Search NASA⌕ Search

SEARCH · Search NASA

Results for “Large deviations”

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 199 records · Page 11
All sourcesNASA NTRSDOE OSTI
NASA NTRS1985

Derivation of Model Topography

The Fourth-Order model necessitates representation of the topography. The problem of the representation of the topography at grid points is addressed. The attempted was to derive an envelope topography. The TI is obtained by taking local mean plus one standard deviation at each grid point and sigma filtering it. The method was greatly influenced by large standard deviations at steep mountains. The O1 topography is the local mean. The S1 is obtained by Sigma filtering in both latitude and longitude the mean O1. The S2 is when the operation is applied twice and S3 thrice, the Q3 is the sigma filtered local mean of the upper third quantile of the source data.

Balgovind, R. C.↗
DOE OSTI2023

Finite-time correlations boost large voltage angle fluctuations in electric power grids

Abstract Decarbonization in the energy sector has been accompanied by an increased penetration of new renewable energy sources in electric power systems. Such sources differ from traditional productions in that, first, they induce larger, undispatchable fluctuations in power generation and second, they lack inertia. Recent measurements have indeed reported long, non-Gaussian tails in the distribution of local voltage frequency data. Large frequency deviations may induce grid instabilities, leading in worst-case scenarios to cascading failures and large-scale blackouts. In this article, we investigate how correlated noise disturbances, characterized by the cumulants of their distribution, propagate through meshed, high-voltage power grids. For a single source of fluctuations, we show that long noise correlation times boost non-Gaussian voltage angle fluctuations so that they propagate similarly to Gaussian fluctuations over the entire network. However, they vanish faster, over short distances if the noise fluctuates rapidly. We furthermore demonstrate that a Berry–Esseen theorem leads to the vanishing of non-Gaussianities as the number of uncorrelated noise sources increases. Our predictions are corroborated by numerical simulations on realistic models of power grids.

24 POWER TRANSMISSION AND DISTRIBUTION↗
DOE OSTI2015

SU-E-T-98: An Analysis of TG-51 Electron Beam Calibration Correction Factor Uncertainty

Purpose: To analyze the uncertainty of the TG-51 electron beam calibration correction factors for farmer type ion chambers currently used by institutions visited by IROC Houston. Methods: TG-51 calibration data were collected from 181 institutions visited by IROC Houston physicists for 1174 and 197 distinct electron beams from modern Varian and Elekta accelerators, respectively. Data collected and analyzed included ion chamber make and model, nominal energy, N{sub D,w}, I{sub 50}, R{sub 50}, k’R{sub 50}, d{sub ref}, P{sub gr} and pdd(d{sub ref}). k’R{sub 50} data for parallel plate chambers were excluded from the analysis. Results: Unlike photon beams, electron nominal energy is a poor indicator of the actual energy as evidenced by the range of R{sub 50} values for each electron beam energy (6–22MeV). The large range in R{sub 50} values resulted k’R{sub 50} values with a small standard deviation but large range between maximum value used and minimum value (0.001–0.029) used for a specific Varian nominal energy. Varian data showed more variability in k’R{sub 50} values than the Elekta data (0.001–0.014). Using the observed range of R{sub 50} values, the maximum spread in k’R{sub 50} values was determined by IROC Houston and compared to the spread of k’R{sub 50} values used in the community. For Elekta linacs the spreads were equivalent, but for Varian energies of 6 to 16MeV, the community spread was 2 to 6 times larger. Community P{sub gr} values had a much larger range of values for 6 and 9 MeV values than predicted. The range in Varian pdd(d{sub ref} ) used by the community for low energies was large, (1.4–4.9 percent), when it should have been very close to unity. Exradin, PTW Roos and PTW farmer chambers N{sub D,w} values showed the largest spread, ≥11 percent. Conclusion: While the vast majority of electron beam calibration correction factors used are accurate, there is a surprising spread in some of the values used.

07 ISOTOPE AND RADIATION SOURCES↗
NASA NTRS2000

Accuracy of Modelled Stratospheric Temperatures in the Winter Arctic Vortex from Infra Red Montgolfier Long Duration Balloon Measurements

The temperature of the stratosphere has been measured in the Arctic vortex every 9-10 minutes along the trajectory of four Infra Red Montgolfier long duration balloons flown for 7 to 22 days during the winters of 1997 and 1999. From a number of comparisons to independent sensors, the accuracy of the measurements is demonstrated to be plus or minus 0.5 K during nighttime and at altitude below 28 km (10 hPa). The performances of the analyses of global meteorological models, European Center for Medium Range Weather Forecasts (ECMWF) 31 and 50 levels, United Kingdom Meteorological Office (UKMO), Data Assimilation Office (DAO), National Climatic Prediction Center (NCEP) and NCEP/NCAR reanalysis, used in photochemical simulations of ozone destruction and interpretation of satellite data, are evaluated by comparison to this large (3500 data points) and homogeneous experimental data set. Most of models, except ECMWF31 in 1999, do show a smal1 average warm bias of between 0 and 1.6 K, with deviations particularly large, up to 20 K at high altitude (5hPa) in stratospheric warming conditions in 1999. Particularly wrong was ECMWF 31 levels near its top level at 10 hPa in 1999 where temperature 25 K colder than the real atmosphere were reported. The average dispersion between models and measurements varies from plus or minus 1.0 to plus or minus 3.0 K depending on the model and the year. It is shown to be the result of three contributions. The largest is a long wave modulation likely caused by the displacement of the temperature field in the analyses compared to real atmosphere. The second is the overestimation of the vertical gradient of temperature particularly in warming conditions, which explains the increase of dispersion from 1997 to 1999. Unexpectedly, the third and smallest (plus or minus 0.6-0.7 K) is the contribution of meso and subgrid scale vertical and horizontal features associated to the vertical propagation of orographic or gravity waves. Compared to other models, the newly available ECMWF 50 levels version assimilating the high vertical resolution radiances of the space borne Advanced Microwave Sounding Unit, performs significantly better (0.03 plus or minus 1.12 K on average between 10 and 140 hPa in 1999) than other models.

Pommereau, J.-P.↗
DOE OSTI2024

Testing quantum gravity using pulsed optomechanical systems

An interesting idea, dating back to Feynman [Report from Chapel Hill Conference, edited by C. M. DeWitt and D. Rickles (1957)], argues that quantum mechanics may break down for large masses if one entertains the possibility that gravity can be “classical,” thereby leading to predictions different from conventional low-energy quantum gravity. Despite the technical difficulty in testing such deviations, a large number of experimental proposals have been put forward due to the high level of fundamental interest. Here, we consider the Schrödinger-Newton (SN) theory and the correlated worldline (CWL) theory, and show that they can be distinguished from conventional quantum mechanics, as well as each other, by performing pulsed optomechanics experiments. For CWL specifically we develop a framework resembling the commonly used “Heisenberg-picture” treatment of coupled oscillators, allowing one to perform simple calculations for such systems without delving into the deeper path-integral formalism. We find that discriminating between the theories will be very difficult until experimental control over low frequency quantum optomechanical systems is pushed much further. Furthermore, the predicted departures of SN and CWL from quantum mechanics occur at the same scale, so both alternative models could in principle be probed by a single experiment.

71 CLASSICAL AND QUANTUM MECHANICS, GENERAL PHYSIC↗
DOE OSTI2025

Identifying High Ionic Conductivity Compositions of Ionic Liquid Electrolytes Using Features of the Solvation Environment

Binary mixtures of ionic liquids with molecular solvents are gaining interest in electrochemical applications due to the improvement in their performance over neat ionic liquids. Dilution with suitable molecular solvents can reduce the viscosity and facilitate faster diffusion of ions, thereby yielding substantially higher ionic conductivity than that for a pure ionic liquid. Although viscosity and diffusion coefficients typically behave as monotonic functions of concentration, ionic conductivity often passes through a peak value at an optimum molar ratio of the molecular solvent to the ionic liquid. The ionic conductivity maximum is generally explained in terms of a balance between the ease of charge transport and the concentration of the charge carriers. In this work, fluctuation in the local environment surrounding an ion is invoked as a plausible explanation for the ionic conductivity mechanism with a binary mixture of 1-ethyl-3-methylimidazolium tetrafluoroborate and ethylene glycol as an example. The magnitude of the dynamism in the local environment is captured by measuring the spatial and temporal features of the solvation environment. Standard deviation in the number of ions in the solvation environment serves as a spatial feature, while the cage correlation lifetimes for oppositely charged ions within the first solvation shell serve as a temporal feature. Large standard deviations in the cluster ion population and short cage correlation lifetimes are indicators of highly dynamic ionic environment at the molecular level and consequently yield high ionic conductivity. Such compositions were found to be in good agreement with the optimum ionic liquid mole fractions obtained through experimental measurement. Short cage correlation lifetimes enable the identification of optimum mixture compositions using simulation trajectories significantly shorter than those required to implement the Nernst–Einstein or Einstein formalisms for calculating ionic conductivity. We validated the applicability of this approach across force fields and in six ionic liquid-molecular solvent electrolytes formed with combination of cations, anions, and solvents. We offer a computationally efficient approach of screening ionic liquid-molecular solvent binary mixture electrolytes to identify molar ratios that yield high ionic conductivity.

25 ENERGY STORAGE↗
NASA NTRS1986

Galaxy distances and deviations from the Hubble flow - Summary remarks

A summary is presented of the issues addressed by the papers delivered at this conference. Papers addressing classical cosmological topics, including the distances to nearby objects, the intermediate scale distance determination, the local peculiar velocity, the local Hubble constant and other age parameters, and the density parameter from Virgo infall are reviewed. Papers discussing new cosmological developments, including large-scale inhomogeneities, large-scale deviations from Hubble flow, and the origin of the velocity and density inhomogeneities are briefly summarized. Prospects for the next 10 years are examined.

Ostriker, Jeremiah P.↗
DOE OSTI2015

SU-E-T-401: Evaluation of TG-43 Dose Calculation Accuracy for SAVI-Based Accelerated Partial Breast Irradiation (APBI) Via Monte Carlo Simulations

Purpose: The current standard TG-43 dose calculation method for SAVI-based Accelerated Partial Breast Irradiation (APBI) assumes an ideal geometry of infinite homogeneous water. However, in SAVI treatments, the air cavity inside the device and the short source-to-skin distance raise concerns about the dose accuracy of the TG-43 method. This study is to evaluate TG-43 dose calculation accuracy in SAVI treatments using Monte Carlo (MC) simulations. Methods: We recalculated the dose distributions of 15 APBI patients treated with SAVI devices, including five cases with a size of 6–1, five with 8−1 and five with 10−1, using our in-house developed fast MC dose package for HDR brachytherapy (gBMC). A phase-space file was used to model the Ir-192 HDR source. For each case, the patient CT was converted into a voxelized phantom and the dwell positions and times were extracted from treatment plans for MC dose calculations. Clinically relevant dosimetric parameters of the recalculated dose were compared to those computed via the TG-43 approach. Results: A systematic overestimation of doses was found for the 15 cases in TG-43 results, with D90, V150, and V200 for PTV-eval 2.8±1.8%, 2.0±2.2%, and 1.8±3.5% higher than MC results. TG-43 also overestimated the dose to skin with the maximum dose 4.4±8.4% higher on average. The relatively large standard deviation seen in the difference of maximum skin dose is partially ascribed to the statistical uncertainty of MC simulations when computing the maximum dose. It took gBMC ∼1 minute to compute dose for a SAVI plan. Conclusion: The high efficiency of our gBMC package facilitated the studies with a relatively large number of cases. An overestimation of TG-43 doses was found when using this MC package to recompute doses in SAVI cases. Clinical utilization of TG-43 dose calculation method in this scenario should be aware of this fact.

60 APPLIED LIFE SCIENCES↗
NASA NTRS2003

Investigation of the Short-Time Variability of Tropical Tropospheric Ozone

Since 1998, a ground based tropospheric ozone lidar has been running at Reunion Island and has been involved with a daily measurement campaign that was performed in the latter part of the biomass burning season, during November-December 1999. The averaged ozone profile obtained during November-December 1 999 agrees well with averaged ozone profile obtained from ozonesondes launch at Reunion during November-December (1992- 2001). Comparing weekly sonde launches (part of the Southern Hemisphere Additional Ozonesondes: SHADOZ program) with the daily ground-based lidar observations shows that some striking features of the day to day variability profiles are not observed in the sonde measurements. Ozone profiles respond to the nature of disturbances which vary from the one day to the next. The vertical ozone distribution at Reunion is examined as a function of prevailing atmospheric circulation. Backtrajectories show that most of the enhanced ozone crossed over biomass burning and convectively active regions in Madagascar and the southern African continent. The analyses of the meteorological data show that ozone stratification profiles are in agreement with the movement of the synoptical situations in November-December 1999. Three different sequences of transport are explained using wind fields. The first sequence from 23 to 25 November is characterized by Northerly transport, the second sequence from 26 to 30 November, the air masses are influenced by meridional transport. The third sequence from 2 to 6 December is characterized by westerly transport associated with the subtropical jet stream. The large standard deviations of lidar profiles in the middle and upper troposphere are in agreement with the upper wind variabilities which evidence passing ridge and trough disturbances. During the transition period between the dry season and the wet season, multiple ozone sources including stratosphere-troposphere exchanges, convection and biomass burning contribute to tropospheric ozone at Reunion Island through sporadic events characterized by a large spatial and temporal variability.

Randriambelo, Tantely↗
NASA NTRS1991

Payload and General Support Computer (PGSC) Detailed Test Objective (DTO) number 795 postflight report: STS-41

Since 1983, the Space Transportation System (STS) had routinely flown the GRiD 1139 (80286) laptop computer as a portable onboard computing resource. In the spring of 1988, the GRiD 1530, an 80386 based machine, was chosen to replace the GRiD 1139. Human factors ground evaluations and detailed test objectives (DTO) examined the usability of the available display types under different lighting conditions and various angle deviations. All proved unsuitable due to either flight qualification of usability problems. In 1990, an Electroluminescent (EL) display for the GRiD 1530 became flight qualified and another DTO was undertaken to examine this display on-orbit. Under conditions of indirect sunlight and low ambient light, the readability of the text and graphics was only limited by the observer's distance from the display. Although a problem of direct sunlight viewing still existed, there were no problems with large angular deviations nor dark adaptation. No further evaluations were deemed necessary. The GRiD 1530 with the EL display was accepted by the STS program as the new standard for the PGSC.

Adolf, Jurine A.↗
DOE OSTI2022

Spin and charge density waves in quasi-one-dimensional KMn 6 Bi 5

The recent observation that pressure could suppress antiferromagnetic (AFM) order in quasi-one-dimensional AMn 6 Bi 5 Mn-cluster chain materials (A=Na, K, Rb, and Cs) and lead to a superconducting dome offers an alternative Mn-based class of materials with which to study unconventional superconductivity. Using neutron diffraction, we elucidate the exact nature of the previously unknown AFM ground state of KMn 6 Bi 5 and report finding transverse incommensurate spin density waves (SDWs) for the Mn atoms with a propagating direction along the chains. The SDWs have distinct refined amplitudes of ~2.46μ B for the Mn atoms in the pentagons and ~0.29μ B with a large standard deviation for Mn atoms at the center between the pentagons. AFM coupling dominates both the nearest-neighbor Mn-Mn interactions within the pentagon and next-nearest-neighbor Mn-Mn interactions out of the pentagon (along the propagating wave). The SDWs exhibit both local and itinerant characteristics potentially due to cooperative interactions between local magnetic exchange and conduction electrons. Single crystal x-ray diffraction below the AFM transition revealed satellite peaks originating from charge density waves along the chain direction with a q vector twice as large as that of the SDW, pointing to a strong real space coupling between them. Additionally, we report a significant magnetoelastic effect during the AFM transition, especially along the chain direction, observed in temperature-dependent x-ray powder diffraction. Our work not only reveals fascinating intertwined spin, charge, and lattice orders in one-dimensional KMn 6 Bi 5 , but also provides an essential piece of information on its magnetic structure to understand the mechanism of superconductivity in this Mn-based family.

1-dimensional systems↗
DOE OSTI2022

Propagation of non-Gaussian voltage angle fluctuations in high-voltage power grids

Recent measurements have reported non-Gaussian tails in the distribution of frequency data in electric power grids. Large frequency deviations may induce grid instabilities and it is therefore crucial to understand how noise disturbances with long, non-Gaussian tails propagate. Here, we investigate how fluctuations in power feed-in, characterized by non-zero cumulants of their distribution, propagate through high-voltage power grids. Unlike previous investigations which focused on the white-noise limit, we consider the limit of long noise correlation time, where power feed-in fluctuates over times longer than the inherent dynamical time scales of the grid - the relevant regime for large-scale, high-voltage distribution grids. In this work, we show that in this limit, the skewness and kurtosis of the power feed-in distribution propagate similarly as its variance, independently of the distribution of inertia. Non-Gaussianities from individual sources of noise therefore persist throughout the entire network. This finding is corroborated by numerical results on a realistic model of the synchronous grid of continental Europe.

42 ENGINEERING↗
DOE OSTI2016

TH-CD-209-07: Preliminary Experimental Comparison of Spot- and Continuous Line Scanning with Or Without Rescanning for Gated Proton Therapy

Purpose: We aim to compare the performance of discrete spot- or continuous line scanning combined with rescanning in mitigating residual organ motion during gated proton therapy treatments. Methods: The Quasar respiratory phantom was used to move a 2D scintillation detector on a linear trajectory with sinusoidal motion pattern (sin{sup 4}), 20 mm peak-to-peak amplitude and 5 sec period. Its motion was monitored using a customized solution based on optical tracking technology. We compared spot and line scanning plans for a monoenergetic 150 MeV circular field, 50.4 mm radius at isocenter. Transverse dose distributions at 13 cm depth in PMMA (15.47 mm water equivalent) were measured to compare three options for motion mitigation: rescanning (10× factor), gating and their combination. The gating window was centered in the trajectory plateau to simulate end-exhale gated treatment in presence of 2 mm and 4 mm residual motion, parallel or perpendicular to the primary scanning direction. Results: When the target moves perpendicular to the primary scanning direction, large dose deviations are measured (γ3%/3mm=47%) without mitigation techniques. Beam gating combined with rescanning restores target coverage (γ3%/3mm=91%). For parallel target motion, observed dose distortions in the non-compensated irradiation are smaller (γ3%/3mm=77%). Beam gating alone recovers the 100% gamma pass-rate at 3%/3mm. Continuous line scanning reduces delivery time by up to 60% with respect to discrete spot scanning in presence of motion mitigation, and improves homogeneity when rescanning is applied (up to 20%, perpendicular motion, 4 mm residual motion). Conclusion: The direction of motion has a large impact on the target dose coverage. Nevertheless, even in the worst case scenario, gating combined with rescanning could mitigate the impact of motion on dose deposition. Moreover, continuous line rescanning improves the robustness against residual motion in the gating window. This study has received funding from the European Community’s Seventh Framework Programme (FP7/2007–2013) under grant agreement n.290605 (PSI-FELLOW/COFUND) and ‘Giuliana and Giorgio Stefanini Foundation’.

60 APPLIED LIFE SCIENCES↗
DOE OSTI2025

The CAI Database: 26 Al– 26 Mg Isotope Systematics

We present a publicly available calcium–aluminum-rich inclusion (CAI) database that focuses on the initial 26 Al/ 27 Al 0 ratio in CAIs, designed in a way that researchers in cosmochemistry and astrophysics may find useful. To date, the database contains 497 CAIs from 75 peer-reviewed papers. The CAIs are from all chondrite groups and cover different CAI types, textures, and sizes. The database includes the paper; the host meteorite; the CAI name and type; the 26 Al/ 27 Al 0 , δ 26 Mg$^*_0$, and δ 25 Mg values and their uncertainties; the number of regression points; the maximum 27 Al/ 24 Mg; the mean-squared weighted deviation; the CAI size; and CAI descriptions. We grouped the CAIs in different ways to discuss 26 Al/ 27 Al 0 ratio distributions with implications for the CAI formation timeline. Overall, we agree with previous authors that CAIs have a bimodal 26 Al distribution: CAIs with robust isochrons (n = 151) have a median 26 Al/ 27 Al 0 = 4.8 × 10 −5 (with a 1σ standard error of 0.1), while those with isotopic anomalies (n = 87) have a median 26 Al/ 27 Al 0 = 0.3 × 10 −5 (with a 1σ standard error of 0.2). However, the large standard deviation of both groups (1.3 and 2.3, respectively) indicates that the 26 Al/ 27 Al 0 values scatter significantly within each population. CAI types and groups can have distinct 26 Al/ 27 Al 0 and δ 26 Mg$^*_0$, but the unmelted inclusions (n = 33) have the highest median 26 Al/ 27 Al 0 = 5.1 × 10 −5 and a low median δ 26 Mg$^*_0$ = −0.05‰. We find slightly different 26 Al/ 27 Al 0 distributions between CAI chondrite types, but no differences between petrographic types or sizes. These observations can help us to understand CAI formation in the context of astrophysical models.

Astronomy and AstroPhysics↗
NASA NTRS2000

Congenitally blind individuals rapidly adapt to coriolis force perturbations of their reaching movements

Reaching movements made to visual targets in a rotating room are initially deviated in path and endpoint in the direction of transient Coriolis forces generated by the motion of the arm relative to the rotating environment. With additional reaches, movements become progressively straighter and more accurate. Such adaptation can occur even in the absence of visual feedback about movement progression or terminus. Here we examined whether congenitally blind and sighted subjects without visual feedback would demonstrate adaptation to Coriolis forces when they pointed to a haptically specified target location. Subjects were tested pre-, per-, and postrotation at 10 rpm counterclockwise. Reaching to straight ahead targets prerotation, both groups exhibited slightly curved paths. Per-rotation, both groups showed large initial deviations of movement path and curvature but within 12 reaches on average had returned to prerotation curvature levels and endpoints. Postrotation, both groups showed mirror image patterns of curvature and endpoint to the per-rotation pattern. The groups did not differ significantly on any of the performance measures. These results provide compelling evidence that motor adaptation to Coriolis perturbations can be achieved on the basis of proprioceptive, somatosensory, and motor information in the complete absence of visual experience.

NASA Discipline Neuroscience↗
DOE OSTI2015

Cascading dust inflation in Born-Infeld gravity

In the framework of Born-Infeld inspired gravity theories, which deviates from General Relativity (GR) in the high curvature regime, we discuss the viability of Cosmic Inflation without scalar fields. For energy densities higher than the new mass scale of the theory, a gravitating dust component is shown to generically induce an accelerated expansion of the Universe. Within such a simple scenario, inflation gracefully exits when the GR regime is recovered, but the Universe would remain matter dominated. In order to implement a reheating era after inflation, we then consider inflation to be driven by a mixture of unstable dust species decaying into radiation. Because the speed of sound gravitates within the Born-Infeld model under consideration, our scenario ends up being predictive on various open questions of the inflationary paradigm. The total number of e-folds of acceleration is given by the lifetime of the unstable dust components and is related to the duration of reheating. As a result, inflation does not last much longer than the number of e-folds of deceleration allowing a small spatial curvature and large scale deviations to isotropy to be observable today. Energy densities are self-regulated as inflation can only start for a total energy density less than a threshold value, again related to the species' lifetime. Above this threshold, the Universe may bounce thereby avoiding a singularity. Another distinctive feature is that the accelerated expansion is of the superinflationary kind, namely the first Hubble flow function is negative. We show however that the tensor modes are never excited and the tensor-to-scalar ratio is always vanishing, independently of the energy scale of inflation.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2022

Double Paddle‐Wheel Enhanced Sodium Ion Conduction in an Antiperovskite Solid Electrolyte

Abstract Antiperovskite structure compounds (X 3 AB, where X is an alkali cation and A and B are anions) have the potential for highly correlated motion between the cation and a cluster anion on the A or B site. This so‐called “paddle‐wheel” mechanism may be the basis for enhanced cation mobility in solid electrolytes. Through combined experiments and modeling, the first instance of a double paddle‐wheel mechanism, leading to fast sodium ion conduction in the antiperovskite Na 3− x O 1− x (NH 2 ) x (BH 4 ), is shown. As the concentration of amide (NH 2 − ) cluster anions is increased, large positive deviations in ionic conductivity above that predicted from a vacancy diffusion model are observed. Using electrochemical impedance spectroscopy, powder X‐ray diffraction, synchrotron X‐ray diffraction, neutron diffraction, ab initio molecular dynamics simulations, and NMR, the cluster anion rotational dynamics are characterized and it is found that cation mobility is influenced by the rotation of both NH 2 − and BH 4 − species, resulting in sodium ion conductivity a factor of 10 2 higher at x = 1 than expected for the vacancy mechanism alone. Generalization of this phenomenon to other compounds could accelerate fast ion conductor exploration and design.

25 ENERGY STORAGE↗
DOE OSTI2023

Residuals-based distributionally robust optimization with covariate information

We consider data-driven approaches that integrate a machine learning prediction model within distributionally robust optimization (DRO) given limited joint observations of uncertain parameters and covariates. Our framework is flexible in the sense that it can accommodate a variety of regression setups and DRO ambiguity sets. We investigate asymptotic and finite sample properties of solutions obtained using Wasserstein, sample robust optimization, and phi-divergence-based ambiguity sets within our DRO formulations, and explore cross-validation approaches for sizing these ambiguity sets. Through numerical experiments, we validate our theoretical results, study the effectiveness of our approaches for sizing ambiguity sets, and illustrate the benefits of our DRO formulations in the limited data regime even when the prediction model is misspecified.

97 MATHEMATICS AND COMPUTING↗