Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Gradient information”

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 649 records · Page 36
All sourcesNASA NTRSDOE OSTI
DOE OSTI2026

Geothermal Heat Pump System Showcase: Short-Term Validation of Borehole Heat Exchanger Performance from Field Data to Numerical Modeling: Preprint

Since 2011, a geothermal heat pump (GHP) system has been operating to provide space heating and cooling for the Solar Radiation and Research Laboratory building at the National Laboratory of the Rockies (NLR) in Golden, Colorado. The system consists of 23 vertical boreholes, each extending to a depth of 300 ft (91 m), connected to 11 water-to-air heat pump units and four circulation pumps. Between fiscal years 2023 and 2025, additional power meters and temperature sensors were retrofitted to support detailed system performance assessment and model development. This study presents preliminary monitoring results and the development of an initial numerical model of the borehole heat exchanger field. The model incorporated site-specific geometry, ground thermal properties derived from thermal response tests, and ambient temperatures, and simulated system behavior over a representative operating day in September. Model predictions of outlet temperatures were compared against corresponding field measurements. Results showed that modeling initialized with a simplified linear subsurface temperature gradient presents systematic discrepancies in outlet temperature, whereas incorporating depth-resolved borehole temperature measurements for initialization yields substantially improved agreement with observations. The findings highlight the sensitivity of short-term predictive modeling to the representation of initial subsurface thermal conditions and underscore the value of high-resolution field measurements for model calibration and validation. These preliminary results inform ongoing efforts to extend the modeling framework to longer time horizons and to refine monitoring and modeling strategies that support the design guidance and operational optimization of GHP systems in research and commercial buildings.

15 GEOTHERMAL ENERGY↗
DOE OSTI2026

HfTi_BCC_SolidSolution_128atoms_VASP6

We performed density functional theory (DFT) calculations for body-centered-cubic (BCC) structures with 128 lattices sites of solid solution binary alloys hafnium-titanium (Hf-Ti). The electronic structures of alloys have been calculated using Vienna Ab initio Simulation Package (VASP). Within this package the DFT approach is used to reduce many-body Schrodinger equation to set of single particle Kohn-Sham (KS) equations. The generalized electronic exchange-correlation functional is described by generalized gradient approximation with the Perdew-Burke-Ernzerhof parametrization. The electron-ion interactions is described by pseudopotentials developed within the plane-wave basis projector augmented-wave (PAW) approach. These pseudopotentials are available at the VASP portal (http://cms.mpi.univie.ac.at/vasp/). Our calculations have been run with the pseudopotentials treating s and p semi-core states as valence in case for the elements Hf and Ti. The electronic densities and potentials are expanded over plane-waves with energy cutoff of 350 eV. 2x2x2 k-mesh and normal precision were used. The alloys were modeled by supercell containing 128 randomly distributed atoms. At initial step the atoms occupy perfect bcc lattice cites. This initial structure was optimized until energy changes less than 1e-6 eV, while forces acting on atoms don't exceed 1e-2 eV/angstrom. The electron-ion interaction is described by PAW pseudopotentials. The calculations have been collected by sampling chemical compositions across the entire compositional range. The chemical compositions have been sampled by progressively changing the number of atoms per constituent by 4. For each chemical composition of binaries and ternaries, the first-principle calculations have been run for 100 randomized arrangements of the constituents on the BCC lattice sites. We collected data for a total of 3,100 randomized atomic structures over 31 chemical compositions. The calculations have been collected on Air Force HPC11 cluster using the VASP 6.5.1. Additional methodology and file structure information is available in the dataset README.txt file.

36 MATERIALS SCIENCE↗
DOE OSTI2026

TiV_BCC_SolidSolution_128atoms_VASP6

We performed density functional theory (DFT) calculations for body-centered-cubic (BCC) structures with 128 lattices sites of solid solution binary alloys titanium-vanadium (Ti-V). The electronic structures of alloys have been calculated using Vienna Ab initio Simulation Package (VASP). Within this package the DFT approach is used to reduce many-body Schrodinger equation to set of single particle Kohn-Sham (KS) equations. The generalized electronic exchange-correlation functional is described by generalized gradient approximation with the Perdew-Burke-Ernzerhof parametrization. The electron-ion interactions is described by pseudopotentials developed within the plane-wave basis projector augmented-wave (PAW) approach. These pseudopotentials are available at the VASP portal (http://cms.mpi.univie.ac.at/vasp/). Our calculations have been run with the pseudopotentials treating s and p semi-core states as valence in case for the elements Ti and V. The electronic densities and potentials are expanded over plane-waves with energy cutoff of 350 eV. 2x2x2 k-mesh and normal precision were used. The alloys were modeled by supercell containing 128 randomly distributed atoms. At initial step the atoms occupy perfect bcc lattice cites. This initial structure was optimized until energy changes less than 1e-6 eV, while forces acting on atoms don't exceed 1e-2 eV/angstrom. The electron-ion interaction is described by PAW pseudopotentials. The calculations have been collected by sampling chemical compositions across the entire compositional range. The chemical compositions have been sampled by progressively changing the number of atoms per constituent by 4. For each chemical composition of binaries and ternaries, the first-principle calculations have been run for 100 randomized arrangements of the constituents on the BCC lattice sites. We collected data for a total of 3,100 randomized atomic structures over 31 chemical compositions. The calculations have been collected on Air Force HPC11 cluster using the VASP 6.5.1. Additional methodology and file structure information is available in the dataset README.txt file.

36 MATERIALS SCIENCE↗
DOE OSTI2024

Choosing the Right Tool: A Comparative Study of Wetland Assessment Approaches

There are over 700 aquatic ecological assessment approaches across the globe that meet specific institutional goals. However, in many cases, multiple assessment tools are designed to meet the same management need, resulting in a confusing array of overlapping options. Here, we look at six riverine wetland assessments currently in use in Montana, USA, and ask which tool (1) best captures the condition across a disturbance gradient and (2) has the most utility to meet the regulatory or management needs. We used descriptive statistics to compare wetland assessments (n = 18) across a disturbance gradient determined by a landscape development intensity. Factor analysis showed that many of the tools had internal metrics that did not correspond well with overall results, hindering the tool’s ability to act as designed. We surveyed regional wetland managers (n = 56) to determine the extent of their use of each of the six tools and how well they trusted the information the assessment tool provided. We found that the Montana Wetland Assessment Methodology best measured the range of disturbance and had the highest utility to meet Clean Water Act (CWA§ 404) needs. Montana Department of Environmental Quality was best for the CWA§ 303(d) & 305(b) needs. The US Natural Resources Conservation Service’s Riparian Assessment Tool was the third most used by managers but was the tool that had the least ability to distinguish across a disturbance, followed by the US Bureau of Land Management’s Proper Functioning Condition.

54 ENVIRONMENTAL SCIENCES↗
DOE OSTI2020

Advances in Kriging-Based Autonomous X-Ray Scattering Experiments

Autonomous experimentation is an emerging paradigm for scientific discovery, wherein measurement instruments are augmented with decision-making algorithms, allowing them to autonomously explore parameter spaces of interest. We have recently demonstrated a generalized approach to autonomous experimental control, based on generating a surrogate model to interpolate experimental data, and a corresponding uncertainty model, which are computed using a Gaussian process regression known as ordinary Kriging (OK). We demonstrated the successful application of this method to exploring materials science problems using x-ray scattering measurements at a synchrotron beamline. Here, we report several improvements to this methodology that overcome limitations of traditional Kriging methods. The variogram underlying OK is global and thus insensitive to local data variation. We augment the Kriging variance with model-based measures, for instance providing local sensitivity by including the gradient of the surrogate model. As with most statistical regression methods, OK minimizes the number of measurements required to achieve a particular model quality. However, in practice this may not be the most stringent experimental constraint; e.g. the goal may instead be to minimize experiment duration or material usage. We define an adaptive cost function, allowing the autonomous method to balance information gain against measured experimental cost. We provide synthetic and experimental demonstrations, validating that this improved algorithm yields more efficient autonomous data collection.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Periodic Narrowband Radio Wave Emissions and Inward Plasma Transport at Saturn's Magnetosphere

The abrupt brightening of an Energetic Neutral Atom (ENA) blob or cloud has been interpreted as plasma injection in the Kronian magnetosphere (termed ENA injection herein). Morphologically, there appears to be two types of abrupt ENA cloud brightening: (1) the brightening of a large cloud usually seen at distances >10–12 R {sub s} (R {sub s} ∼ 60,268 km) in the midnight or postmidnight region; (2) the brightening of a smaller cloud usually seen at distances <10–12 R {sub s} around 21–03 magnetic local time. Among many radio waves observed at Saturn, type 2 ENA injections correlate best with the 5 kHz narrowband (NB) waves. Using Cassini Ion and Neutral Camera (INCA) and Radio and Plasma Wave Science (RPWS) data, we examine the periodicities of the type 2 ENA injections and the 5 kHz NB emissions as well as their cross-correlations, which have been previously used to measure the lag times or phase differences. Because correlational analysis can only establish linear relationships, we also use mutual information to establish linear and nonlinear relationships. On average, the peak of the 5 kHz NB emission lags those of the type 2 ENA injection by about a few minutes to 2 hr. The injection of hot plasma to the inner magnetosphere can lead to temperature anisotropy, which can generate electrostatic upper hybrid waves, which upon encountering the high-density gradient at the outer edge of the Enceladus density torus, can mode convert to Z mode and then to O mode. The 5 kHz NB waves commonly propagate in the O mode.

79 ASTRONOMY AND ASTROPHYSICS↗
NASA NTRS1962

Ionospheric Studies with the Differential Doppler Technique

This report is essentially a review of the measurements related to the ionosphere utilizing the differential Doppler technique. A brief study of the theory is presented, in which the error incurred in measurements of ionospheric electron content, caused by the assumption of common-path propagation, is shown as a function of the zenith angle of observation for different values of the critical frequency foF2. Some measurements of electron densities, electron content, horizontal gradients, irregularities, and magnetic-storm effects are discussed. Comments are also made about the method of combining the differential Doppler with the Faraday rotation measurements to establish a necessary constant of integration in the measurements made with satellites. It is concluded that the differential Doppler technique, with harmonically related frequencies transmitted to or from rockets or satellites, can provide accurate information about the electron density or electron content of the medium, and that the knowledge of these quantities is valuable in understanding the morphology of the ionospheric processes.

IONOSPHERE↗
NASA NTRS1994

A microwave technique for mapping thin sea ice

A technique is presented for mapping the distribution of new, young and first-year sea ice in seasonal sea ice zones that utilizes microwave spectral and polarization information from the Defense Meteorological Satellite Program Special Sensor Microwave/Imager (DMSP SSM/I). The motivation for this work stems from the need for accurate estimates of open water and thin ice within the Arctic ice pack. The technique utilizes the microwave polarization and spectral characteristics of these three ice types through two microwave radiance ratios: the 19.4 GHz polarization and the spectral gradient ratio, which is a measure of the spectral difference between the 19.4-GHz and the 37.0-GHz vertically polarized radiance components. The combined use of the spectral gradient ratio and polarization reduces the low ice concentration bias generally associated with the presence of thin ice types. The microwave polarization, which is sensitive to changes in ice thickness and ice surface characteristics, is used to classify new, young, and first-year ice types.

Cavalieri, Donald J.↗
DOE OSTI2026

NbTi_BCC_SolidSolution_128atoms_VASP6

We performed density functional theory (DFT) calculations for body-centered-cubic (BCC) structures with 128 lattices sites of solid solution binary alloys niobium-titanium (Nb-Ti). The electronic structures of alloys have been calculated using Vienna Ab initio Simulation Package (VASP), https://github.com/ORNL/ScalableWorkflow_VASP_Calculations. Within this package the DFT approach is used to reduce many-body Schrodinger equation to set of single particle Kohn-Sham (KS) equations. The generalized electronic exchange-correlation functional is described by generalized gradient approximation with the Perdew-Burke-Ernzerhof parametrization. The electron-ion interactions is described by pseudopotentials developed within the plane-wave basis projector augmented-wave (PAW) approach. These pseudopotentials are available at the VASP portal (http://cms.mpi.univie.ac.at/vasp/). Our calculations have been run with the pseudopotentials treating s and p semi-core states as valence in case for the elements Nb and Ti. The electronic densities and potentials are expanded over plane-waves with energy cutoff of 350 eV. 2x2x2 k-mesh and normal precision were used. The alloys were modeled by supercell containing 128 randomly distributed atoms. At initial step the atoms occupy perfect bcc lattice cites. This initial structure was optimized until energy changes less than 1e-6 eV, while forces acting on atoms don't exceed 1e-2 eV/angstrom. The electron-ion interaction is described by PAW pseudopotentials. The calculations have been collected by sampling chemical compositions across the entire compositional range. The chemical compositions have been sampled by progressively changing the number of atoms per constituent by 4. For each chemical composition of binaries and ternaries, the first-principle calculations have been run for 100 randomized arrangements of the constituents on the BCC lattice sites. We collected data for a total of 3,100 randomized atomic structures over 31 chemical compositions. The calculations have been collected on Air Force HPC11 cluster using the VASP 6.5.1. Additional methodology and file structure information is available in the dataset README.txt file.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Barren Plateaus Preclude Learning Scramblers

Scrambling processes, which rapidly spread entanglement through many-body quantum systems, are difficult to investigate using standard techniques, but are relevant to quantum chaos and thermalization. In this Letter, we ask if quantum machine learning (QML) could be used to investigate such processes. In this study we prove a no-go theorem for learning an unknown scrambling process with QML, showing that it is highly probable for any variational Ansatz to have a barren plateau landscape, i.e., cost gradients that vanish exponentially in the system size. This implies that the required resources scale exponentially even when strategies to avoid such scaling (e.g., from Ansatz -based barren plateaus or no-free-lunch theorems) are employed. Furthermore, we numerically and analytically extend our results to approximate scramblers. Hence, our work places generic limits on the learnability of unitaries when lacking prior information.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2020

Null Space Monte Carlo Evaluation of the Plateau to River Model

The Plateau to River Groundwater Model (P2R Model) is a groundwater flow and contaminant fate and transport (F&T) simulation model used to support remedial activities conducted by CH2M HILL Plateau Remediation Company at the Hanford Site in Washington State. Figure 1-1 illustrates the P2R Model extents, discretization, and boundary conditions. The P2R Model provides a computational framework to simulate the F&T of contaminants in groundwater associated with the 200-PO-1, 200-UP-1, 200-BP-5, and 200-ZP-1 Groundwater Operable Units (OUs) in the Hanford Site Central Plateau. In addition, the model includes adjacent areas and facilities (e.g., the State Approved Land Disposal Site). Intended and anticipated uses of the model include calculating water levels, hydraulic gradients, and groundwater flows throughout the model domain (encompassing the 200 West and 200 East Areas) for use in subsequent F&T calculations for contaminants of concern and developing scale-appropriate, telescopic-mesh refinement models for detailed evaluation of areas within the model domain where required. The overall objective of the modeling effort is to provide a basis for making informed remedial action decisions based on descriptions of current and expected future contaminant concentrations in groundwater at decision points within the OU boundaries. The objective for the model development phase is to create a common modeling platform that can be used for investigations of the Central Plateau groundwater OUs and areas downgradient toward the Columbia River. The P2R Model calibration to historical data observed at the Hanford Site is documented in CP-57037, Model Package Report for the Plateau to River Model Version 8.3. The purpose of this environmental calculation is to describe a null space Monte Carlo (NSMC) evaluation was conducted with the historic calibration of the P2R model. Use of numerical groundwater models is always accompanied with uncertainty in the results produced by a model because models are approximations of reality. Thus, by definition, lack the detail to fully represent observed behavior. Use of numerical techniques, such as a NSMC analysis, can help in identifying and quantifying the potential uncertainties associated with a numerical model such as the P2R Model. Use of the NSMC approach results in 100 groundwater flow models that are variants of the calibrated P2R Model. These variant models can be used to evaluate uncertainty in model predictions made by the calibrated P2R Model for other analyses. A secondary purpose of the environmental calculation is to establish these variant models for use with other applications.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗
DOE OSTI2024

A Multiphysics Thermoelastoviscoplastic Damage Internal State Variable Constitutive Model including Magnetism

We present a macroscale constitutive model that couples magnetism with thermal, elastic, plastic, and damage effects in an Internal State Variable (ISV) theory. Previous constitutive models did not include an interdependence between the internal magnetic (magnetostriction and magnetic flux) and mechanical fields. Although constitutive models explaining the mechanisms behind mechanical deformations caused by magnetization changes have been presented in the literature, they mainly focus on nanoscale structure–property relations. A fully coupled multiphysics macroscale ISV model presented herein admits lower length scale information from the nanoscale and microscale descriptions of the multiphysics behavior, thus capturing the effects of magnetic field forces with isotropic and anisotropic magnetization terms and moments under thermomechanical deformations. For the first time, this ISV modeling framework internally coheres to the kinematic, thermodynamic, and kinetic relationships of deformation using the evolving ISV histories. For the kinematics, a multiplicative decomposition of deformation gradient is employed including a magnetization term; hence, the Jacobian represents the conservation of mass and conservation of momentum including magnetism. The first and second laws of thermodynamics are used to constrain the appropriate constitutive relations through the Clausius–Duhem inequality. The kinetic framework employs a stress–strain relationship with a flow rule that couples the thermal, mechanical, and magnetic terms. Experimental data from the literature for three different materials (iron, nickel, and cobalt) are used to compare with the model’s results showing good correlations.

36 MATERIALS SCIENCE↗
DOE OSTI2024

A Multiphysics Thermoelastoviscoplastic Damage Internal State Variable Constitutive Model including Magnetism

We present a macroscale constitutive model that couples magnetism with thermal, elastic, plastic, and damage effects in an Internal State Variable (ISV) theory. Previous constitutive models did not include an interdependence between the internal magnetic (magnetostriction and magnetic flux) and mechanical fields. Although constitutive models explaining the mechanisms behind mechanical deformations caused by magnetization changes have been presented in the literature, they mainly focus on nanoscale structure–property relations. A fully coupled multiphysics macroscale ISV model presented herein admits lower length scale information from the nanoscale and microscale descriptions of the multiphysics behavior, thus capturing the effects of magnetic field forces with isotropic and anisotropic magnetization terms and moments under thermomechanical deformations. For the first time, this ISV modeling framework internally coheres to the kinematic, thermodynamic, and kinetic relationships of deformation using the evolving ISV histories. For the kinematics, a multiplicative decomposition of deformation gradient is employed including a magnetization term; hence, the Jacobian represents the conservation of mass and conservation of momentum including magnetism. The first and second laws of thermodynamics are used to constrain the appropriate constitutive relations through the Clausius–Duhem inequality. The kinetic framework employs a stress–strain relationship with a flow rule that couples the thermal, mechanical, and magnetic terms. Experimental data from the literature for three different materials (iron, nickel, and cobalt) are used to compare with the model’s results showing good correlations.

36 MATERIALS SCIENCE↗
NASA NTRS1981

VISSR Atmospheric Sounder (VAS) simulation experiment for a severe storm environment

Radiance fields were simulated for prethunderstorm environments in Oklahoma to demonstrate three points: (1) significant moisture gradients can be seen directly in images of the VISSIR Atmospheric Sounder (VAS) channels; (2) temperature and moisture profiles can be retrieved from VAS radiances with sufficient accuracy to be useful for mesoscale analysis of a severe storm environment; and (3) the quality of VAS mesoscale soundings improves with conditioning by local weather statistics. The results represent the optimum retrievability of mesoscale information from VAS radiance without the use of ancillary data. The simulations suggest that VAS data will yield the best soundings when a human being classifies the scene, picks relatively clear areas for retrieval, and applies a "local" statistical data base to resolve the ambiguities of satellite observations in favor of the most probable atmospheric structure.

Chesters, D.↗
NASA NTRS2001

Dynamics of the spatial scale of visual attention revealed by brain event-related potentials

The temporal dynamics of the spatial scaling of attention during visual search were examined by recording event-related potentials (ERPs). A total of 16 young participants performed a search task in which the search array was preceded by valid cues that varied in size and hence in precision of target localization. The effects of cue size on short-latency (P1 and N1) ERP components, and the time course of these effects with variation in cue-target stimulus onset asynchrony (SOA), were examined. Reaction time (RT) to discriminate a target was prolonged as cue size increased. The amplitudes of the posterior P1 and N1 components of the ERP evoked by the search array were affected in opposite ways by the size of the precue: P1 amplitude increased whereas N1 amplitude decreased as cue size increased, particularly following the shortest SOA. The results show that when top-down information about the region to be searched is less precise (larger cues), RT is slowed and the neural generators of P1 become more active, reflecting the additional computations required in changing the spatial scale of attention to the appropriate element size to facilitate target discrimination. In contrast, the decrease in N1 amplitude with cue size may reflect a broadening of the spatial gradient of attention. The results provide electrophysiological evidence that changes in the spatial scale of attention modulate neural activity in early visual cortical areas and activate at least two temporally overlapping component processes during visual search.

Non-NASA Center↗
NASA NTRS1937

The Transition Phase in the Take-Off of an Airplane, Special Report

An investigation was undertaken to determine the character and importance of the transition phase between the ground run and steady climb in the takeoff of an airplane and the effects of various factors on this phase and on the airborne part of the takeoff as a whole. The information was obtained from a series of step-by-step integrations, which defined the motion of the airplane during the transition and which were based on data derived from actual takeoff tests of a Verville AT airplane. Both normal and zoom takeoffs under several loading and takeoff speed conditions were considered. The effects of a moderate wind with a corresponding wind gradient and the effect of proximity of the ground were also investigated. The results show that, for normal takeoffs, the best transition was realized at the lowest possible takeoff speed. Moreover, this speed gave the shortest overall takeoff distance for normal takeoffs. Zoom takeoffs required a shorter overall takeoff run than normal takeoffs, particularly with a heavy landing, if the obstacle to be cleared was sufficiently high (greater than 50 feet); no advantage was indicated to the airplane with a light loading if the height to be cleared was less. The error resulting from the neglect of the transition in the calculation of the airborne distance of takeoff was found to vary from 4% with the heaviest loading considered to -4% with the lightest loading for normal takeoffs over a 100-ft obstacle; the percentage error was twice as great for a 50-foot obstacle. For zoom takeoffs the error attained much greater values. The average wind gradient corresponding to a 5-mile-per-hour surface wind reduced the airborne distance required to clear a 50-foot obstacle by about 9% with the lightest loading and 16% with the heaviest loading; for both cases. The overall reduction due to this wind was approximately twice that resulting from the wind gradient alone. A simple expression for the reduction of observed takeoff performance to no-wind conditions is presented. Ground effect is shown to reduce the airborne distance to attain a height of 50 foot by 10% with the lightest loading and 16% with the heaviest loading; for a 100-foot obstacle the percentage reduction was about 1/2 as great.

Wetmore, J. W.↗
NASA NTRSDate not supplied

Predicting Airport Runway Configurations for Decision-Support Using Supervised Learning

One of the most challenging tasks for air traffic controllers is runway configuration management (RCM). It deals with the optimal selection of runways to operate on (for arrivals and departures) based on traffic, surface wind speed, wind direction, other environmental variables, noise constraints, and several other airport-specific factors. It affects the efficiency of the National Airspace System (NAS) and both surface and airspace operations can benefit from better understanding future runway configurations. In this paper, we present a comprehensive implementation of predictive models for runway configuration estimation from large volumes of historical data. Specifically, operational data from two full years (2018 and 2019) is collected, analyzed, and fused together to build the data product used in this work. The data set differs from prior work in the field in terms of its scope, resolution, and variety of factors collected and considered. Meteorological data is collected from two different sources – current weather conditions from METAR (Meteorological Terminal Aviation Routine Weather Report) and forecast weather conditions from Localized Aviation MOS Program (LAMP). Operational data from the Federal Aviation Administration (FAA) Aviation System Performance Metrics (ASPM) related to scheduled and actual number of arrivals and departures, average taxi times, etc. are collected. NASA’s Sherlock Data Warehouse is used to identify critical information such as go-arounds, and other events that might impact RCM decision-making. All data is collected and aggregated over 15-minute intervals throughout the two years. This provides a resolution like the timescales that might be necessary for runway configuration management decision-making. A variety of supervised learning algorithms are tested including Support Vector Machine, Random Forest, Gradient Boosting, etc. including tuning of the model hyperparameters. The modeling process is applied and presented on two representative U.S. airports – Charlotte Douglas International Airport (KCLT) and Denver International Airport (KDEN). The two airports present different levels of complexity in terms of the total number of configurations used and provide a balanced perspective on the generalizability of the developed approach to other airports in the NAS. Initial results are promising (F1 score of 0.91 at KCLT and 0.83 at KDEN) for data in the test set. The final paper will contain a comprehensive comparison between different models and model building strategies as well as further refined results. Most important predictors for each airport will be identified along with a discussion and recommendations on adapting the framework to other scenarios.

Tejas G Puranik↗
DOE OSTI2024

Speeding up and reducing memory usage for scientific machine learning via mixed precision

Scientific machine learning (SciML) has emerged as a versatile approach to address complex computational science and engineering problems. Within this field, physics-informed neural networks (PINNs) and deep operator networks (DeepONets) stand out as the leading techniques for solving partial differential equations by incorporating both physical equations and experimental data. However, training PINNs and DeepONets require significant computational resources, including long computational times and large amounts of memory. In search of computational efficiency, training neural networks using half precision (float16) rather than the conventional single (float32) or double (float64) precision has gained substantial interest, given the inherent benefits of reduced computational time and memory consumed. However, we find that float16 cannot be applied to SciML methods, because of gradient divergence at the start of training, weight updates going to zero, and the inability to converge to a local minima. To overcome these limitations, we explore mixed precision, which is an approach that combines the float16 and float32 numerical formats to reduce memory usage and increase computational speed. Our experiments showcase that mixed precision training not only substantially decreases training times and memory demands but also maintains model accuracy. Here, we also reinforce our empirical observations with a theoretical analysis. The research has broad implications for SciML in various computational applications.

97 MATHEMATICS AND COMPUTING↗