Search NASA⌕ Search

SEARCH · Search NASA

Results for “statistical selection”

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 307 records · Page 17

United States Industrial Sector Energy End Use Analysis

The United States Department of Energy’s (DOE) Energy Information Administration (EIA) conducts the Manufacturing Energy Consumption Survey (MECS) to provide detailed data on energy consumption in the manufacturing sector. The survey is a sample of approximately 15,000 manufacturing establishments selected from the Economic Census - Manufacturing Sector. MECS provides statistics on the consumption of energy by end uses (e.g., boilers, process, electric drives, etc.) disaggregated by North American Industry Classification System (NAICS) categories. The manufacturing sector (NAICS Sector 31-33) consists of all manufacturing establishments in the 50 States and the District of Columbia. According to the NAICS, the manufacturing sector comprises establishments engaged in the mechanical, physical, or chemical transformation of materials, substances, or components into new products. The establishments are physical facilities such as plants, factories, or mills. For many of the sectors in the MECS datasets, information is missing because the reported energy use is less than 0.5 units or BTUs, or is withheld to avoid disclosing data for individual establishments, or is withheld because the standard error is greater than 50%. We infer what the missing information likely are using several approximations techniques. First, much of the missing data can be easily calculated by adding or subtracting other values reported by MECS. If this is not possible (e.g. two data are missing), we look at historic MECS reports to help identify the breakdown of energy use in the past and assume it remained the same for the current MECS. Lastly, if historic data is also missing, we assume that 3 digit NAICS classifications predict energy use in their 4, 5, or 6 digit NAICS sub-classifications, or vice versa. Along with addressing data gaps, end use energy is disaggregated beyond the specified MECS allocations using additional industry specific energy consumption data. The result is a completed table of energy end use by sector with mechanical drives broken down by pumps, fans, compressed air, and drives.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Progress on a High Sensitivity, High Purity, One-Electron-One-Proton Search for the MiniBooNE Low Energy Excess in the MicroBooNE Detector

MicroBooNE is a large liquid argon time projection chamber (LArTPC) located at Fermilab. One of the experiments primary aims is to investigate the CCQE-like excess of ν ϵ events observed by MiniBooNE at low energy. A novel approach to analyzing LArTPC data using an amalgam of Deep Learning and traditional algorithms has been developed and is capable of end to end reconstruction and selection. We report here an overview and progress to event selection which yields greater than 99.999% rejection of backgrounds. Statistical, flux, and some cross section systematics have also been evaluated. Detector systematics and additional cross section systematics are still under study.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Funding for the 2ND IAEA technical meeting on fusion data processing, validation and analysis

The International Atomic Energy Agency (IAEA) will organize the second Technical Meeting on Fusion Da Processing, Validation and Analysis from 30 May to 02 June, 2017, in Cambridge, MA USA. The meeting w be hosted by the MIT Plasma Science and Fusion Center (PSFC). The objective of the meeting is to provide a platform where a set of topics relevant to fusion data processing, validation and analysis are discussed with the view of extrapolation needs to next step fusion devices such as ITER. The validation and analysis of experimental data obtained from diagnostics used to characterize fusion plasmas are crucial for a knowledge based understanding of the physical processes governing the dynamics of these plasmas. The meeting will aim at fostering, in particular, discussions of research and development results that set out or underline trends observed in the current major fusion confinement devices. General information on the IAEA, including its mission and organization, can be found at the IAEA websit Uncertainty quantification (UQ) Model selection, validation, and verification (V&V) Probability theory and statistical analysis Inverse problems & equilibrium reconstru ction Integrated data analysis Real time data analysis Machine learning Signal/image proc essing & pattern recognition Experimental design and synthetic diagnostics Data management

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Multimodel Predictive System for Carbon Dioxide Solubility in Saline Formation Waters

The prediction of carbon dioxide solubility in brine at conditions relevant to carbon sequestration (i.e., high temperature, pressure, and salt concentration (T-P-X)) is crucial when this technology is applied. Eleven mathematical models for predicting CO{sub 2} solubility in brine are compared and considered for inclusion in a multimodel predictive system. Model goodness of fit is evaluated over the temperature range 304–433 K, pressure range 74–500 bar, and salt concentration range 0–7 m (NaCl equivalent), using 173 published CO{sub 2} solubility measurements, particularly selected for those conditions. The performance of each model is assessed using various statistical methods, including the Akaike Information Criterion (AIC) and the Bayesian Information Criterion (BIC). Different models emerge as best fits for different subranges of the input conditions. A classification tree is generated using machine learning methods to predict the best-performing model under different T-P-X subranges, allowing development of a multimodel predictive system (MMoPS) that selects and applies the model expected to yield the most accurate CO{sub 2} solubility prediction. Statistical analysis of the MMoPS predictions, including a stratified 5-fold cross validation, shows that MMoPS outperforms each individual model and increases the overall accuracy of CO{sub 2} solubility prediction across the range of T-P-X conditions likely to be encountered in carbon sequestration applications.

54 ENVIRONMENTAL SCIENCES↗

WE-H-207A-03: The Universality of the Lognormal Behavior of [F-18]FLT PET SUV Measurements

Purpose: Log transforming [F-18]FDG PET standardized uptake values (SUVs) has been shown to lead to normal SUV distributions, which allows utilization of powerful parametric statistical models. This study identified the optimal transformation leading to normally distributed [F-18]FLT PET SUVs from solid tumors and offers an example of how normal distributions permits analysis of non-independent/correlated measurements. Methods: Forty patients with various metastatic diseases underwent up to six FLT PET/CT scans during treatment. Tumors were identified by nuclear medicine physician and manually segmented. Average uptake was extracted for each patient giving a global SUVmean (gSUVmean) for each scan. The Shapiro-Wilk test was used to test distribution normality. One parameter Box-Cox transformations were applied to each of the six gSUVmean distributions and the optimal transformation was found by selecting the parameter that maximized the Shapiro-Wilk test statistic. The relationship between gSUVmean and a serum biomarker (VEGF) collected at imaging timepoints was determined using a linear mixed effects model (LMEM), which accounted for correlated/non-independent measurements from the same individual. Results: Untransformed gSUVmean distributions were found to be significantly non-normal (p<0.05). The optimal transformation parameter had a value of 0.3 (95%CI: −0.4 to 1.6). Given the optimal parameter was close to zero (which corresponds to log transformation), the data were subsequently log transformed. All log transformed gSUVmean distributions were normally distributed (p>0.10 for all timepoints). Log transformed data were incorporated into the LMEM. VEGF serum levels significantly correlated with gSUVmean (p<0.001), revealing log-linear relationship between SUVs and underlying biology. Conclusion: Failure to account for correlated/non-independent measurements can lead to invalid conclusions and motivated transformation to normally distributed SUVs. The log transformation was found to be close to optimal and sufficient for obtaining normally distributed FLT PET SUVs. These transformations allow utilization of powerful LMEMs when analyzing quantitative imaging metrics.

60 APPLIED LIFE SCIENCES↗

A Parameter Subset Selection Algorithm for Mixed-Effects Models

Mixed-effects models are commonly used to statistically model phenomena that include attributes associated with a population or general underlying mechanism as well as effects specific to individuals or components of the general mechanism. This can include individual effects associated with data from multiple experiments. However, the parameterizations used to incorporate the population and individual effects are often unidentifiable in the sense that parameters are not uniquely specified by the data. As a result, the current literature focuses on model selection, by which insensitive parameters are fixed or removed from the model. Model selection methods that employ information criteria are applicable to both linear and nonlinear mixed-effects models, but such techniques are limited in that they are computationally prohibitive for large problems due to the number of possible models that must be tested. To limit the scope of possible models for model selection via information criteria, we introduce a parameter subset selection (PSS) algorithm for mixed-effects models, which orders the parameters by their significance. In conclusion, we provide examples to verify the effectiveness of the PSS algorithm and to test the performance of mixed-effects model selection that makes use of parameter subset selection.

97 MATHEMATICS AND COMPUTING↗

Generalized correlation integral vectors: A distance concept for chaotic dynamical systems

Several concepts of fractal dimension have been developed to characterise properties of attractors of chaotic dynamical systems. Numerical approximations of them must be calculated by finite samples of simulated trajectories. In principle, the quantities should not depend on the choice of the trajectory, as long as it provides properly distributed samples of the underlying attractor. In practice, however, the trajectories are sensitive with respect to varying initial values, small changes of the model parameters, to the choice of a solver, numeric tolerances, etc. The purpose of this paper is to present a statistically sound approach to quantify this variability. We modify the concept of correlation integral to produce a vector that summarises the variability at all selected scales. The distribution of this stochastic vector can be estimated, and it provides a statistical distance concept between trajectories. Here, we demonstrate the use of the distance for the purpose of estimating model parameters of a chaotic dynamic model. The methodology is illustrated using computational examples for the Lorenz 63 and Lorenz 95 systems, together with a framework for Markov chain Monte Carlo sampling to produce posterior distributions of model parameters.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Spectroscopic failures in photometric redshift calibration: cosmological biases and survey requirements

We use N-body-spectro-photometric simulations to investigate the impact of incompleteness and incorrect redshifts in spectroscopic surveys to photometric redshift training and calibration and the resulting effects on cosmological parameter estimation from weak lensing shear-shear correlations. The photometry of the simulations is modeled after the upcoming Dark Energy Survey and the spectroscopy is based on a low/intermediate resolution spectrograph with wavelength coverage of 5500{\AA} < {\lambda} < 9500{\AA}. The principal systematic errors that such a spectroscopic follow-up encounters are incompleteness (inability to obtain spectroscopic redshifts for certain galaxies) and wrong redshifts. Encouragingly, we find that a neural network-based approach can effectively describe the spectroscopic incompleteness in terms of the galaxies' colors, so that the spectroscopic selection can be applied to the photometric sample. Hence, we find that spectroscopic incompleteness yields no appreciable biases to cosmology, although the statistical constraints degrade somewhat because the photometric survey has to be culled to match the spectroscopic selection. Unfortunately, wrong redshifts have a more severe impact: the cosmological biases are intolerable if more than a percent of the spectroscopic redshifts are incorrect. Moreover, we find that incorrect redshifts can also substantially degrade the accuracy of training set based photo-z estimators. The main problem is the difficulty of obtaining redshifts, either spectroscopically or photometrically, for objects at z > 1.3. We discuss several approaches for reducing the cosmological biases, in particular finding that photo-z error estimators can reduce biases appreciably.

79 ASTRONOMY AND ASTROPHYSICS↗

THE ABSOLUTE MAGNITUDE OF RRc VARIABLES FROM STATISTICAL PARALLAX

We present the first definitive measurement of the absolute magnitude of RR Lyrae c-type variable stars (RRc) determined purely from statistical parallax. We use a sample of 242 RRc variables selected from the All Sky Automated Survey for which high-quality light curves, photometry, and proper motions are available. We obtain high-resolution echelle spectra for these objects to determine radial velocities and abundances as part of the Carnegie RR Lyrae Survey. We find that M{sub V,RRc} = 0.59 ± 0.10 at a mean metallicity of [Fe/H] = –1.59. This is to be compared with previous estimates for RRab stars (M{sub V,RRab} = 0.76 ± 0.12) and the only direct measurement of an RRc absolute magnitude (RZ Cephei, M{sub V,RRc} = 0.27 ± 0.17). We find the bulk velocity of the halo relative to the Sun to be (W{sub π}, W{sub θ}, W{sub z} ) = (12.0, –209.9, 3.0) km s{sup –1} in the radial, rotational, and vertical directions with dispersions (σ{sub W{sub π}},σ{sub W{sub θ}},σ{sub W{sub z}}) = (150.4, 106.1, 96.0) km s{sup -1}. For the disk, we find (W{sub π}, W{sub θ}, W{sub z} ) = (13.0, –42.0, –27.3) km s{sup –1} relative to the Sun with dispersions (σ{sub W{sub π}},σ{sub W{sub θ}},σ{sub W{sub z}}) = (67.7,59.2,54.9) km s{sup -1}. Finally, as a byproduct of our statistical framework, we are able to demonstrate that UCAC2 proper-motion errors are significantly overestimated as verified by UCAC4.

79 ASTRONOMY AND ASTROPHYSICS↗

Cold Commissioning Simulant Selection for the Hanford Tank Waste Treatment Plant LAW Facility - 18465

Commissioning the Low-Activity Waste (LAW) Facility of the Hanford Tank Waste Treatment and Immobilization Plant (WTP) will require large volumes (∼2,000,000 liters) of a non-radioactive, chemically analogous waste simulant. Cold commissioning will provide for operator training in an integrated facility, demonstration of mechanical process equipment capacity, environmental performance demonstration of emissions control systems, and process control systems shakedown and tuning. Selection of the cold commissioning simulant is following a structured four-phase process to arrive at the preferred composition and recipe. Phases I, II and III are complete. Phase I laid the framework for decision making with preparation of a Project Management Plan, an update to the Simulant Development, Approval, Validation, and Documentation procedure, and a kickoff meeting to gather stakeholder input into the simulant definition and selection criteria. Phase II consisted of a comprehensive literature search and compilation of LAW simulant recipes, followed by initial screening against the selection criteria identified in Phase I. This process identified 72 candidate simulant recipes and down-selected 9 for further study. Phase III studies involved a) statistical comparisons of the similarity of candidate simulant compositions to the tank waste feed vector, b) process modelling of the commissioning sequence with each simulant to estimate glass former and simulant material consumption, offgas and effluent volumes and compositions, and product glass volumes and compositions, and c) evaluation of simulants against the immobilized LAW glass composition algorithm to determine allo able waste loading to ensure compliance with product specifications and process constraints. The results of these studies identified two of the candidate recipes to be carried forward to full-scale testing. Phase IV consists of testing these two recipes at vendor facilities to determine scalability, fabricability, and stability of the simulants when produced at industrial scale. Scalability is the ability to take a recipe from bench scale (1 to 40 liters) batch size to full scale (>35,000 liters) batch size. Fabricability is the ability of a recipe to be prepared in large batches without the need for complex or expensive mechanical equipment, heating, or cooling systems. Stability is the ability of a simulant to withstand long term storage and variations in environmental conditions without unacceptable changes in key characteristics (the set of chemical, physical, and rheological parameters that the simulant is intended to exhibit during cold commissioning tests). Simulants will also be tested for compatibility with other startup material (i.e., frit), tuning simulants, and environmental performance demonstration test spiking solutions to ensure against unexpected situations. Using a structured process is essential when making high impact decisions such as waste simulant selection. Proper planning and a clear explanation of how the decision process will unfold is key to obtaining stakeholder buy-in as more information is gathered. The documentation resulting from this structured selection process will be compiled and summarized in a Cold Commissioning Simulant Definition deliverable due to the Department of Energy 24 months before the start of Cold Commissioning. (authors)

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

Selection Algorithm Improvement for MicroBooNE

Data selection is an extremely important part of data analysis for any experiment. Finding a physics result is often the result of sifting through a massive amount of data, keeping data that we believe to be signal and throwing out data we do not. This process is called data selection. Creating a selection algorithm is an intensive process that must balance keeping enough data to have statistics and maximizing the signal purity of that data. In this study, we used three different reconstruction tools, Pandora, WireCell, and LANTERN, for the MicroBooNE experiment in conjunction to improve the selection algorithm for analysis. For the case of this study, we look into the charged current N proton 0 pions (CCNp0$\pi$) interaction channel. This is the dominant channel for the Short Baseline Neutrino (SBN) program and is expected to be a large contributor to the Deep Underground Neutrino Experiment (DUNE). We first investigated each of the three tools to find out more about their strengths and weaknesses as reconstructions. We then put together a direct comparison of the three methods to find which method or combination of methods would return the best result for us. While the study is ongoing, we have learned a lot about data selection for the experiment and the differences between the reconstruction tools.

Dillon, Brayden [Michigan State U.]↗

Statistical Characterization of Medium-Duty Electric Vehicle Drive Cycles: Preprint

In an effort to help commercialize technologies for electric vehicles (EVs) through deployment and demonstration projects, the U.S. Department of Energy’s (DOE's) American Recovery and Reinvestment Act (ARRA) provided funding to participating U.S. companies to cover part of the cost of purchasing new EVs. Within the medium- and heavy-duty commercial vehicle segment, both Smith Electric Newton and and Navistar eStar vehicles qualified for such funding opportunities. In an effort to evaluate the performance characteristics of the new technologies deployed in these vehicles operating under real world conditions, data from Smith Electric and Navistar medium-duty EVs were collected, compiled, and analyzed by the National Renewable Energy Laboratory's (NREL) Fleet Test and Evaluation team over a period of 3 years. More than 430 Smith Newton EVs have provided data representing more than 150,000 days of operation. Similarly, data have been collected from more than 100 Navistar eStar EVs, resulting in a comparative total of more than 16,000 operating days. Combined, NREL has analyzed more than 6 million kilometers of driving and 4 million hours of charging data collected from commercially operating medium-duty electric vehicles in various configurations. In this paper, extensive duty-cycle statistical analyses are performed to examine and characterize common vehicle dynamics trends and relationships based on in-use field data. The results of these analyses statistically define the vehicle dynamic and kinematic requirements for each vehicle, aiding in the selection of representative chassis dynamometer test cycles and the development of custom drive cycles that emulate daily operation. In this paper, the methodology and accompanying results of the duty-cycle statistical analysis are presented and discussed. Results are presented in both graphical and tabular formats illustrating a number of key relationships between parameters observed within the data set that relate to medium duty EVs.

33 ADVANCED PROPULSION SYSTEMS↗

Statistical Characterization of Medium-Duty Electric Vehicle Drive Cycles

In an effort to help commercialize technologies for electric vehicles (EVs) through deployment and demonstration projects, the U.S. Department of Energy's (DOE's) American Recovery and Reinvestment Act (ARRA) provided funding to participating U.S. companies to cover part of the cost of purchasing new EVs. Within the medium- and heavy-duty commercial vehicle segment, both Smith Electric Newton and and Navistar eStar vehicles qualified for such funding opportunities. In an effort to evaluate the performance characteristics of the new technologies deployed in these vehicles operating under real world conditions, data from Smith Electric and Navistar medium-duty EVs were collected, compiled, and analyzed by the National Renewable Energy Laboratory's (NREL) Fleet Test and Evaluation team over a period of 3 years. More than 430 Smith Newton EVs have provided data representing more than 150,000 days of operation. Similarly, data have been collected from more than 100 Navistar eStar EVs, resulting in a comparative total of more than 16,000 operating days. Combined, NREL has analyzed more than 6 million kilometers of driving and 4 million hours of charging data collected from commercially operating medium-duty electric vehicles in various configurations. In this paper, extensive duty-cycle statistical analyses are performed to examine and characterize common vehicle dynamics trends and relationships based on in-use field data. The results of these analyses statistically define the vehicle dynamic and kinematic requirements for each vehicle, aiding in the selection of representative chassis dynamometer test cycles and the development of custom drive cycles that emulate daily operation. In this paper, the methodology and accompanying results of the duty-cycle statistical analysis are presented and discussed. Results are presented in both graphical and tabular formats illustrating a number of key relationships between parameters observed within the data set that relate to medium duty EVs.

33 ADVANCED PROPULSION SYSTEMS↗

The LSST DESC data challenge 1: generation and analysis of synthetic images for next-generation surveys

Data Challenge 1 (DC1) is the first synthetic data set produced by the Rubin Observatory Legacy Survey of Space and Time (LSST) Dark Energy Science Collaboration (DESC). DC1 is designed to develop and validate data reduction and analysis and to study the impact of systematic effects that will affect the LSST data set. DC1 is comprised of r -band observations of 40 deg 2 to 10 yr LSST depth. In this paper, we present each stage of the simulation and analysis process: (a) generation, by synthesizing sources from cosmological N -body simulations in individual sensor-visit images with different observing conditions; (b) reduction using a development version of the LSST Science Pipelines; and (c) matching to the input cosmological catalogue for validation and testing. We verify that testable LSST requirements pass within the fidelity of DC1. We establish a selection procedure that produces a sufficiently clean extragalactic sample for clustering analyses and we discuss residual sample contamination, including contributions from inefficiency in star–galaxy separation and imperfect deblending. We compute the galaxy power spectrum on the simulated field and conclude that: (i) survey properties have an impact of 50 per cent of the statistical uncertainty for the scales and models used in DC1; (ii) a selection to eliminate artefacts in the catalogues is necessary to avoid biases in the measured clustering; and (iii) the presence of bright objects has a significant impact (2σ–6σ) in the estimated power spectra at small scales (ℓ > 1200), highlighting the impact of blending in studies at small angular scales in LSST.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

Towards intelligent diagnostic system employing integration of mathematical and engineering model

The development of medical diagnostic system has been one of the main research fields during years. The goal of the medical diagnostic system is to place a nosological system that could ease the diagnostic evaluation normally performed by scientists and doctors. Efficient diagnostic evaluation is essentials and requires broad knowledge in order to improve conventional diagnostic system. Several approaches on developing the medical diagnostic system have been designed and tested since the earliest 60s. Attempts on improving their performance have been made which utilizes the fields of artificial intelligence, statistical analyses, mathematical model and engineering theories. With the availability of the microcomputer and software development as well as the promising aforementioned fields, medical diagnostic prototypes could be developed. In general, the medical diagnostic system consists of several stages, namely the 1) data acquisition, 2) feature extraction, 3) feature selection, and 4) classifications stages. Data acquisition stage plays an important role in converting the inputs measured from the real world physical conditions to the digital numeric values that can be manipulated by the computer system. One of the common medical inputs could be medical microscopic images, radiographic images, magnetic resonance image (MRI) as well as medical signals such as electrocardiogram (ECG) and electroencephalogram (EEG). Normally, the scientist or doctors have to deal with myriad of data and redundant to be processed. In order to reduce the complexity of the diagnosis process, only the significant features of the raw data such as peak value of the ECG signal or size of lesion in the mammogram images will be extracted and considered in the subsequent stages. Mathematical models and statistical analyses will be performed to select the most significant features to be classified. The statistical analyses such as principal component analysis and discriminant analysis as well as mathematical model of clustering technique have been widely used in developing the medical diagnostic systems. The selected features will be classified using mathematical models that embedded engineering theory such as artificial intelligence, support vector machine, neural network and fuzzy-neuro system. These classifiers will provide the diagnostic results without human intervention. Among many publishable researches, several prototypes have been developed namely NeuralPap, Neural Mammo, and Cervix Kit. The former system (NeuralPap) is an automatic intelligent diagnostic system for classifying and distinguishing between the normal and cervical cancerous cells. Meanwhile, the Cervix Kit is a portable Field-programmable gate array (FPGA)-based cervical diagnostic kit that could automatically diagnose the cancerous cell based on the images obtained during sampling test. Besides the cervical diagnostic system, the Neural Mammo system is developed to specifically aid the diagnosis of breast cancer using a fine needle aspiration image.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Cosmological implications of the CMB large-scale structure

The Wilkinson Microwave Anisotropy Probe (WMAP) and Planck may have uncovered several anomalies in the full cosmic microwave background (CMB) sky that could indicate possible new physics driving the growth of density fluctuations in the early universe. These include an unusually low power at the largest scales and an apparent alignment of the quadrupole and octopole moments. In a ΛCDM model where the CMB is described by a Gaussian Random Field, the quadrupole and octopole moments should be statistically independent. The emergence of these low probability features may simply be due to posterior selections from many such possible effects, whose occurrence would therefore not be as unlikely as one might naively infer. If this is not the case, however, and if these features are not due to effects such as foreground contamination, their combined statistical significance would be equal to the product of their individual significances. In the absence of such extraneous factors, and ignoring the biasing due to posterior selection, the missing large-angle correlations would have a probability as low as ∼0.1% and the low-l multipole alignment would be unlikely at the ∼4.9% level; under the least favorable conditions, their simultaneous observation in the context of the standard model could then be likely at only the ∼0.005% level. In this paper, we explore the possibility that these features are indeed anomalous, and show that the corresponding probability of CMB multipole alignment in the R{sub h}=ct universe would then be ∼7–10%, depending on the number of large-scale Sachs–Wolfe induced fluctuations. Since the low power at the largest spatial scales is reproduced in this cosmology without the need to invoke cosmic variance, the overall likelihood of observing both of these features in the CMB is ⩾7%, much more likely than in ΛCDM, if the anomalies are real. The key physical ingredient responsible for this difference is the existence in the former of a maximum fluctuation size at the time of recombination, which is absent in the latter because of inflation.

79 ASTRONOMY AND ASTROPHYSICS↗

Importance of Engineered and Learned Molecular Representations in Predicting Organic Reactivity, Selectivity, and Chemical Properties

Machine-readable chemical structure representations are foundational in all attempts to harness machine learning for the prediction of reactivities, selectivities, and chemical properties directly from molecular structure. The featurization of discrete chemical structures into a continuous vector space is a critical phase undertaken before model selection, and the development of new ways to quantitatively encode molecules is an active area of research. Here, we highlight the application and suitability of different representations, from expert-guided “engineered” descriptors to automatically “learned” features, in different prediction tasks relevant to organic and organometallic chemistry, where differing amounts of training data are available. These tasks include statistical models of stereo- and enantioselectivity, thermochemistry, and kinetics developed using experimental and quantum chemical data. The use of expert-guided molecular descriptors provides an opportunity to incorporate chemical knowledge, domain expertise, and physical constraints into statistical modeling. In applications to stereoselective organic and organometallic catalysis, where data sets may be relatively small and 3D-geometries and conformations play an important role, mechanistically informed features can be used successfully to obtain predictive statistical models that are also chemically interpretable. We provide an overview of several recent applications of this approach to obtain quantitative models for reactivity and selectivity, where topological descriptors, quantum mechanical calculations of electronic and steric properties, along with conformational ensembles, all feature as essential ingredients of the molecular representations used. Alternatively, more flexible, general-purpose molecular representations such as attributed molecular graphs can be used with machine learning approaches to learn the complex relationship between a structure and prediction target. This approach has the potential to out-perform more traditional representation methods such as “hand-crafted” molecular descriptors, particularly as data set sizes grow. One area where this is particularly relevant is in the use of large sets of quantum mechanical data to train quantitative structure–property relationships. A general approach toward curating useful data sets and training highly accurate graph neural network models is discussed in the context of organic bond dissociation enthalpies, where this strategy outperforms regression using precomputed descriptors. Finally, we describe how graph neural network predictions can be incorporated into mechanistically informed statistical models of chemical reactivity and selectivity. Once trained, this approach avoids the expensive computational overhead associated with quantum mechanical calculations, while maintaining chemical interpretability. We illustrate examples for which fast predictions of bond dissociation enthalpy and of the identities of radicals formed through cleavage of a molecule’s weakest bond are used in simple physical models of site-selectivity and reactivity.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Validation of LOCA2 and STAR-ESDM Statistically Downscaled Products

The National Climate Assessment (NCA) is the preeminent national report examining current and future risks posed by climate change. Countless agencies, policymakers, stakeholders and other end-users rely upon guidance from the NCA to plan for an uncertain future. These groups all depend on modern curated data, provided alongside the NCA, to quantify the impact of climate change on metrics of relevance for their decision processes. In its fifth iteration (NCA5), two statistically downscaled ensemble products, each providing data at grid spacing of approximately 5km over the contiguous United States, were selected to accompany the report. These include LOCalized Analogs version 2 (LOCA2) and Seasonal Trends and Analysis of Residuals Empirical-Statistical Downscaling Model (STAR-ESDM). Both data products are produced through a process known as statistical downscaling, where relatively coarse Global Climate Model (GCM) data is refined to locally relevant scales through the application of scientifically-supported empirical and algorithmic relationships. In support of the NCA effort, this report provides an independent validation of these two products against historical observations, with a focus on precipitation and near-surface temperature variables. Based on the results of this validation, several recommendations are provided related to the use of these data products. The structure of this report is as follows: In section 2, we review three gridded observational products that are used as part of our intercomparison. In section 3, we describe the two statistical downscaling techniques and their corresponding datasets that are the focus of this study. In section 4, the methodology we employ for validation is described. Section 5 provides results of the validation, which in turn motivate our recommendations on the use of these data products. A brief summary is provided in section 6.

54 ENVIRONMENTAL SCIENCES↗