Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Distance Learning”

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 469 records · Page 26

Neural Scaling Laws for Jet Generation

Recently observed empirical scaling laws describe the performance of foundation-type models as three independent key quantities -- dataset size, compute, and model parameters -- are modified. Extracting these scaling laws informs the training of large complex models for which the tuning of hyperparameters in traditional ways is not feasible. This work for the first time explores if scaling laws can also be observed for the task of particle jet generation -- both relevant as a pre-training objective for foundation models and as in-situ simulation by itself. We indeed replicate the key logarithmic scaling law behavior for model-size scaling. Beyond studying the next token prediction validation loss of the generative model, we also study the sliced Wasserstein distance of five physical quantities that are not immediately available to the model during training. Our study shows that this quantity is monotonically related to the next token prediction validation loss, meaning that this loss is indeed a good proxy for the physics performance. For the scaling with dataset size and compute, we observe substantially weaker scaling behavior of both the loss and the sliced Wasserstein distance. We analyze this behavior by introducing the concept of a learnable window, and argue that autoregressive next token prediction on jet constituents exhibits comparatively rapid saturation relative to language-model studies. We discuss possible origins of this behavior, including the stochastic nature of QCD radiation and differences between generative and supervised learning tasks in collider physics.

Amram, Oz [Fermilab]↗

Reinforcement learning-based design of shape-changing metamaterials

During the last decade, artificially architected materials have been designed to obtain properties unreachable by naturally occurring materials, whose properties are determined by their atomic structure and chemical composition. In this work, we implement a new reinforcement learning (RL) method able to rationally design unique metamaterial structures at the nano-, micro-, and macroscale, which change shape during operational conditions. As an example, we apply this method to design nanostructured silicon anodes for Li-ion batteries (LIBs). The RL model is designed to apply different actions and predict change during operational conditions. The multi-component reward function comprises an increase in the total storage capacity of the resulting battery electrode and structural parameters, such as the minimum distance between the individual components of the nanostructure. Upon experimental validation using a polymer-based 3D printing technique, we expect that the newly discovered structures improve the current Si-based LIB anodes state-of-the-art by almost three times and almost ten times the current commercial LIB based on a graphitic anode. Furthermore, this RL-based optimization method opens up vast design space for other responsive metamaterials with tailored properties and pre-programmed structural transformation.

25 ENERGY STORAGE↗

Confinement Effects on Proton Transfer in TiO 2 Nanopores from Machine Learning Potential Molecular Dynamics Simulations

Improved understanding of proton transfer in nanopores is critical for a wide range of emerging applications, yet experimentally probing mechanisms and energetics of this process remains a significant challenge. To help reveal details of this process, we developed and applied a machine learning potential derived from first-principles calculations to examine water reactivity and proton transfer in TiO 2 slit-pores. Here, we find that confinement of water within pores smaller than 0.5 nm imposes strong and complex effects on water reactivity and proton transfer. Although the proton transfer mechanism is similar to that at a TiO 2 interface with bulk water, confinement reduces the activation energy of this process, leading to more frequent proton transfer events. This enhanced proton transfer stems from the contraction of oxygen–oxygen distances dictated by the interplay between confinement and hydrophilic interactions. Our simulations also highlight the importance of the surface topology, where faster proton transport is found in the direction where a unique arrangement of surface oxygens enables the formation of an ordered water chain. In a broader context, our study demonstrates that proton transfer in hydrophilic nanopores can be enhanced by controlling pore size, surface chemistry, and topology.

36 MATERIALS SCIENCE↗

Self-learning control system for plug-in hybrid vehicles

A system is provided to instruct a plug-in hybrid electric vehicle how optimally to use electric propulsion from a rechargeable energy storage device to reach an electric recharging station, while maintaining as high a state of charge (SOC) as desired along the route prior to arriving at the recharging station at a minimum SOC. The system can include the step of calculating a straight-line distance and/or actual distance between an orientation point and the determined instant present location to determine when to initiate optimally a charge depleting phase. The system can limit extended driving on a deeply discharged rechargeable energy storage device and reduce the number of deep discharge cycles for the rechargeable energy storage device, thereby improving the effective lifetime of the rechargeable energy storage device. This "Just-in-Time strategy can be initiated automatically without operator input to accommodate the unsophisticated operator and without needing a navigation system/GPS input.

DeVault, Robert C↗

A fingerprint based metric for measuring similarities of crystalline structures

Measuring similarities/dissimilarities between atomic structures is important for the exploration of potential energy landscapes. However, the cell vectors together with the coordinates of the atoms, which are generally used to describe periodic systems, are quantities not directly suitable as fingerprints to distinguish structures. Based on a characterization of the local environment of all atoms in a cell, we introduce crystal fingerprints that can be calculated easily and define configurational distances between crystalline structures that satisfy the mathematical properties of a metric. This distance between two configurations is a measure of their similarity/dissimilarity and it allows in particular to distinguish structures. The new method can be a useful tool within various energy landscape exploration schemes, such as minima hopping, random search, swarm intelligence algorithms, and high-throughput screenings.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗

Composite Qdrift-product formulas for quantum and classical simulations in real and imaginary time

Recent study has shown that it can be advantageous to implement a composite channel that partitions the Hamiltonian H for a given simulation problem into subsets A and B such that H = A + B , where the terms in A are simulated with a Trotter-Suzuki channel and the B terms are randomly sampled via the Qdrift algorithm. Here we extend Qdrift and composite product formulas to imaginary time, formulating candidate classical algorithms for quantum Monte Carlo calculations. We upper bound the induced Schatten- 1 → 1 norm on both imaginary-time Qdrift and composite channels. Another recent result demonstrated that simulations of lattice Hamiltonians containing geometrically local interactions can be improved using a Lieb-Robinson argument to decompose H into subsets that contain only terms supported on that subset of the lattice. Here, we provide a quantum algorithm by unifying this result with the composite approach into “local composite channels” and we upper bound the diamond distance. We provide exact numerical simulations of algorithmic cost by counting the number of gates of the form e − i H j t and e − H j β to meet a certain error tolerance ε . In doing so, we optimize the partitioning into sets A and B using gradient boosted tree models from machine learning. These numerical studies are important given that product formulas have been historically known to outperform analytic upper bounds. We show constant factor advantages for a variety of interesting Hamiltonians, the maximum of which is a ≈ 20 -fold speedup that occurs in the simulation of Jellium. Published by the American Physical Society 2024

Pocrnic, Matthew (ORCID:0000000203089376)↗

Climate data for bioclimate research of Amazon Basin central region near the city of Manaus

The current research/data set deals with environmental data related to the region around the city of Manaus, Brazil, located in the Amazon Basin central region. The research problem encompasses the answers to the questions: 'What correlations can arise from the data of climate and fauna in the Amazon Basin?" "How useful is the data from instrumented stations related to that area, related to the fauna and climatic conditions?" To answer these questions, an initial data set was generated, taking into account the main pollutants with sources from NASA satellites, the Amazon Tall Tower Observatory (ATTO) experiment, and the GOAmazon project, from 2014 to 2015, along with other local data, such as the position of thermal electric power stations. It has been worth noting the data available was not coherent among the sources, due to different measuring time intervals, heights, and geographical locations. Equally important, for future bioclimatic research, a grid space of 5-km interval and a 30-minute time period was considered within the data set, covering a surface area within the boundaries of Latitude 2.00 S to 4.25 S and Longitude 59.00 to 61.25 W. The data from NASA satellites, ATTO, and GOAmazon do not have this kind of detailed grid and time mesh. To carry on, the machine learning (ML) Random Forest (RF) algorithm was selected and used, among other ML algorithms, to fill in the data into this new grid. The features considered were: data from NASA satellites and from the two projects above, covering the pollutants: ozone, NOx, particulates with diameter down to 10 nanometers, isoprene, CO, and acetonitrile. Moreover, distances and bearings to Manaus and 11 thermal power stations nearby, night and day time periods, dry and wet seasons, were also considered to compose the features. The interpolation produced by the Random Forest algorithm aimed at the pollutants: acetonitrile, ozone, NOx, Particulate N, isoprene, and CO, which are related to bioclimatic research commonly. In total, a data set was created with 51,148,800 rows, with 44 columns, 38 for the features and 6 for the targets.

54 ENVIRONMENTAL SCIENCES↗

The Effect of Different Operations Modes on Science Capabilities During the 2010 Desert-RATS Test: Insights from the Geologist Crewmembers

The 2010 Desert RATS field test utilized two Space Exploration Vehicles (prototype planetary rovers) and four crewmembers (2 per rover) to conduct a geologic traverse across northern Arizona while testing continuous and twice-per-day communications paired with operation modes of separating and exploring individually (Divide & Conquer) and exploring together (Lead & Follow), respectively. This report provides qualitative conclusions from the geologist crewmembers involved in this test as to how these modes of communications and operations affected our ability to conduct field geology. Each mode of communication and operation provided beneficial capabilities that might be further explored for future Human Spaceflight Missions to other solar system objects. We find that more frequent interactions between crews and an Apollo-style Science Team on the Earth best enables scientific progress during human exploration. However, during multiple vehicle missions, this communication with an Earth-based team of scientists, who represent "more minds on the problem", should not come at the exclusion of (or significantly decrease) communication between the crewmembers in different vehicles who have the "eyes on the ground". Inter-crew communications improved when discussions with a backroom were infrequent. Both aspects are critical and cannot be mutually exclusive. Increased vehicle separation distances best enable encounters with multiple geologic units. However, seemingly redundant visits by multiple vehicles to the same feature can be utilized to provide improved process-related observations about the development and modification of the local terrain. We consider the value of data management, transfer, and accessibility to be the most important lesson learned. Crews and backrooms should have access to all data and related interpretations within the mission in as close to real-time conditions as possible. This ensures that while on another planetary surface, crewmembers are as educated as possible with respect to the observations and data they will need to collect at any moment.

Bleacher, Jacob E.↗

Detecting Outliers in Network Transfers with Feature Extraction

Reliable file transfers are essential for successful science computations and experiments that need large complex data files to be moved across long distances. Networks use protocols such as TCP and UDP to support file transfers, and their performance degrades under packet losses and duplications, thereby adversely affecting science applications. Hence, scientists and engineers often monitor network statistics to find and repair network problems that cause such degradations.Outliers or anomalies can be detected using statistical and machine learning methods that highlight abnormal behaviors. However, without knowing what is normal, it is very difficult and subjective to determine the anomalies. In this paper, we investigate statistical and machine learning approaches to extract unknown or unsupervised features from TCP and perfSONAR data sets, with the aim to identify anomalies such as packet loss, duplication and retransmission sequencing that affect file transfer performance. Simple statistical and feature extraction methods (e.g. PCA) have shown that detecting anomalies is simpler if labeled data sets of normal behavior are available. While perfSONAR logs record losses,they do not detect duplications and reordering. We show that PCA applied to TCP statistics is able to determine abnormal behavior in the above cases. In particular, our results show that simple feature extraction techniques provide building blocks for more advanced network monitoring tools that can quickly flag anomalies using simple measurements, thereby helping to build confidence in file transfer reliability.

Rao, Nageswara S.↗

Disconnecting structure and dynamics in glassy thin films

Nanometrically thin glassy films depart strikingly from the behavior of their bulk counterparts. We investigate whether the dynamical differences between a bulk and thin film polymeric glass former can be understood by differences in local microscopic structure. Machine learning methods have shown that local structure can serve as the foundation for successful, predictive models of particle rearrangement dynamics in bulk systems. By contrast, in thin glassy films, we find that particles at the center of the film and those near the surface are structurally indistinguishable despite exhibiting very different dynamics. Next, we show that structure-independent processes, already present in bulk systems and demonstrably different from simple facilitated dynamics, are crucial for understanding glassy dynamics in thin films. Our analysis suggests a picture of glassy dynamics in which two dynamical processes coexist, with relative strengths that depend on the distance from an interface. Here, one of these processes depends on local structure and is unchanged throughout most of the film, while the other is purely Arrhenius, does not depend on local structure, and is strongly enhanced near the free surface of a film.

36 MATERIALS SCIENCE↗

Converting tabular data into images for deep learning with convolutional neural networks

Abstract Convolutional neural networks (CNNs) have been successfully used in many applications where important information about data is embedded in the order of features, such as speech and imaging. However, most tabular data do not assume a spatial relationship between features, and thus are unsuitable for modeling using CNNs. To meet this challenge, we develop a novel algorithm, image generator for tabular data (IGTD), to transform tabular data into images by assigning features to pixel positions so that similar features are close to each other in the image. The algorithm searches for an optimized assignment by minimizing the difference between the ranking of distances between features and the ranking of distances between their assigned pixels in the image. We apply IGTD to transform gene expression profiles of cancer cell lines (CCLs) and molecular descriptors of drugs into their respective image representations. Compared with existing transformation methods, IGTD generates compact image representations with better preservation of feature neighborhood structure. Evaluated on benchmark drug screening datasets, CNNs trained on IGTD image representations of CCLs and drugs exhibit a better performance of predicting anti-cancer drug response than both CNNs trained on alternative image representations and prediction models trained on the original tabular data.

59 BASIC BIOLOGICAL SCIENCES↗

Reducing ridesourcing empty vehicle travel with future travel demand prediction

Ridesourcing services provide alternative mobility options in several cities. Their market share has grown exponentially due to the convenience they provide. The use of such services may be associated with car-light or car-free lifestyles. However, there are growing concerns regarding their impact on urban transportation operations performance due to empty, unproductive miles driven without a passenger (commonly referred to as deadheading). This paper is motivated by the potential to reduce deadhead mileage of ridesourcing trips by providing drivers with information on future ridesourcing trip demand. Future demand information enables the driver to wait in place for the next rider’s request without cruising around and contributing to congestion. A machine learning model is employed to predict hourly and 10-minute future interval travel demand for ridesourcing at a given location. Using future demand information, we propose algorithms to (i) assign drivers to act on received demand information by waiting in place for the next rider, and (ii) match these drivers with riders to minimize deadheading distance. Real-world data from ridesourcing providers in Austin, TX (RideAustin) and Chengdu, China (DiDi Chuxing) are leveraged. Results show that this process achieves 68%–82% and 53%–60% reduction of trip-level deadheading miles for the RideAustin and DiDi Chuxing sample operations respectively, under the assumption of unconstrained availability of short-term parking. Deadheading savings increase slightly as the maximum tolerable waiting time for the driver increases. Further, it is observed that significant deadhead savings per trip are possible, even when a small percent of the ridesourcing driver pool is provided with future ridesourcing demand information.

32 ENERGY CONSERVATION, CONSUMPTION, AND UTILIZATI↗

Ricci focusing, shearing, and the expansion rate in an almost homogeneous Universe

The Universe is inhomogeneous, and yet it seems to be incredibly well-characterised by a homogeneous relativistic model. One of the current challenges is to accurately characterise the properties of such a model. In this paper we explore how inhomogeneities may affect the overall optical properties of the Universe by quantifying how they can bias the redshift-distance relation in a number of toy models that mimic the real Universe. The models that we explore are statistically homogeneous on large scales. We find that the effect of inhomogeneities is of order of a few percent, which can be quite important in precise estimation of cosmological parameters. We discuss what lessons can be learned to help us tackle a more realistic inhomogeneous universe.

79 ASTRONOMY AND ASTROPHYSICS↗

Machine-learned Identification of RR Lyrae Stars from Sparse, Multi-band Data: The PS1 Sample

RR Lyrae stars may be the best practical tracers of Galactic halo (sub-)structure and kinematics. The PanSTARRS1 (PS1) 3π survey offers multi-band, multi-epoch, precise photometry across much of the sky, but a robust identification of RR Lyrae stars in this data set poses a challenge, given PS1's sparse, asynchronous multi-band light curves (≲12 epochs in each of five bands, taken over a 4.5 year period). We present a novel template fitting technique that uses well-defined and physically motivated multi-band light curves of RR Lyrae stars, and demonstrate that we get accurate period estimates, precise to 2 s in >80% of cases. We augment these light-curve fits with other features from photometric time-series and provide them to progressively more detailed machine-learned classification models. From these models, we are able to select the widest (three-fourths of the sky) and deepest (reaching 120 kpc) sample of RR Lyrae stars to date. The PS1 sample of ∼45,000 RRab stars is pure (90%) and complete (80% at 80 kpc) at high galactic latitudes. It also provides distances that are precise to 3%, measured with newly derived period–luminosity relations for optical/near-infrared PS1 bands. With the addition of proper motions from Gaia and radial velocity measurements from multi-object spectroscopic surveys, we expect the PS1 sample of RR Lyrae stars to become the premier source for studying the structure, kinematics, and the gravitational potential of the Galactic halo. The techniques presented in this study should translate well to other sparse, multi-band data sets, such as those produced by the Dark Energy Survey and the upcoming Large Synoptic Survey Telescope Galactic plane sub-survey.

79 ASTRONOMY AND ASTROPHYSICS↗

Lessons Learned from Previous Space-Borne Sounders as a Guide to Future Sounder Development

Space-borne radio sounding is considered to be the gold standard for electron-density (N(sub e)) measurements compared to other techniques even under low-density conditions, such as N(sub e) < 1/cu cm, when other techniques are known to experience difficulties. These reliable measurements are not restricted to in-situ N(sub e) determinations since a spaceborne sounder can provide vertical N(sub e) profiles (N(sub e)(h)) from the spacecraft altitude to the altitude of maximum N(sub e). Near-conjunction studies involving the International Satellites for Ionospheric Studies (ISIS) satellites in the topside ionosphere and Dynamics Explorer 2 (DE 2) near the altitude of the F-region peak density have verified that, even at the greatest distance from the sounder, the ISIS-derived N(sub e)(h) profiles agree with the DE-2 Langmuir-probe measurements to within about 30% over a density range of more than two decades. Space-borne sounders can also provide N(sub e) profiles along the magnetic-field B, by inverting echoes that are ducted along field-aligned irregularities (FAI), and can provide information about the terrain beneath the satellite by examining surface reflections in the frequency range above the ionospheric penetration frequency. Many nations have launched rocket and satellite radio sounders in geospace over more than 4 decades and there have been sounders on space-probes and in orbit around other planets. Here we will summarize some of the lessons learned from these accomplishments by analyzing data from radio sounders on the Alouette and ISIS satellites and the OEDIPUS and other rockets in the terrestrial ionosphere, the IMAGE satellite in the terrestrial magnetosphere, the Ulysses space probe in Jupiter's 10 plasma torus and the MARSIS satellite in orbit around Mars. The emphasis will be on information deduced concerning (1) fundamental plasma processes and gradients in N, and B in the vicinity of the sounders from sounder-stimulated plasma resonances and short-range echoes involving ion as well as electron motions, (2) the importance of the antenna orientation relative to B for the detection of different plasma resonances, (3) sounder-stimulated plasma phenomena, including FAI, when special plasma conditions are satisfied, (4) the minimum power required for long-range echoes, as indicated by echoes from frequency components of the transmitted pulse and by multiple ducted echoes, and (5) the terrain beneath the satellite from surface reflections. Knowledge of these results should enable the optimum design of a future sounder to satisfy specific science requirements with minimal spacecraft resources.

Benson, Robert F.↗

Accurate prediction of global-density-dependent range-separation parameters based on machine learning

In this work, we develop an accurate and efficient XGBoost machine learning model for predicting the global-density-dependent range-separation parameter, ωGDD, for long-range corrected functional (LRC)-ωPBE. This ωGDDML model has been built using a wide range of systems (11 466 complexes, ten different elements, and up to 139 heavy atoms) with fingerprints for the local atomic environment and histograms of distances for the long-range atomic correlation for mapping the quantum mechanical range-separation values. The promising performance on the testing set with 7046 complexes shows a mean absolute error of 0.001 117 a0−1 and only five systems (0.07%) with an absolute error larger than 0.01 a0−1, which indicates the good transferability of our ωGDDML model. In addition, the only required input to obtain ωGDDML is the Cartesian coordinates without electronic structure calculations, thereby enabling rapid predictions. LRC-ωPBE(ωGDDML) is used to predict polarizabilities for a series of oligomers, where polarizabilities are sensitive to the asymptotic density decay and are crucial in a variety of applications, including the calculations of dispersion corrections and refractive index, and surpasses the performance of all other popular density functionals except for the non-tuned LRC-ωPBE. Finally, LRC-ωPBE (ωGDDML) combined with (extended) symmetry-adapted perturbation theory is used in calculating noncovalent interactions to further show that the traditional ab initio system-specific tuning procedure can be bypassed. The present study not only provides an accurate and efficient way to determine the range-separation parameter for LRC-ωPBE but also shows the synergistic benefits of fusing the power of physically inspired density functional LRC-ωPBE and the data-driven ωGDDML model.

Chemistry↗

Exploration Medical Capability Clinical Decision Support System Architecture Recommendation

A new era in space exploration has arrived with the goal of establishing a long-term presence on the Moon and using those learning to take the next giant leap: sending the first astronauts to Mars. These ambitious goals will require significant changes in in-flight medical care due to constraints on mass, volume, power, crew time, and medical evacuation capabilities. Furthermore, while crews currently rely on real-time communications with ground-based medical providers, as distance from Earth increases, so do communication delays and disruptions. The crew will need to autonomously detect, diagnose, treat, and prevent medical events. These constraints require development of transformative solutions and new technologies. Through participation in the Human Research Program’s (HRP) Exploration Medical Capability (ExMC) research, NASA has developed and demonstrated a Medical Data Architecture (MDA), a platform upon which a robust capability of Clinical Decision Support (CDS) can be built. As a result, NASA is now positioned to build an advanced autonomous CDS System (CDSS) which will aid in crew health decision-making. The CDSS combines data management aspects of a system that would lead to the autonomy required by NASA-STD-3001, Section 3, Health and Medical Care Standards. The project addresses the ExMC gap Medical-701: We need to increase inflight medical capabilities and identify new capabilities that (a) maximize benefit and/or (b) reduce “costs” on human system/mission/vehicle resources. This architecture recommendation document introduces an integrated data architecture that provides self-sufficient medical care for crews on long-duration spaceflight missions.

ExMC↗

Graphical Gaussian Process Regression Model for Aqueous Solvation Free Energy Prediction of Organic Molecules in Redox Flow Battery

The solvation free energy of organic molecules is a critical parameter in determining emergent properties such as solubility, liquid-phase equilibrium constants, and pKa and redox potentials in an organic redox flow battery. In this work, we present a machine learning (ML) model that can learn and predict the aqueous solvation free energy of an organic molecule using Gaussian process regression method based on a new molecular graph kernel. To investigate the performance of the ML model on electrostatic interaction, the nonpolar interaction contribution of solvent and the conformational entropy of solute in solvation free energy, three data sets with implicit or explicit water solvent models, and contribution of conformational entropy of solute are tested. We demonstrate that our ML model can predict the solvation free energy of molecules at chemical accuracy with a mean absolute error of less than 1 kcal/mol for subsets of the QM9 dataset and the Freesolv database. To solve the general data scarcity problem for a graph-based ML model, we propose a dimension reduction algorithm based on the distance between molecular graphs, which can be used to examine the diversity of the molecular data set. It provides a promising way to build a minimum training set to improve prediction for certain test sets where the space of molecular structures is predetermined.

25 ENERGY STORAGE↗