Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “density estimation”

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 1,297 records · Page 72
All sourcesNASA NTRSDOE OSTI
DOE OSTI2021

Screening of metal complexes and organic solvents using the COSMOSAC-LANL model to enhance the energy density in a non-aqueous redox flow cell: an insight into the solubility

In this paper, we have proposed a first-principles methodology to screen transition metal complexes against a particular organic solvent and organic solvents against a particular transition metal complex based on their solubility information without the knowledge of heat of fusion and melting temperature. The energy density of a non-aqueous redox flow cell directly depends on the solubility of the redox active species in the non-aqueous medium. Here, we have used the “COSMOSAC-LANL” activity coefficient model which is based on first-principles COSMO calculations where the microscopic information is passed to the macroscopic world via a dielectric continuum solvation model, followed by a post-statistical thermodynamic treatment of the self-consistent properties of the solute particle to calculate the solubility. To model the activity coefficient at infinite dilution for the binary mixtures, a 3-suffix Margules (3sM) function is introduced for the quantitative estimation of the asymmetric interactions and, for the combinatorial term, the Staverman–Guggenheim (SG) form is used. The new activity coefficient model is separately called the “LANL” activity coefficient model. The metal complex and the organic solvent have been treated as a simple binary mixture. The present model has been applied to a set of 14 different organic solvents and 16 different transition metal complexes. Using the new LANL activity coefficient model in combination with the ADF-COSMOSAC-2013 model, we have shown how one can improve the solubility of a transition metal complex in an organic solvent. We applied our model to screen 84 binary mixtures to predict the compatible pair of redox active species and organic solvent to increase the energy density. The solvation mechanism of the transition metal complexes in the organic solvents was obtained using the new model. The results have been compared with the experimental and theoretical results where they are available.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2021

Photodissociation and Infrared Spectroscopy of Uranium–Nitrogen Cation Complexes

Laser vaporization of uranium in a pulsed supersonic expansion of nitrogen is used to produce complexes of the form U + (N 2 ) n (n = 1–8). These ions are mass selected in a reflectron time-of-flight spectrometer and studied with visible and UV laser fixed-frequency photodissociation and with tunable infrared laser photodissociation spectroscopy. The dissociation patterns and spectroscopy of U + (N 2 ) n indicate that N 2 ligands are intact molecules and that there is no insertion chemistry resulting in UN + or NUN + . Fixed frequency photodissociation at 532 and 355 nm indicate that the U + –N 2 bond dissociation energy varies little with changing coordination. The photon energy and the number of ligands eliminated allow an estimate of the average U + –N 2 dissociation energy of 12 kcal/mol. Infrared bands are observed for these complexes near the N–N stretch vibration via elimination of N 2 molecules. These resonances are observed to be shifted about 130 cm–1 to the red from the free-N 2 frequency for complexes with n = 3–8. Density functional theory indicates that U + is most stable in the sextet state in these complexes and that N 2 molecules bind in end-on configurations. Furthermore, the fully coordinated complex is predicted to be U + (N 2 ) 8 , which has a cubic structure. The vibrational frequencies predicted by theory are consistently lower than those in the experiment, independent of the isomeric structure or spin state of the complexes. Despite its failure to reproduce the infrared spectra, theory provides an average U + –N 2 dissociation energy of 11.8 ± 0.5 kcal/mol, in good agreement with the value from the experiments.

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS↗
DOE OSTI2022

Superconductivity from energy fluctuations in dilute quantum critical polar metals

Abstract Superconductivity in low carrier density metals challenges the conventional electron-phonon theory due to the absence of retardation required to overcome Coulomb repulsion. Here we demonstrate that pairing mediated by energy fluctuations, ubiquitously present close to continuous phase transitions, occurs in dilute quantum critical polar metals and results in a dome-like dependence of the superconducting T c on carrier density, characteristic of non-BCS superconductors. In quantum critical polar metals, the Coulomb repulsion is heavily screened, while the critical transverse optical phonons decouple from the electron charge. In the resulting vacuum, long-range attractive interactions emerge from the energy fluctuations of the critical phonons, resembling the gravitational interactions of a chargeless dark matter universe. Our estimates show that this mechanism may explain the critical temperatures observed in doped SrTiO 3 . We provide predictions for the enhancement of superconductivity near polar quantum criticality in two- and three-dimensional materials that can be used to test our theory.

75 CONDENSED MATTER PHYSICS, SUPERCONDUCTIVITY AND↗
DOE OSTI2021

Laser transport and backscatter in low-density SiO 2 and Ta2O 5 foams

Experiments using a single 527 nm wavelength beam interacting with sub- and supercritical density SiO 2 and Ta 2 O 5 foams examined laser propagation and backscatter from laser–plasma instabilities such as Stimulated Brillouin Scattering (SBS). Two densities of each material were examined, and multiple diagnostics were used to characterize the propagation and backscatter. For 5 mg/cc SiO 2 (n e /n c = 0.375), the laser propagation distance was well approximated by treating the foam as a gas. However, for the 2 mg/cc SiO 2 foam (n e /n c = 0.15), the same model over-predicts the propagation distance by ~40%. Furthermore, existing analytical theories on propagation through subcritical foams were able to account for this difference. The laser heat wave propagated ~1/2 as far in Ta 2 O 5 than SiO 2 foams with similar electron density. We showed that this difference is due to the increased radiation losses in the higher Z foam. The fraction of backscattered light scales linearly with incident laser intensity for the range of intensities examined. Ta 2 O 5 foams had significantly lower levels of backscatter (1–3%) than the SiO 2 (4–8%), which is consistent with estimates of large Landau damping due to the presence of the oxygen atoms. The measured fraction of SBS backscattered laser energy for a 2 mg/cc SiO 2 foam shot was ~4 times lower than predicted by simulations assuming a gas-like foam. We found that we needed to assume increased ion heating such that T i /T e ~1.2–1.5 in the plasma to agree with the measured SBS reflectivity. Analytical models of laser-heated foams predict preferential heating of the ions as has been observed in previous experiments.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2026

Cosmological constraints from a joint DESI DR1 Full-Shape and DR2 BAO

We present a cosmological analysis combining full-shape (FS) clustering measurements from the Dark Energy Spectroscopic Instrument (DESI) DR1 with baryon acoustic oscillation (BAO) measurements from DESI DR2. To achieve a robust combination that accounts for the correlation between the two data releases, we employ the ShapeFit compression method and estimate the joint covariance using EZmocks. This compressed approach inherently mitigates the prior volume effects that have previously dominated Bayesian constraints from DESI data with minimal external priors. Consequently, we obtain — for the first time within a Bayesian framework — reliable DESI-only constraints on extensions to ΛCDM using only a Big Bang Nucleosynthesis prior on the baryon density and a wide prior on the spectral index. In flat ΛCDM, we find Ω m = 0.3035 ± 0.0085, h = 0.6876 ± 0.0059, and σ 8 = 0.822 ± 0.034. For the w 0 w a CDM dynamical dark energy model, we measure w 0 = -0.49 ± 0.25 and w a = -1.52 ± 0.77, improving constraints by ∼ 30% relative to the analogous DR1 measurement and reducing the discrepancy with ΛCDM to 1.4σ when compared to BAO only analyses. We also report competitive limits on the sum of neutrino masses and spatial curvature. This work demonstrates that the ShapeFit compression provides a prior-robust and computationally efficient pathway to constrain beyond-ΛCDM physics with large-scale structure.

baryon acoustic oscillations↗
DOE OSTI2023

MaNGA DynPop – I. Quality-assessed stellar dynamical modelling from integral-field spectroscopy of 10K nearby galaxies: a catalogue of masses, mass-to-light ratios, density profiles, and dark matter

ABSTRACT This is the first paper in our series on the combined analysis of the Dynamics and stellar Population (DynPop) for the MaNGA survey in the final SDSS Data Release 17 (DR17). Here, we present a catalogue of dynamically determined quantities for over 10 000 nearby galaxies based on integral-field stellar kinematics from the MaNGA survey. The dynamical properties are extracted using the axisymmetric Jeans Anisotropic Modelling (JAM) method, which was previously shown to be the most accurate for this kind of study. We assess systematic uncertainties using eight dynamical models with different assumptions. We use two orientations of the velocity ellipsoid: either cylindrically aligned JAMcyl or spherically aligned JAMsph. We also make four assumptions for the models’ dark versus luminous matter distributions: (1) mass-follows-light, (2) free NFW dark halo, (3) cosmologically constrained NFW halo, (4) generalized NFW dark halo, i.e. with free inner slope. In this catalogue, we provide the quantities related to the mass distributions (e.g. the density slopes and enclosed mass within a sphere of a given radius for total mass, stellar mass, and dark matter mass components). We also provide the complete models which can be used to compute the full luminous and mass distribution of each galaxy. Additionally, we visually assess the qualities of the models to help with model selections. We estimate the observed scatter in the measured quantities which decreases as expected with improvements in quality. For the best data quality, we find a remarkable consistency of measured quantities between different models, highlighting the robustness of the results.

Astronomy & Astrophysics↗
DOE OSTI2020

Mars Upper Atmospheric Temperature and Atomic Oxygen Density Derived from the O i 130.4 nm Emission Observed by NASA’s MAVEN Mission

The abundance of atomic oxygen is a major unknown for the Mars upper atmosphere, and systematic measurements of this key species is a primary objective of NASA’s Mars Atmosphere and Volatile Evolution (MAVEN) mission. Moreover, the Martian dayside temperature structure and its variability has been a subject of considerable debate in the past. In this study, an inversion analysis of the periapsis limb scans of the O i 130.4 nm emission observed by MAVEN is performed to quantify the neutral temperature and atomic oxygen abundance in the Mars upper atmosphere. Our analysis yields unprecedented resolution of the spatial variation of Mars atomic oxygen, showing that its exobase density decreases with the solar zenith angle (SZA) from ∼0° to 60° by about an order of magnitude, from ∼10{sup 8} to ∼10{sup 7} cm{sup −3} near perihelion (2015 April) and from ∼10{sup 7} to ∼10{sup 6} cm{sup −3} near aphelion (2017 November). For a given SZA the exobase densities near perihelion are about an order of magnitude higher than those near aphelion. The [O]/[CO{sub 2}] mixing ratio at 130 km altitude can vary from ∼0.1% to 4.6%, which also decreases with increasing SZA. Moreover, it is shown that the Mars exobase temperature does not vary significantly with SZA from ∼0° to ∼60°, which is estimated to be ∼239.2 ± 27.6 K near perihelion and ∼162.4 ± 19.9 K near aphelion, with a difference of 76.8 K between these two seasonal extremes.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Sooting tendencies of terpenes and hydrogenated terpenes as sustainable transportation biofuels

Terpenes are a diverse group of molecules that are synthesized by plants and microorganisms through combining units of isoprene (2-methyl-1,3-butadiene). They typically contain rings and methyl branches, which gives them high energy densities and low freezing points and makes them appealing candidates for sustainable transportation biofuels. Between the original biosynthesis and upgrading options such as hydrogenation, they have a large degree of freedom of structures, e.g., different carbon skeletons, positions of double bonds, and functional groups. Therefore, structure-property data is needed to downselect potential fuel candidates. Here, we measured the sooting tendencies of 17 C10 monoterpenes and 7 of their hydrogenated analogues. The hydrogenated compounds were custom synthesized, so the quantities were too small for conventional smoke point measurements. Thus, the sooting tendencies were quantified with yield sooting index (YSI), which is based on the soot yield in a fuel-doped non-premixed methane flame. Derived smoke points (DSPs) were estimated from a correlation between YSI and smoke point for other hydrocarbons. The YSI of terpenes and their derivatives varies widely from 85.6 to 248.5. The YSI follows the trend: terpenes > dihydroterpenes > tetrahydroterpenes. The DSPs of all the tetrahydroterpenes and some dihydroterpenes are higher than that of a Jet-A fuel sample, suggesting that they offer soot reduction benefits. Further, the YSIs depend strongly on molecular structure; for example, α-pinene and β-pinene have identical carbon skeletons and differ only in the position of one carbon-carbon double bond, but the YSI of α-pinene is 34% higher than that of β-pinene. Detailed decomposition analysis via density functional theory (DFT) suggests that compared with β-pinene, α-pinene requires fewer steps to form the first aromatic ring and the process is more thermodynamically favorable. The YSI difference between the pinenes is mainly affected by the identity of the products from the dominant decomposition pathways.

09 BIOMASS FUELS↗
DOE OSTI2021

CO adsorption on MnO(100): Experimental benchmarks compared to DFT

In this study, CO adsorption on the MnO(100) surface was studied using temperature programmed desorption (TPD) and density functional theory (DFT). TPD results show that CO is weakly-bound on MnO(100), with an experimental adsorption energy of -35.6 ± 2.1 kJ/mol at terrace sites in the low coverage limit. PBE simulations suggest that CO adsorption causes an implausible (2 × 2) surface reconstruction. PBE+U simulations show no signs of surface reconstruction, and provide an accurate estimate of the adsorption energy (-36.4 kJ/mol) when combined with the DFT-D3 method with Becke-Jonson damping to correct for van der Waals interactions. This simulation also shows that CO adsorbs C-down onto the Mn 2+ terrace site in a tilted geometry, which is also observed experimentally and computationally on the similarly-structured NiO(100) transition metal oxide surface. TPD results for large doses show a plateauing of the coverage at about 0.4 monolayers of CO at 85 K, with a defect coverage equivalent to 0.08 monolayers. Adsorption associated with defect sites is indicated by a high-temperature desorption tail which is not satisfactorily explained by DFT simulations of simple step or oxygen vacancy defects.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗
DOE OSTI2023

ac susceptibility studies of intra- and intergrain properties of high-$J_c$ Bi-2212 wires

Bi 2 Sr 2 CaCu 2 O x (Bi-2212) is the only high-$T_c$ superconductor (HTS) available as a multifilamentary round wire with multiple architectures and it is a very promising conductor for the realization of high-field applications. Despite their relatively simple wire fabrication by the powder-in-tube technique, Bi-2212 wires require a tightly controlled overpressure heat treatment (HT) with a multi-parameter time-temperature schedule to achieve high critical current density, $J_c$. Further, the variation of these HT parameters, changes in the wire design, wire diameter and powder quality can lead to variations in both the microstructure and the superconducting performance. Particularly noticeable are variations in $J_c$ performance and degree of filament bridging. In this work, we focus on the use of different magnetic characterization techniques to estimate the bridging level and assess the balance of INTER-grain and intra-grain superconducting properties including the irreversibility field ($H_{\text{irr}}$) and the pinning energy ($U_0$) in differently processed wires. Regardless of the actual bridging level, we find that the supercurrent flows at the filament bundle level, not just at the individual filament level. Moreover, using AC susceptibility we identify two distinct supercurrent contributions, one related to the intra-grain and one to the INTER-grain properties, whose irreversibility fields are different but without large sample-to-sample variation. Moreover, an additional component of intra-grain pinning mechanism becomes effective at low temperatures with positive effects also on the INTER-grain performance. The work clearly shows that detailed magnetic characterizations can become valuable tools to investigate the performance of differently processed Bi-2212 wires, correlating their microstructure and overall transport $J_c$, to obtain a deeper understanding of the causes of performance variation and paths to achieve further improvement.

2212↗
DOE OSTI2020

Performance simulation of the soft gamma-ray concentrator

The soft gamma-ray concentrator is a telescope mission concept utilizing a suitable arrangement of bent multilayer structures of alternating low- and high-density materials. This lens is able to channel gamma-ray photons via total external reflection and concentrate the incident radiation to a point. The channeling technique offers the potential for concentrating gamma rays with focal lengths <10 m and energies >100 keV, beyond the reach of current grazing-incidence hard x-ray mirrors. For the performance estimation of such an instrument, we have developed a flexible set of computer modeling tools to compute the optical properties of multilayer structures, predict the channeling efficiency for a given multilayer configuration, and aid in the optimization of potential gamma-ray concentrator-based telescope designs. This modeling includes the multilayer optical properties calculated by the IMD software, the ray tracing using an IDL code, and the focal plane detector simulation by MEGAlib. We illustrate the potential of this approach by presenting simulated astronomical observations from a balloon-borne platform. The final result, including simulated effective area, instrument sensitivity, and polarization performance, shows that the gamma-ray concentrator will provide greatly increased sensitivity for next-generation soft gamma-ray missions with modest cost and complexity.

47 OTHER INSTRUMENTATION↗
DOE OSTI2021

Evaluation of Oxygen Consumption as a Sensitive Measure of Electrical Cable Polymer Insulation Degradation

Polymers used as dielectric materials in nuclear grade electrical cables, such as cross-linked polyethylene (XLPE) and ethylene-propylene rubber (EPR), are long-lived and durable. Estimating the decades-long useful service life of polymers experimentally in a feasible timeframe of weeks or months requires greatly accelerated aging in the laboratory at elevated temperatures and/or gamma dose rates. Many limitations of accelerated aging, such as due to artefacts of the process including diffusion limited oxidation and dose rate effects, have been identified that can make accurate lifetime prediction at service conditions challenging. This can be due to differences in degradation mechanisms between rapid and extended aging or to the presence of induction periods during which early aging is undetectable using common metrics of polymer insulation characterization such as tensile elongation at break. Conceptually, an ideal measure of aging would track clearly and predictably with material changes throughout the lifecycle of the evolving material state from early aging through end of useful life. With such a metric accelerated aging could be performed at much milder conditions, conditions closer to those found in long term service, and end of life calculated from the early data. Oxygen consumption has previously been proposed as a sensitive measure of insulation polymer oxidation that might meet such desirable characteristics. In this work, nuclear cable insulation samples of the two most common types, XLPE and EPR, from two of the most sourced manufacturers, RSCC and The Okonite Company, were subjected to gamma irradiation at dose rates of 100, 200, and 300 Gy/hr for up to 42 days at 26°C. The aged samples were used to evaluate oxygen consumption as a sensitive measure of aging for these materials and these conditions. For comparison and validation, the aged samples were also characterized using the more conventional tests of mass change, density, carbonyl index, and yellowness index. For both materials, oxygen consumption during irradiation was found to track consistently with applied dose—oxygen consumption rate was linear with time. Mass change and carbonyl index were also clearly proportional to exposure dose. Yellowness index, a surface sensitive measure, was observed to increase linearly with dose at low exposure levels before plateauing. Sample density was not observed to change under the conditions explored. Oxygen consumption was found to be a particularly sensitive metric to measure the degradation of polymeric cable insulation exposed to radiation. Oxygen consumption was compared to Arrhenius and time-temperature-superposition methods for activation energy calculation using literature date and found to be a useful and viable alternative for lifetime prediction. The techniques developed in this work are anticipated to be valuable in the estimation of insulation aging rates under mild conditions approaching service conditions. Direction comparison of useful life prediction by oxygen consumption with prediction via Arrhenius methodology for thermal aging and existing methods for radiation and combined thermal/radiation aging would serve to further validate this promising approach and refine our understanding of material-specific aging rates.

36 MATERIALS SCIENCE↗
DOE OSTI2020

Tuned and screened range-separated hybrid density functional theory for describing electronic and optical properties of defective gallium nitride

We apply a hybrid density functional theory approach, based on a tuned and screened range-separated hybrid (SRSH) exchange-correlation functional, to describe the optoelectronic properties of defective gallium nitride (GaN). SRSH and time-dependent SRSH (TDSRSH) are tuned to produce accurate energetics for the pristine material and applied to the study of a series of point defects in bulk GaN, a blue light-emitting material that degrades in the presence of defects. We first establish the accuracy of the method by comparing the predicted quasiparticle gap and low-energy excitation spectra of (TD)SRSH and many-body perturbation theory for both pristine GaN and GaN containing a single nitrogen vacancy. Aided by the reduced computational cost of (TD)SRSH, we then report on three additional technologically relevant point defects and defect complexes in GaN: the gallium vacancy, the carbon interstitial, and the carbon-silicon complex. We compute the low-energy optical absorption spectra for these defects and show the presence of defect-centered transitions. Furthermore, by estimating the Stokes shift, we predict, in agreement with previous studies, that the carbon substitutional defect is a candidate for the detrimental yellow luminescence in GaN. Furthermore, this study indicates that TDSRSH is a promising and computationally feasible approach for quantitatively accurate, first-principles modeling of defective semiconductors.

36 MATERIALS SCIENCE↗
DOE OSTI2021

Measurements of electron temperature in high-energy-density plasmas using gated x-ray pinhole imaging

We present measurements of spatially and temporally resolved electron temperature in high-energy-density plasmas using gated x-ray pinhole imagers. A 2D image of bremsstrahlung x-ray self-emission from laser-driven plasma plumes is detected at the same time through two pinholes covered with different filter materials. By comparing the attenuated signal through each filter, a spatially resolved electron temperature as low as 0.1 keV can be estimated. Measurements of the plasma plume taken from different directions indicate that imaging through extended plasmas has a negligible effect on the temperature estimates. Methods for estimating the expected signal, selecting filters, and incorporating the response of the detector are discussed.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
DOE OSTI2020

Large scale structure reconstruction with short-wavelength modes

Large scale density modes are difficult to measure because they are sensitive to systematic observational errors in galaxy surveys, but we can study them indirectly by observing their impact on small scale perturbations. Cosmological perturbation theory predicts that second-order density inhomogeneities are a convolution of a short- and a long-wavelength mode. This arises physically because small scale structures grow at different rates depending on the large scale environment in which they reside. This induces an off-diagonal term in the two-point statistics in Fourier space that we use as the basis for a quadratic estimator for the large scale field. Here, we demonstrate that this quadratic estimator works well on an N-body simulation of size (2.5 h -1 Gpc) 3 . In particular, the quadratic estimator successfully reconstructs the long-wavelength modes using only small-scale information. This opens up novel opportunities to study structure on the largest observable scales.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Connecting GRBs from binary neutron star mergers to nuclear properties of neutron stars

The fate of the binary neutron star (NS) merger remnants hinges sensitively upon the NS equation of state and the threshold mass, M ls , that separates a long-lived from a short-lived NS remnant. The nature of the electromagnetic counterparts is also influenced by the remnant type, particularly in determining whether a gamma-ray burst from a compact binary merger (cbGRB) is of short or long duration. We propose a novel approach to probe Mls by linking it to the estimated observed ratio of long to short cbGRBs. We find that current observations broadly favor a relatively high value for this transition, M ls ≃ 1.3M TOV , for which M TOV ≲ 2.6M ⊙ , consistent with numerical simulations, as also shown here. Furthermore, our results disfavor nuclear physics scenarios that would lead to catastrophic pressure loss at a few times nuclear density and temperatures of tens of MeV, leading to a rapid gravitational collapse of binaries with total mass M ≲ 1.3M TOV . Future individual gravitational wave events with on-axis cbGRBs can further bound Mls.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2025

Diabatic error and propagation of Majorana zero modes in interacting quantum dots systems

Motivated by recent experimental progress in realizing Majorana zero modes (MZMs) using quantum dot systems, we investigate the diabatic errors associated with the movement of those MZMs. The movement is achieved by tuning time-dependent gate potentials applied to individual quantum dots, effectively creating a moving potential wall. To probe the optimized movement of MZMs, we calculate the experimentally accessible time-dependent fidelity and local density-of-states using many-body time-dependent numerical methods. Furthermore, our analysis reveals that an optimal potential wall height is crucial to preserve the well-localized nature of the MZM during its movement. Moreover, we analyze diabatic errors in realistic quantum-dot systems, incorporating the effects of repulsive Coulomb interactions and disorder in both hopping and pairing terms. Additionally, we provide a comparative study of diabatic errors arising from the simultaneous versus sequential tuning of multiple gates during the MZMs movement. Finally, we estimate the timescale required for MZM transfer in a six-quantum-dot system, demonstrating that MZM movement is feasible and can be completed well within the qubit's operational lifetime in practical quantum-dot setups.

Density of states↗
DOE OSTI2022

Fast and Accurate Predictions of Total Energy for Solid Solution Alloys with Graph Convolutional Neural Networks

We use graph convolutional neural networks (GCNNs) to produce fast and accurate predictions of the total energy of solid solution binary alloys. GCNNs allow us to abstract the lattice structure of a solid material as a graph, whereby atoms are modeled as nodes and metallic bonds as edges. This representation naturally incorporates information about the structure of the material, thereby eliminating the need for computationally expensive data pre-processing which would be required with standard neural network (NN) approaches. We train GCNNs on ab-initio density functional theory (DFT) for copper-gold (CuAu) and iron-platinum (FePt) data that has been generated by running the LSMS-3 code, which implements a locally self-consistent multiple scattering method, on OLCF supercomputers Titan and Summit. GCNN outperforms the ab-initio DFT simulation by orders of magnitude in terms of computational time to produce the estimate of the total energy for a given atomic configuration of the lattice structure. We compare the predictive performance of GCNN models against a standard NN such as dense feedforward multi-layer perceptron (MLP) by using the root-mean-squared errors to quantify the predictive quality of the deep learning (DL) models. We find that the attainable accuracy of GCNNs is at least an order of magnitude better than that of the MLP.

Lupo Pasini, Massimiliano↗