Scalable uncertainty quantification for deep operator networks using randomized priors
Not provided.
SEARCH · Search NASA
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.
Not provided.
We describe best practices for providing convenient, high-speed, secure access to large data via research data portals. Here, we capture these best practices in a new design pattern, the Modern Research Data Portal, that disaggregates the traditional monolithic web-based data portal to achieve orders-of-magnitude increases in data transfer performance, support new deployment architectures that decouple control logic from data storage, and reduce development and operations costs. We introduce the design pattern; explain how it leverages high-performance data enclaves and cloud-based data management services; review representative examples at research laboratories and universities, including both experimental facilities and supercomputer sites; describe how to leverage Python APIs for authentication, authorization, data transfer, and data sharing; and use coding examples to demonstrate how these APIs can be used to implement a range of research data portal capabilities. Sample code at a companion web site, https://docs.globus.org/mrdp , provides application skeletons that readers can adapt to realize their own research data portals.
All known small solar system bodies have diameters between a few meters and a few thousands of kilometers. Based on the collisional evolution of solar system bodies, a larger number of asteroids with diameters down to ~2 m is thought to exist. As all solar system bodies, small bodies can be passive sources of high-energy gamma-rays, produced by the interaction of energetic cosmic rays impinging on their surfaces. Since the majority of known asteroids are in orbits between Mars and Jupiter (in a region known as the Main Belt), we expect them to produce a diffuse emission close to the ecliptic plane. In this work, we have studied the gamma-ray emission coming from the ecliptic using the data collected by the Large Area Telescope (LAT) onboard the Fermi satellite. We have fit the results with simulations of the gamma-ray intensity at the source level (calculated with the software FLUKA) to constrain the small solar system bodies population. Finally, we have proposed a model describing the distribution of asteroid sizes and we have used the LAT data to constrain the gamma-ray emission expected from this model and, in turn, on the model itself.
Here, we present an overview of the High-Luminosity (HL-LHC) program at the Large Hadron Collider (LHC), its scientific potential and technological challenges for both the accelerator and detectors. The HL-LHC program is expected to start circa 2027 and aims to increase the integrated luminosity delivered by the LHC by an order of magnitude at the collision energy of 14 TeV. This requires upgrades to the injector system, accelerator complex and luminosity levelling. The two experiments, ATLAS and CMS, require substantial upgrades to most of their systems in order to cope with the increased interaction rate, and much higher radiation levels than at the current LHC. We present selected examples based on novel ideas and technologies for applications at a hadron collider. Both experiments will replace their tracking systems. We describe the ATLAS pixel detector upgrade featuring novel tilted modules, and the CMS Outer Tracker upgrade with a new module design enabling use of tracks in the level-1 trigger system. CMS will also install state-of-the-art highly segmented calorimeter endcaps. Finally, we describe new picosecond precision timing detectors of both experiments. In addition, we discuss how the upgrades will enhance the physics performance of the experiments, and solve the computing challenges posed by the expected large data sets. The physics program of the HL-LHC is focused on precision measurements probing the limits of the Standard Model (SM) of particle physics and discovering new physics. We present a selection of studies that have been carried out to motivate the HL-LHC program. A central topic of exploration will be the characterization of the Higgs boson. The large HL-LHC data samples will extend the sensitivity of searches for new particles or new interactions whose existence has been hypothesized in order to explain shortcomings of the SM. Finally, we comment on the nature of large scientific collaborations.
Here we describe best practices for providing convenient, high-speed, secure access to large data via research data portals. We capture these best practices in a new design pattern, the Modern Research Data Portal, that disaggregates the traditional monolithic web-based data portal to achieve orders-of-magnitude increases in data transfer performance, support new deployment architectures that decouple control logic from data storage, and reduce development and operations costs. We introduce the design pattern; explain how it leverages high-performance Science DMZs and cloud-based data management services; review representative examples at research laboratories and universities, including both experimental facilities and supercomputer sites; describe how to leverage Python APIs for authentication, authorization, data transfer, and data sharing; and use coding examples to demonstrate how these APIs can be used to implement a range of research data portal capabilities. Sample code at a companion web site, https://docs.globus.org/mrdp, provides application skeletons that readers can adapt to realize their own research data portals.
Soil organic carbon (SOC) regulates terrestrial ecosystem functioning, provides diverse energy sources for soil microorganisms, governs soil structure, and regulates the availability of organically bound nutrients. Investigators in increasingly diverse disciplines recognize how quantifying SOC attributes can provide insight about ecological states and processes. Today, multiple research networks collect and provide SOC data, and robust, new technologies are available for managing, sharing, and analyzing large data sets. Here, we advocate that the scientific community capitalize on these developments to augment SOC data sets via standardized protocols. We describe why such efforts are important and the breadth of disciplines for which it will be helpful, and outline a tiered approach for standardized sampling of SOC and ancillary variables that ranges from simple to more complex. We target scientists ranging from those with little to no background in soil science to those with more soil-related expertise, and offer examples of the ways in which the resulting data can be organized, shared, and discoverable.
We introduce the deep post Hartree–Fock (DeePHF) method, a machine learning-based scheme for constructing accurate and transferable models for the ground-state energy of electronic structure problems. DeePHF predicts the energy difference between results of highly accurate models such as the coupled cluster method and low accuracy models such as the Hartree–Fock (HF) method, using the ground-state electronic orbitals as the input. It preserves all the symmetries of the original high accuracy model. The added computational cost is less than that of the reference HF or DFT and scales linearly with respect to system size. We examine the performance of DeePHF on organic molecular systems using publicly available data sets and obtain the state-of-art performance, particularly on large data sets.
We present a combination of high-resolution Hubble Space Telescope and wide-field ground-based and Galaxy Evolution Explorer data of the Galactic globular cluster M10 (NGC 6254). By using this large data set, we determined the center of gravity of the cluster and we built its density profile from star counts over its entire radial extension. We find that the density profile is well reproduced by a single-mass King model with structural parameters c = 1.41 and r{sub c} = 41''. We also studied the blue straggler star (BSS) population and its radial distribution. We count a total number of 120 BSS within the tidal radius. Their radial distribution is bimodal: highly peaked in the cluster center, decreasing at intermediate distances, and rising again outward. We discuss these results in the context of the dynamical clock scheme presented by Ferraro et al. and of recent results about the radial distribution of binary systems in this cluster.
ABSTRACT Galaxy morphology is a fundamental quantity, which is essential not only for the full spectrum of galaxy-evolution studies, but also for a plethora of science in observational cosmology (e.g. as a prior for photometric-redshift measurements and as contextual data for transient light-curve classifications). While a rich literature exists on morphological-classification techniques, the unprecedented data volumes, coupled, in some cases, with the short cadences of forthcoming ‘Big-Data’ surveys (e.g. from the LSST), present novel challenges for this field. Large data volumes make such data sets intractable for visual inspection (even via massively distributed platforms like Galaxy Zoo), while short cadences make it difficult to employ techniques like supervised machine learning, since it may be impractical to repeatedly produce training sets on short time-scales. Unsupervised machine learning, which does not require training sets, is ideally suited to the morphological analysis of new and forthcoming surveys. Here, we employ an algorithm that performs clustering of graph representations, in order to group image patches with similar visual properties and objects constructed from those patches, like galaxies. We implement the algorithm on the Hyper-Suprime-Cam Subaru-Strategic-Program Ultra-Deep survey, to autonomously reduce the galaxy population to a small number (160) of ‘morphological clusters’, populated by galaxies with similar morphologies, which are then benchmarked using visual inspection. The morphological classifications (which we release publicly) exhibit a high level of purity, and reproduce known trends in key galaxy properties as a function of morphological type at z < 1 (e.g. stellar-mass functions, rest-frame colours, and the position of galaxies on the star-formation main sequence). Our study demonstrates the power of unsupervised machine learning in performing accurate morphological analysis, which will become indispensable in this new era of deep-wide surveys.
We update a previous investigation of cosmological effects of a nonstandard interaction between neutrinos and dark matter. Parametrizing the elastic-scattering cross section between the two species as a function of the temperature of the Universe, the resulting neutrino-dark matter fluid has a nonzero pressure, which determines diffusion-damped oscillations in the matter power spectrum similar to the acoustic oscillations generated by the photon-baryon fluid. Using cosmic microwave background data in combination with large scale structure experiment results, we then put constraints on the fraction of the interacting dark matter component as well as on the corresponding opacity.
Scientific applications produce vast amounts of data, posing grand challenges in the underlying data management and analytic tasks. Progressive compression is a promising way to address this problem, as it allows for on-demand data retrieval with significantly reduced data movement cost. However, most existing progressive methods are designed for CPUs, leaving a gap for them to unleash the power of today’s heterogeneous computing systems with GPUs.In this work, we propose HP-MDR, a high-performance and portable data refactoring and progressive retrieval framework for GPUs. Our contributions are four-fold: (1) We carefully optimize the bitplane encoding and lossless encoding, two key stages in progressive methods, to achieve high performance on GPUs; (2) We propose pipeline optimization and incorporate it with data refactoring and progressive retrieval workflows to further enhance the performance for large data process; (3) We leverage our framework to enable high-performance data retrieval with guaranteed error control for common Quantities of Interest; (4) We evaluate HP-MDR and compare it with state of the arts using five real-world datasets. Experimental results demonstrate that HP-MDR delivers an average 13.68 × and 6.31 × throughput in data refactoring and progressive retrieval tasks, respectively. It also leads to 11.22 × throughput for recomposing required data representations under Quantity-of-Interest error control and 6.04 × performance for the corresponding end-to-end data retrieval, when compared with state-of-the-art solutions.
Graphical abstract: - Highlights: • An automated workflow to optimize force-field parameters. • Used the workflow to optimize force-field parameter for a system containing nanodiamond and tRNA. • The mechanism relies on molecular dynamics simulation and neutron scattering experimental data. • The workflow can be generalized to any other experimental and simulation techniques. - Abstract: Large-scale simulations and data analysis are often required to explain neutron scattering experiments to establish a connection between the fundamental physics at the nanoscale and data probed by neutrons. However, to perform simulations at experimental conditions it is critical to use correct force-field (FF) parameters which are unfortunately not available for most complex experimental systems. In this work, we have developed a workflow optimization technique to provide optimized FF parameters by comparing molecular dynamics (MD) to neutron scattering data. We describe the workflow in detail by using an example system consisting of tRNA and hydrophilic nanodiamonds in a deuterated water (D{sub 2}O) environment. Quasi-elastic neutron scattering (QENS) data show a faster motion of the tRNA in the presence of nanodiamond than without the ND. To compare the QENS and MD results quantitatively, a proper choice of FF parameters is necessary. We use an efficient workflow to optimize the FF parameters between the hydrophilic nanodiamond and water by comparing to the QENS data. Our results show that we can obtain accurate FF parameters by using this technique. The workflow can be generalized to other types of neutron data for FF optimization, such as vibrational spectroscopy and spin echo.
Current methods for characterizing Earth’s subsurface, such as standard inverse techniques, are not sufficiently accurate to meet the needs of modern applications in the fields of energy exploration, environmental management, and global security. While increasing the quantity of field measurements and robustness of the applied data-/model-analysis methods can improve accuracy, such approaches can be computationally impractical for large data sets and complex site conditions. Thus, there is a need to develop economically-feasible and robust computational methods while maintaining accuracy. For example, in-field drilling for geothermal operations may yield high failure rates, resulting in unacceptably high costs; errors and/or large uncertainties in the estimated subsurface characteristics are the main impediment to the successful siting of an in-field well. This problem is not uniquely geothermal. Accurate characterization of uncertain subsurface properties is also essential for monitoring storage of carbon dioxide, estimating pathways of subsurface contaminant transport, and supervising ground-based nuclear-explosion tests.
The Atmospheric Radiation Measurement (ARM) large-scale forcing data developed based on the constrained variational analysis (VARANAL) value-added product (VAP) (Zhang and Lin 1997, Zhang et al. 2001, Xie et al. 2004, Tang et al. 2019) has been widely used for single-column models (SCMs), cloud-resolving models (CRMs), and large-eddy simulation models (LESs) to understand and improve physical processes in models. Recently, the U.S. Department of Energy (DOE) ARM user facility conducted several major field campaigns using ship-based moving observational platforms. For example, the Marine ARM GPCI Investigation of Clouds (MAGIC) field campaign focused on the role of subtropical marine-boundary layer (MBL) clouds, and the Multidisciplinary Drifting Observatory for the Study of Arctic Climate (MOSAiC) field campaign aimed to improve understanding of the coupled climate systems in the Arctic. Observations from moving platforms are critical to provide a comprehensive characterization of coupled-system processes associated with all stages of the cloud and/or sea-ice life cycle. Traditional ARM large-scale forcing data have been developed at fixed locations. They need to be extended to include these moving platforms to address data needs for ship-based field campaigns or to support LES modeling in a Lagrangian framework. With these considerations in mind, we develop ARM-type Lagrangian large-scale forcing data sets based on the lagtraj framework (Boeing et al. 2020) with notable enhancements in generating forcings that are more suitable for ARM field campaigns. The lagtraj is a novel tool that generates forcings for LES and SCM simulation in both Lagrangian and Eulerian perspective. This technical report focuses on the major changes we performed on the lagtraj algorithm and provides an overview of the ARM Lagrangian Large-Scale Forcing Data (ARMLAGTRAJ) value-added products.
Investment in brighter sources and larger and faster detectors has accelerated the speed of data acquisition at national user facilities. The accelerated data acquisition offers many opportunities for the discovery of new materials, but it also presents a daunting challenge. The rate of data acquisition far exceeds the current speed of data quality assessment, resulting in less than optimal data and data coverage, which in extreme cases forces recollection of data. Herein, we show how this challenge can be addressed through the development of an approach that makes routine data assessment automatic and instantaneous. By extracting and visualizing customized attributes in real time, data quality and coverage, as well as other scientifically relevant information contained in large data sets, is highlighted. Deployment of such an approach not only improves the quality of data but also helps optimize the usage of expensive characterization resources by prioritizing measurements of the highest scientific impact. We anticipate our approach will become a starting point for a sophisticated decision-tree that optimizes data quality and maximizes scientific content in real time through automation. Finally, with these efforts to integrate more automation in data collection and analysis, we can truly take advantage of the accelerating speed of data acquisition.
Combinatorial experiments involve synthesis of sample libraries with lateral composition gradients requiring spatially-resolved characterization of structure and properties. Here, due to maturation of combinatorial methods and their successful application in many fields, the modern combinatorial laboratory produces diverse and complex data sets requiring advanced analysis and visualization techniques. In order to utilize these large data sets to uncover new knowledge, the combinatorial scientist must engage in data science. For data science tasks, most laboratories adopt common-purpose data management and visualization software. However, processing and cross-correlating data from various measurement tools is no small task for such generic programs. Here we describe COMBIgor, a purpose-built open-source software package written in the commercial Igor Pro environment, designed to offer a systematic approach to loading, storing, processing, and visualizing combinatorial data sets. It includes (1) methods for loading and storing data sets from combinatorial libraries, (2) routines for streamlined data processing, and (3) data analysis and visualization features to construct figures. Most importantly, COMBIgor is designed to be easily customized by a laboratory, group, or individual in order to integrate additional instruments and data-processing algorithms. Utilizing the capabilities of COMBIgor can significantly reduce the burden of data management on the combinatorial scientist.
This paper develops a physically justified reduced-order capacity fade model from accelerated calendar- and cycle-aging data for 32 lithium-ion (Li-ion) graphite/nickel-manganese-cobalt (NMC) cells. The large data set reveals temperature-, charge C-rate-, depth-of-discharge-, and state of charge (SOC)-dependent degradation patterns that would be unobserved in a smaller test matrix. Model structure is informed by incremental capacity analysis that shows loss of lithium inventory and cathode-material loss as the dominant capacity fade mechanisms. The model includes terms attributable to solid-electrolyte interface (SEI) growth, electrode cracking, cycling-driven acceleration of SEI growth, and "break-in" mechanisms that slightly decrease or increase available Li inventory early in life. The study explores what mathematical couplings of these mechanisms best describe calendar aging, cycle aging, and mixed calendar/cycle aging. Various approaches are discussed for extracting relevant stress factors from complex cycling profiles to predict lifetime during real-world battery loads using models trained on constant-current laboratory test results. The complexity of the present human-driven model identification process motivates future work in machine learning to more widely search and statistically discern the optimal model that correctly extrapolates capacity fade based on physical knowledge.
OpenStudio is a cross-platform collection of software tools to support whole building energy modeling using EnergyPlus and advanced daylight analysis using Radiance. This disclosure includes two OpenStudio measures: (1) data center prototype models, and (2) occupant schedule generator. The data center models cover two sizes (small computer server rooms, stand-alone large data centers) and two IT load densities (average and high).