation between the light-curve period and color/spectral type for close and contact eclipsing binary stars (β Lyrae and W UMa): as the color-based spectral type varies from K4 to F5, the median period increases from 5.9 hr to 8.8 hr. These large samples of robustly classified variable stars will enable detailed statistical studies of the Galactic structure and physics of binary and other stars and we make these samples publicly available.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2015

Systematic assessment of wellbore integrity for geologic carbon storage projects using regulatory and industry information

Under this three year project, the condition of legacy oil and gas wells in the Midwest United States was evaluated through analysis of well records, well plugging information, CBL evaluation, sustained casing pressure (SCP) field testing, and analysis of hypothetical CO 2 test areas to provide a realistic description of wellbore integrity factors. The research included a state-wide review of oil and gas well records for Ohio and Michigan, along with more detailed testing of wells in Ohio. Results concluded that oil and gas wells are clustered along fields in areas. Well records vary in quality, and there may be wells that have not been identified in records, but there are options for surveying unknown wells. Many of the deep saline formations being considered for CO 2 storage have few wells that penetrate the storage zone or confining layers. Research suggests that a variety of well construction and plugging approaches have been used over time in the region. The project concluded that wellbore integrity is an important issue for CO 2 storage applications in the Midwest United States. Realistic CO 2 storage projects may cover an area in the subsurface with several hundred legacy oil and gas wells. However, closer inspection may often establish that most of the wells do not penetrate the confining layers or storage zone. Therefore, addressing well integrity may be manageable. Field monitoring of SCP also indicated that tested wells provided zonal isolation of the reservoirs they were designed to isolate. Most of these wells appeared to exhibit gas pressure originating from intermediate zones. Based on these results, more flexibility in terms of cementing wells to surface, allowing well testing, and monitoring wells may aid operators in completing CO 2 storage project. Several useful products were developed under this project for examining wellbore integrity for CO 2 storage applications including, a database of over 4 million items on well integrity parameters in the study areas, a systematic CBL evaluation tool for rating cement in boreholes, SCP field testing procedures and analysis methodology, a process for summarizing well integrity at CO 2 storage fields, a statistical analysis of well integrity indicators, and an assessment of practical methods and costs necessary to repair/remediate typical wells in the region based on assessment of six test study areas. Project results may benefit both CO 2 storage and improved oil recovery applications. This study of wellbore integrity is a useful precursor to support development of geologic storage in the Midwest United States because it sheds more light on the actual well conditions (rather than the perceived condition) of historic oil and gas wells in the region.

42 ENGINEERING↗

DOE OSTI2016

Biased deposition of Nb-ON-Cu for SRF accelerators, Final Report

The Phase I project objectives were: Task 1: Design, assemble, and commission a new coating facility that uses the RF cavity as the vacuum chamber Task 3: Coat and RF test biased Cu cavity Task 4: Evaluate test results and prepare Ph-II bid. As for project accomplishments, AASC has completed all of the tasks of Ph-I. However, for Task 2, we found a limit to the bias voltage on a substrate placed at the radius of the beam pipe. A bias voltage higher than -50 V can induce arcing from the substrate to the grounded anode. Arc spots form on the substrate and damage the film. The main problem addressed by this SBIR project is that the price of niobium dominates the cost of superconducting RF cavity production. The most effective route to significant cost reduction is to use cavities that have a thin layer of niobium deposited on a less expensive substrate. However, all attempts to use thin films of niobium in superconducting RF cavities have to date resulted in cavity performance that does not meet the requirements of modern linear accelerators. Alameda Applied Sciences Corporation (AASC) has developed a novel deposition method that can produce high-purity films of niobium. For superconducting radio-frequency (SRF) accelerator cavities, films of niobium on another base material could provide good performance at lower cost than for bulk niobium. The primary goal of all thin film coated approaches must be to demonstrate RF performance that matches that of bulk Nb cavities. This Ph-I was not expected to accomplish this major milestone for thin film coated SRF cavities in just 9 months with limited resources. We set less ambitious goals of demonstrating that using our CED process to coat a biased substrate could result in an increase of RF performance over previous tests with no applied bias voltage. In this Ph-1, we obtained preliminary RF data from Nb coated biased cavity structures so as to move ahead in Ph-II to demonstration of performance that approaches that of bulk Nb. Our task has been made more daunting recently as several labs (Fermilab, Cornell, and JLab) have demonstrated significantly improved performance of bulk Nb cavities using nitrogen doping and rapid cool-down techniques. The bar has just been raised for thin films. At the 2013 SRF thin film workshop held at INFN in Padua Italy, leading scientists from the LHC tossed a gauntlet to the thin film community. The LHC plans a high luminosity upgrade starting in about 4 years. Upper management at CERN has stated that they would prefer not to use bulk Nb for the upgrade, which will require hundreds of new SRF cavities. But the thin film community has a short window in which to demonstrate that a Cu (or Al) cavity that is coated with Nb delivers the kind of performance that we have come to expect from bulk Nb: Q o >10 10 that stays flat up to 20 MV/m or higher, good statistics from multiple cavities and reliable operation over thousands of hours. Breakthroughs are needed to meet these daunting goals. AASC has persisted with this challenge for more than five years now, with the benefit of DOE SBIR funding. We aim to preserve our momentum and build upon our base to try to reach the lofty goals within four more years.

36 MATERIALS SCIENCE↗

NASA NTRS2014

Statistical Mechanics of Turbulent Dynamos

Incompressible magnetohydrodynamic (MHD) turbulence and magnetic dynamos, which occur in magnetofluids with large fluid and magnetic Reynolds numbers, will be discussed. When Reynolds numbers are large and energy decays slowly, the distribution of energy with respect to length scale becomes quasi-stationary and MHD turbulence can be described statistically. In the limit of infinite Reynolds numbers, viscosity and resistivity become zero and if these values are used in the MHD equations ab initio, a model system called ideal MHD turbulence results. This model system is typically confined in simple geometries with some form of homogeneous boundary conditions, allowing for velocity and magnetic field to be represented by orthogonal function expansions. One advantage to this is that the coefficients of the expansions form a set of nonlinearly interacting variables whose behavior can be described by equilibrium statistical mechanics, i.e., by a canonical ensemble theory based on the global invariants (energy, cross helicity and magnetic helicity) of ideal MHD turbulence. Another advantage is that truncated expansions provide a finite dynamical system whose time evolution can be numerically simulated to test the predictions of the associated statistical mechanics. If ensemble predictions are the same as time averages, then the system is said to be ergodic; if not, the system is nonergodic. Although it had been implicitly assumed in the early days of ideal MHD statistical theory development that these finite dynamical systems were ergodic, numerical simulations provided sufficient evidence that they were, in fact, nonergodic. Specifically, while canonical ensemble theory predicted that expansion coefficients would be (i) zero-mean random variables with (ii) energy that decreased with length scale, it was found that although (ii) was correct, (i) was not and the expected ergodicity was broken. The exact cause of this broken ergodicity was explained, after much investigation, by greatly extending the statistical theory of ideal MHD turbulence. The mathematical details of broken ergodicity, in fact, give a quantitative explanation of how coherent structure, dynamic alignment and force-free states appear in turbulent magnetofluids. The relevance of these ideal results to real MHD turbulence occurs because broken ergodicity is most manifest in the ideal case at the largest length scales and it is in these largest scales that a real magnetofluid has the least dissipation, i.e., most closely approaches the behavior of an ideal magnetofluid. Furthermore, the effects grow stronger when cross and magnetic helicities grow large with respect to energy, and this is exactly what occurs with time in a real magnetofluid, where it is called selective decay. The relevance of these results found in ideal MHD turbulence theory to the real world is that they provide at least a qualitative explanation of why confined turbulent magnetofluids, such as the liquid iron that fills the Earth's outer core, produce stationary, large-scale magnetic fields, i.e., the geomagnetic field. These results should also apply to other planets as well as to plasma confinement devices on Earth and in space, and the effects should be manifest if Reynolds numbers are high enough and there is enough time for stationarity to occur, at least approximately. In the presentation, details will be given for both theoretical and numerical results, and references will be provided.

Shebalin, John V.↗

NASA NTRS2014

MHD Turbulence and Magnetic Dynamos

Incompressible magnetohydrodynamic (MHD) turbulence and magnetic dynamos, which occur in magnetofluids with large fluid and magnetic Reynolds numbers, will be discussed. When Reynolds numbers are large and energy decays slowly, the distribution of energy with respect to length scale becomes quasi-stationary and MHD turbulence can be described statistically. In the limit of infinite Reynolds numbers, viscosity and resistivity become zero and if these values are used in the MHD equations ab initio, a model system called ideal MHD turbulence results. This model system is typically confined in simple geometries with some form of homogeneous boundary conditions, allowing for velocity and magnetic field to be represented by orthogonal function expansions. One advantage to this is that the coefficients of the expansions form a set of nonlinearly interacting variables whose behavior can be described by equilibrium statistical mechanics, i.e., by a canonical ensemble theory based on the global invariants (energy, cross helicity and magnetic helicity) of ideal MHD turbulence. Another advantage is that truncated expansions provide a finite dynamical system whose time evolution can be numerically simulated to test the predictions of the associated statistical mechanics. If ensemble predictions are the same as time averages, then the system is said to be ergodic; if not, the system is nonergodic. Although it had been implicitly assumed in the early days of ideal MHD statistical theory development that these finite dynamical systems were ergodic, numerical simulations provided sufficient evidence that they were, in fact, nonergodic. Specifically, while canonical ensemble theory predicted that expansion coefficients would be (i) zero-mean random variables with (ii) energy that decreased with length scale, it was found that although (ii) was correct, (i) was not and the expected ergodicity was broken. The exact cause of this broken ergodicity was explained, after much investigation, by greatly extending the statistical theory of ideal MHD turbulence. The mathematical details of broken ergodicity, in fact, give a quantitative explanation of how coherent structure, dynamic alignment and force-free states appear in turbulent magnetofluids. The relevance of these ideal results to real MHD turbulence occurs because broken ergodicity is most manifest in the ideal case at the largest length scales and it is in these largest scales that a real magnetofluid has the least dissipation, i.e., most closely approaches the behavior of an ideal magnetofluid. Furthermore, the effects grow stronger when cross and magnetic helicities grow large with respect to energy, and this is exactly what occurs with time in a real magnetofluid, where it is called selective decay. The relevance of these results found in ideal MHD turbulence theory to the real world is that they provide at least a qualitative explanation of why confined turbulent magnetofluids, such as the liquid iron that fills the Earth's outer core, produce stationary, large-scale magnetic fields, i.e., the geomagnetic field. These results should also apply to other planets as well as to plasma confinement devices on Earth and in space, and the effects should be manifest if Reynolds numbers are high enough and there is enough time for stationarity to occur, at least approximately. In the presentation, details will be given for both theoretical and numerical results, and references will be provided.

Shebalin, John V↗

DOE OSTI2018

Impact of global atmospheric reanalyses on statistical precipitation downscaling

Statistical downscaling based on a perfect prognosis approach often relies on global reanalyses to infer the statistical relationship between synoptic predictors and the local variable of interest, here daily precipitation. Nowadays, many reanalyses are available and their impact on the downscaled variable is not often considered. The present work assessed the impact of ten reanalyses on the performance of seven variants of analogue methods for statistical precipitation downscaling at 301 stations in Switzerland. Even though the study location is in a data-rich region, significant differences were found between reanalyses and their impact on the performance of the method can be even higher than the choice of the predictor variables. There was no single overall winner, but a selection of recommended reanalyses resulting in higher skill scores depending on the considered predictor variables. The impact of the output spatial resolution was assessed for different types of variables. Output resolutions below one degree were found to be often of low to no interest. Reanalyses with longer archives allow the pool of potential analogues to be increased, resulting in better performance. However, when adding variables affected by errors in a more distant past, the skill score decreased again. The use of multiple members from two reanalyses was also tested over a recent and a past period. The benefit of using members to increase the performance by better incorporating the uncertainties was found to be limited, and even problematic with methods using multiple analogy levels.

Meteorology & Atmospheric Sciences↗

DOE OSTI2023

Predicting the neutrinoless double-β-decay matrix element of 136 Xe using a statistical approach

Calculation of the nuclear matrix elements (NMEs) for double-β decay is of paramount importance for guiding experiments and for analyzing and interpreting the experimental data, especially for the search of the neutrinoless double β decay mode (0v⁢ββ). However, there are currently still large differences between the NME values calculated by different methods, hence a quantification of their uncertainties is very much required. Here, in this paper, we propose a statistical analysis of 0⁢v⁢ββ NME for the 136 Xe isotope, based on the interacting shell model, but using three independent effective Hamiltonians, emphasizing the range of the NMEs' most probable values and its correlations with observables that can be obtained from the existing nuclear data. Consequently, we propose a common probability distribution function for the 0⁢vββ NME, which has a range of (1.55–2.65) at 90% confidence level, with a mean value of 1.99 and a standard deviation of 0.37.

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗

DOE OSTI2023

Multifidelity uncertainty quantification with models based on dissimilar parameters

Multifidelity uncertainty quantification (MF UQ) sampling approaches have been shown to significantly reduce the variance of statistical estimators while preserving the bias of the highest-fidelity model, provided that the low-fidelity models are well correlated. However, maintaining a high level of correlation can be challenging, especially when models depend on different input uncertain parameters, which drastically reduces the correlation. Existing MF UQ approaches do not adequately address this issue. In this work, we propose a new sampling strategy that exploits a shared space to improve the correlation among models with dissimilar parameterization. We achieve this by transforming the original coordinates onto an auxiliary manifold using the adaptive basis (AB) method (Tipireddy and Ghanem, 2014). The AB method has two main benefits: (1) it provides an effective tool to identify the low-dimensional manifold on which each model can be represented, and (2) it enables easy transformation of polynomial chaos representations from high- to low-dimensional spaces. This latter feature is used to identify a shared manifold among models without requiring additional evaluations. Here we present two algorithmic flavors of the new estimator to cover different analysis scenarios, including those with legacy and non-legacy high-fidelity (HF) data. We provide numerical results for analytical examples, a direct field acoustic test, and a finite element model of a nuclear fuel assembly. For all examples, we compare the proposed strategy against both single-fidelity and MF estimators based on the original model parameterization.

42 ENGINEERING↗

DOE OSTI2019

Improving Precipitation Estimation Using Convolutional Neural Network

Precipitation process is generally considered to be poorly represented in numerical weather/climate models. Statistical downscaling (SD) methods, which relate precipitation with model resolved dynamics, often provide more accurate precipitation estimates compared to model's raw precipitation products. Herein we introduce the convolutional neural network model to foster this aspect of SD for daily precipitation prediction. Specifically, we restrict the predictors to the variables that are directly resolved by discretizing the atmospheric dynamics equations. In this sense, our model works as an alternative to the existing precipitation-related parameterization schemes for numerical precipitation estimation. We train the model to learn precipitation-related dynamical features from the surrounding dynamical fields by optimizing a hierarchical set of spatial convolution kernels. We test the model at 14 geogrid points across the contiguous United States. Results show that provided with enough data, precipitation estimates from the convolutional neural network model outperform the reanalysis precipitation products, as well as SD products using linear regression, nearest neighbor, random forest, or fully connected deep neural network. Evaluation for the test set suggests that the improvements can be seamlessly transferred to numerical weather modeling for improving precipitation prediction. Based on the default network, we examine the impact of the network architectures on model performance. Also, we offer simple visualization and analyzing approaches to interpret the models and their results. Our study contributes to the following two aspects: First, we offer a novel approach to enhance numerical precipitation estimation; second, the proposed model provides important implications for improving precipitation-related parameterization schemes using a data-driven approach.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2022

Perturbation theory models for LSST-era galaxy clustering: Tests with subpercent mock catalog measurements in Fourier and configuration space

Here, we analyze the clustering of galaxies using the z = 1.006 snapshot of the CosmoDC2 simulation, a high-fidelity synthetic galaxy catalog designed to validate analysis methods for the Vera C. Rubin Observatory Legacy Survey of Space and Time (LSST). We present sub-percent measurements of the galaxy auto-correlation and galaxy-dark matter cross correlations in Fourier space and configuration space for a magnitude-limited galaxy sample. At these levels of precision, the statistical errors of the measurement are comparable to the systematic effects present in the simulation and measurement procedure; nevertheless, using a hybrid-PT model, we are able to model non-linear galaxy bias with 0.5% precision up to scales of k max = 0.5 h/Mpc and r min = 4 Mpc/h. While the linear bias parameter is measured with 0.01% precision, other bias parameters are determined with considerably weaker constraints and sometimes bimodal posterior distributions. We compare our fiducial model with lower dimensional models where the higher-order bias parameters are fixed at their co-evolution values and find that leaving these parameters free provides significant improvements in our ability to model small scale information. We also compare bias parameters for galaxy samples defined using different magnitude bands and find agreement between samples containing equal numbers of galaxies. Finally, we compare bias parameters between Fourier space and configuration space and find moderate to significant tension between the two approaches. Although our model is often unable to fit the CosmoDC2 galaxy samples within the 0.1% precision of our measurements, our results suggest that the hybrid-PT model used in this analysis is capable of modeling non-linear galaxy bias within the percent level precision needed for upcoming galaxy surveys.

79 ASTRONOMY AND ASTROPHYSICS↗

DOE OSTI2025

Statistical estimates of the binary properties of rotational variables

ABSTRACT We present a model to estimate the average primary masses, companion mass ranges, the inclination limit for recognizing a rotational variable, and the primary mass spreads for populations of binary stars. The model fits a population’s binary mass function distribution and allows for a probability that some mass functions are incorrectly estimated. Using tests with synthetic data, we assess the model’s sensitivity to each parameter, finding that we are most sensitive to the average primary mass and the minimum companion mass, with less sensitivity to the inclination limit and little to no sensitivity to the primary mass spread. We apply the model to five populations of binary spotted rotational variables identified in ASAS-SN, computing their binary mass functions using RV data from APOGEE. Their average primary mass estimates are consistent with our expectations based on their CMD locations ($\sim 0.75 \, {\rm M}_{\odot }$ for lower main sequence primaries and $\sim 0.9$–$1.2 \, {\rm M}_{\odot }$ for RS CVn and sub-subgiants). Their companion mass range estimates allow companion masses down to $M_2/M_1\simeq 0.1$, although the main sequence population may have a higher minimum mass fraction ($\sim 0.4$). We see weak evidence of an inclination limit $\gtrsim 50^{\circ }$ for the main sequence and sub-subgiant groups and no evidence of an inclination limit in the other groups. No groups show strong evidence for a preferred primary mass spread. We conclude by demonstrating that the approach will provide significantly better estimates of the primary mass and the minimum mass ratio and reasonable sensitivity to the inclination limit with 10 times as many systems.

Phillips, Anya (ORCID:000900051914974X)↗

DOE OSTI2023

Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

42 ENGINEERING↗

DOE OSTI2023

Reprint of: Explainable AI (XAI)-driven vibration sensing scheme for surface quality monitoring in a smart surface grinding process

Local Interpretable and Model-agnostic Explanation (LIME), an explainable artificial intelligence (XAI) approach is adapted to identify the globally important time-frequency bands for predicting average surface roughness (Ra) in a smart grinding process. The smart grinding setup consisted of a Supertech CNC precision surface grinding machine, instrumented with a Dytran piezoelectric accelerometer attached to the tailstock quill along the tangential direction (Y-axis). For every grinding pass, vibration signatures were captured, and the ground truth surface roughness values were recorded using a Mahr Marsurf M300C portable surface roughness profilometer. The roughness values ranged from 0.06 to 0.14 microns over the complete set of experiments. Time-frequency domain spectrogram frames were extracted for each of the vibration signals collected during the grinding process. Convolutional Neural Networks (CNNs) were modeled to predict the surface roughness based on these spectrogram frames and their image augmentations. The best CNN model was able to predict the roughness values with an overall R2-score of 0.95, training R2-score of 0.99, and testing R2-score of 0.81 with only 80 sets of vibration signals corresponding to 4 experiments with 20 trials each. Although the data size is not large enough to guarantee such performance metrics in real-world scenarios, one can extract statistically consistent explanations underlying the relationships these complex deep learning models capture. Further, the LIME methodology was implemented on the developed surface roughness CNN model to identify the important time-frequency bands (i.e., the superpixels of a spectrogram) influencing the predictions. Based on the identified important regions on the spectrogram frames, the corresponding frequency characteristics were determined that influence the surface roughness predictions. The important frequency range based on LIME results was approximately 11.7 to 19.1 kHz. The power of XAI was demonstrated by cutting down the sampling rate from 160 kHz to 30, 20, 10, and 5 kHz based on the important frequency range and considering Nyquist criteria. Separate CNN models were developed for these ranges by only extracting time-frequency contents below their corresponding Nyquist cut-offs. A proper data acquisition strategy is proposed by comparing the model performances to argue the selection of a sufficient sampling rate to capture the grinding process successfully and robustly.

47 OTHER INSTRUMENTATION↗