Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “Class VI”

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 181 records · Page 10

The Far-Ultraviolet Spectra of "Cool" PG1159 Stars

We present a comprehensive study of Far Ultraviolet Spectroscopic Explorer (FUSE) spectra (912-1190 A) of two members of the PG1159 spectral class, which consists of hydrogen-deficient (pre-) white dwarfs with effective temperatures in the range T(sub eff) = 75000-200000 K. As two representatives of the cooler objects, we have selected PG1707+427 (T(sub eff) = 85000 K) and PG1424+535 (T(sub eff) = 110000 K), complementing a previous study of the hotter prototype PG1159-035 (T(sub eff) = 140000 K). The helium-dominated atmospheres are strongly enriched in carbon and oxygen, therefore, their spectra are dominated by lines from C III-IV and O III-VI, many of which were never observed before in hot stars. In addition, lines of many other metals (N, F, Ne, Si, P, S, Ar, Fe) are detectable, demonstrating that observations in this spectral region are most rewarding when compared to the near-ultraviolet and optical wavelength bands. We perform abundance analyses of these species and derive upper limits for several undetected light and heavy metals including iron-group and trans-iron elements. The results are compared to predictions of stellar evolution models for neutron-capture nucleosynthesis and good agreement is found.

Werner, K.↗

Wide Band Gap Chalcogenide Semiconductors

Wide band gap semiconductors are essential for today’s electronic devices and energy applications because of their high optical transparency, controllable carrier concentration, and tunable electrical conductivity. The most intensively investigated wide band gap semiconductors are transparent conductive oxides (TCOs), such as tin-doped indium oxide (ITO) and amorphous In–Ga–Zn–O (IGZO), used in displays and solar cells, carbides (e.g., SiC) and nitrides (e.g., GaN) used in power electronics, and emerging halides (e.g., γ-CuI) and 2D electronic materials (e.g., graphene) used in various optoelectronic devices. Compared to these prominent materials families, chalcogen-based (Ch = S, Se, Te) wide band gap semiconductors are less heavily investigated but stand out because of their propensity for p-type doping, high mobilities, high valence band positions (i.e., low ionization potentials), and broad applications in electronic devices such as CdTe solar cells. This manuscript provides a review of wide band gap chalcogenide semiconductors. First, we outline general materials design parameters of high performing transparent semiconductors, as well as the theoretical and experimental underpinnings of the corresponding research methods. We proceed to summarize progress in wide band gap (EG > 2 eV) chalcogenide materials—namely, II–VI MCh binaries, CuMCh 2 chalcopyrites, Cu 3 MCh 4 sulvanites, mixed-anion layered CuMCh(O,F), and 2D materials—and discuss computational predictions of potential new candidates in this family, highlighting their optical and electrical properties. Then, we finally review applications—for example, photovoltaic and photoelectrochemical solar cells, transistors, and light emitting diodes—that employ wide band gap chalcogenides as either an active or passive layer. Overall, by examining, categorizing, and discussing prospective directions in wide band gap chalcogenides, this Review aims to inspire continued research on this emerging class of transparent semiconductors and thereby enable future innovations for optoelectronic devices.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

Ionization and thermal equilibrium models for O star winds based on time-independent radiation-driven wind theory

Ab initio ionization and thermal equilibrium models are calculated for the winds of O stars using the results of steady state radiation-driven wind theory to determine the input parameters. Self-consistent methods are used for the roles of H, He, and the most abundant heavy elements in both the statistical and the thermal equilibrium. The model grid was chosen to encompass all O spectral subtypes and the full range of luminosity classes. Results of earlier modeling of O star winds by Klein and Castor (1978) are reproduced and used to motivate improvements in the treatment of the hydrogen equilibrium. The wind temperature profile is revealed to be sensitive to gross changes in the heavy element abundances, but insensitive to other factors considered such as the mass-loss rate and velocity law. The reduced wind temperatures obtained in observing the luminosity dependence of the Si IV lambda 1397 wind absorption profile are shown to eliminate any prospect of explaining the observed O VI lambda 1036 line profiles in terms of time-independent radiation-driven wind theory.

Drew, J. E.↗

VELOCITY CHARACTERISTICS OF EVAPORATED PLASMA USING HINODE/EUV IMAGING SPECTROMETER

This paper presents a detailed study of chromospheric evaporation using the EUV Imaging Spectrometer (EIS) onboard Hinode in conjunction with hard X-ray (HXR) observations from Reuven Ramaty High-Energy Solar Spectroscopic Imager (RHESSI). The advanced capabilities of EIS were used to measure Doppler shifts in 15 emission lines covering the temperature range T = 0.05-16 MK during the impulsive phase of a C-class flare on 2007 December 14. Blueshifts indicative of the evaporated material were observed in six emission lines from Fe XIV-XXIV (2-16 MK). Upflow velocity (v{sub up}) was found to scale with temperature as v{sub up} (km s{sup -1}) {approx} 8-18T(MK). Although the hottest emission lines, Fe XXIII and Fe XXIV, exhibited upflows of >200 km s{sup -1}, their line profiles were found to be dominated by a stationary component in contrast to the predictions of the standard flare model. Emission from O VI-Fe XIII lines (0.5-1.5 MK) was found to be redshifted by v{sub down} (km s{sup -1}) {approx} 60-17T (MK) and was interpreted as the downward-moving 'plug' characteristic of explosive evaporation. These downflows occur at temperatures significantly higher than previously expected. Both upflows and downflows were spatially and temporally correlated with HXR emission observed by RHESSI that provided the properties of the electron beam deemed to be the driver of the evaporation. The energy flux of the electron beam was found to be {approx}>5 x 10{sup 10} erg cm{sup -2} s{sup -1}, consistent with the value required to drive explosive chromospheric evaporation from hydrodynamic simulations.

79 ASTRONOMY AND ASTROPHYSICS↗

Velocity Characteristics of Evaporated Plasma using Hinode/EIS

This paper presents a detailed study of chromospheric evaporation using the EUV Imaging Spectrometer (EIS) onboard Hinode in conjunction with HXR observat,ions from RHESSI. The advanced capabilities of EIS were used to measure Doppler shifts in 15 emission lines covering the temperature range T=0.05-16 MK during the impulsive phase of a C-class flare on 2007 December 14. Blueshifts indicative of the evaporated material were observed in six emission lines from Fe XIV-XXIV (2-16 MK). Upflow velocity was found to scale with temperature as v(sub up) (kilometers per second) approximately equal to 5-17 T (MK). Although the hottest emission lines, Fe XXIII and Fe XXIV, exhibited upflows of greater than 200 kilometers per second, their line profiles were found to be dominated by a stationary component in stark contrast to the predictions of the standard flare model. Emission from O VI-Fe XIII lines (0.5-1.5 MK) was found to be redshifted by v(sub down) (kilometers per second) approximately equal to 60-17 T (MK) and was interpreted as the downward-moving 'plug' characteristic of explosive evaporation. These downflows occur at temperatures significantly higher than previously expected. Both upflows and downflows were spatially and temporally correlated with HXR emission observed by RHESSI that provided the properties of the electron beam deemed to be the driver of the evaporation. The energy contained in the electron beam was found to be greater than or equal to 10(sup 11) ergs per square centimeter per second consistent with the value required to drive explosive chromospheric evaporation from hydrodynamic simulations.

Milligan, Ryan O.↗

Analysis of correlations between intrinsic ductility and electronic density of states in refractory alloys

High entropy alloys (HEAs) correspond to a new and emerging class of materials that allows us to explore a large composition space to tune mechanical strength and thermal stability. Therefore, to design better alloys, it is important to scan the high-dimensional space of chemistry, composition and temperature. Here, to facilitate this search, we present a method to screen intrinsically ductile body centered cubic (BCC) refractory alloys from electronic structure calculations by using the density of states (DOS) at the Fermi level, g(μ F ). This correlation between intrinsic ductility and g(μ F ) is tested by analyzing group V (V, Nb, Ta) and VI (Mo, W) refractory metals, binary alloys, such as W-Nb, W-V, Mo-Nb and Mo-V, and refractory alloys for which experimental stress-strain measurements are available. In addition, we perform a high-throughput exploration of the entire composition space of a recently proposed alloy system, CrMoNbV, and identify compositions that exhibit high intrinsic ductility.

36 MATERIALS SCIENCE↗

A structured framework for predicting sustainable aviation fuel properties using liquid-phase FTIR and machine learning

Sustainable aviation fuels have the potential to improve efficiency, reduce emissions, and enhance energy security. To help identify viable sustainable aviation fuels and accelerate research, machine learning models have been developed to predict relevant physicochemical properties. However, many models have limited applicability, leverage data from complex analytical techniques with confined spectral ranges, or use feature decomposition methods that offer limited interpretability. Using liquid-phase Fourier Transform Infrared (FTIR) spectra, this study presents a structured method for creating accurate and interpretable property prediction models for neat molecules, aviation fuels, and blends. Liquid FTIR spectra can be collected quickly and consistently, offering high reliability, sensitivity, and component specificity using less than 2 ml of sample. The method first decomposes FTIR spectra into fundamental building blocks using non-negative matrix factorization (NMF) to enable scientific analysis of FTIR spectra attributes and fuel properties. The NMF features are then used to create five ensemble models for predicting final boiling point, flash point, freezing point, density at 15°C, and kinematic viscosity at -20°C. All models were trained using experimental property data from neat molecules, aviation fuels, and blends. The models accurately predict key properties across a broad range of neat molecules and representative fuels and blends, while enabling interpretation of relationships between compositional elements, such as functional groups or chemical classes, and their resulting properties. This demonstrates strong potential to support sustainable aviation fuel research and development. The models and data are available on an interactive web tool.

Fourier transform infrared spectroscopy↗

Bandgap engineering in semiconductor alloy nanomaterials with widely tunable compositions

Over the past decade, tremendous progress has been achieved in the development of nanoscale semiconductor materials with a wide range of bandgaps by alloying different individual semiconductors. These materials include traditional II–VI and III–V semiconductors and their alloys, inorganic and hybrid perovskites, and the newly emerging 2D materials. One important common feature of these materials is that their nanoscale dimensions result in a large tolerance to lattice mismatches within a monolithic structure of varying composition or between the substrate and target material, which enables us to achieve almost arbitrary control of the variation of the alloy composition. As a result, the bandgaps of these alloys can be widely tuned without the detrimental defects that are often unavoidable in bulk materials, which have a much more limited tolerance to lattice mismatches. This class of nanomaterials could have a far-reaching impact on a wide range of photonic applications, including tunable lasers, solid-state lighting, artificial photosynthesis and new solar cells.

Ning, Cun-Zheng↗

Characterizing the Influence of Organic Polymers on the Specific Reactivity of Particulate Remedial Amendments

Commercially available particulate amendments demonstrate high reactivity for effective treatment of water soluble organic and inorganic contaminants in laboratory studies; however, transport of these particles is constrained in the subsurface. In many field applications, particulate amendments are mixed with organic polymers to enhance mobility for direct push applications or stabilize suspensions for high mass loadings. As such, the interactions between particulate amendments, organic polymers and contaminant species need to be systematically investigated to properly understand mechanistic processes that facilitate predictive performance metrics for specific applications in situ . In this study, batch experiments were conducted to quantify the effects of organic polymers (xanthan gum, guar gum, and sodium alginate), polymer concentration (800 and 4,000 mg/L), and aging (up to 28 days) on chromate treatment rate and capacity by two classes of amendments: reductants [granular zero-valent iron (gZVI), micron-ZVI (mZVI), sulfur modified iron (SMI)], and an adsorbent (bismuth sub-nitrate). When particulate amendments were suspended in polymer solutions, reductants retained between 84–100% of the amendment treatment capacity. Conversely, the adsorbent maintained 63–97% relative treatment capacity of the no-polymer control. Polymer solutions had a more pronounced impact on the rate of chromate removal; first order rates of chemical reduction decreased by as much as 70% and adsorption by up to 81% relative to the no-polymer controls. Polymer–amendment aging experiments also showed decreased Cr(VI) treatment capacity; reductants decreased by as much as 24% and adsorption decreased by as much as 44% after 28 days of incubation. While polymer suspensions are needed to aid the injection of particulate amendments into the subsurface, the results from this study indicate potential losses of treatment capacity and a decrease in the rate of remedial performance due to the physical and chemical interactions between polymer suspensions and reactive particulate amendments. Simple batch systems provide baseline characterization of tripartite interactions for the removal of Cr(VI). Additional work is needed to quantify the full impact of polymers on remedial outcomes under site relevant conditions at field scale.

54 ENVIRONMENTAL SCIENCES↗

Hybrid bandgap engineering for super-hetero-epitaxial semiconductor materials, and products thereof

"Super-hetero-epitaxial" combinations comprise epitaxial growth of one material on a different material with different crystal structure. Compatible crystal structures may be identified using a "Tri-Unity" system. New bandgap engineering diagrams are provided for each class of combination, based on determination of hybrid lattice constants for the constituent materials in accordance with lattice-matching equations. Using known bandgap figures for previously tested materials, new materials with lattice constants that match desired substrates and have the desired bandgap properties may be formulated by reference to the diagrams and lattice matching equations. In one embodiment, this analysis makes it possible to formulate new super-hetero-epitaxial semiconductor systems, such as systems based on group IV alloys on c-plane LaF.sub.3; group IV alloys on c-plane langasite; Group III-V alloys on c-plane langasite; and group II-VI alloys on c-plane sapphire.

Park, Yeonjoon↗

Crystal structure of a multi-domain human smoothened receptor in complex with a super stabilizing ligand

Here, the Smoothened receptor (SMO) belongs to the Class Frizzled of the G protein-coupled receptor (GPCR) superfamily, constituting a key component of the Hedgehog signalling pathway. Here we report the crystal structure of the multi-domain human SMO, bound and stabilized by a designed tool ligand TC114, using an X-ray free-electron laser source at 2.9 Å. The structure reveals a precise arrangement of three distinct domains: a seven-transmembrane helices domain (TMD), a hinge domain (HD) and an intact extracellular cysteine-rich domain (CRD). This architecture enables allosteric interactions between the domains that are important for ligand recognition and receptor activation. By combining the structural data, molecular dynamics simulation, and hydrogen-deuterium-exchange analysis, we demonstrate that transmembrane helix VI, extracellular loop 3 and the HD play a central role in transmitting the signal employing a unique GPCR activation mechanism, distinct from other multi-domain GPCRs.

36 MATERIALS SCIENCE↗

CHROMOSPHERIC EVAPORATION IN AN M1.8 FLARE OBSERVED BY THE EXTREME-ULTRAVIOLET IMAGING SPECTROMETER ON HINODE

We discuss observations of chromospheric evaporation for a complex flare that occurred on 2012 March 9 near 03:30 UT obtained from the Extreme-ultraviolet Imaging Spectrometer (EIS) on board the Hinode spacecraft. This was a multiple event with a strong energy input that reached the M1.8 class when observed by EIS. EIS was in raster mode and fortunately the slit was almost at the exact location of a significant energy input. Also, EIS obtained a full-CCD spectrum of the flare, i.e., the entire CCD was readout so that data were obtained for about the 500 lines identified in the EIS wavelength ranges. Chromospheric evaporation characterized by 150-200 km s{sup -1} upflows was observed in multiple locations in multi-million degree spectral lines of flare ions such as Fe XXII, Fe XXIII, and Fe XXIV, with simultaneous 20-60 km s{sup -1} upflows in million degree coronal lines from ions such as Fe XII-Fe XVI. The behavior of cooler, transition region ions such as O VI, Fe VIII, He II, and Fe X is more complex, but upflows were also observed in Fe VIII and Fe X lines. At a point close to strong energy input in space and time, the flare ions Fe XXII, Fe XXIII, and Fe XXIV reveal an isothermal source with a temperature close to 14 MK and no strong blueshifted components. At this location there is a strong downflow in cooler active region lines from ions such as Fe XIII and Fe XIV, on the order of 200 km s{sup -1}. We speculate that this downflow may be evidence of the downward shock produced by reconnection in the current sheet seen in MHD simulations. A sunquake also occurred near this location. Electron densities were obtained from density sensitive lines ratios from Fe XIII and Fe XIV. Atmospheric Imaging Assembly (AIA) observations from the Solar Dynamics Observatory are used with JHelioviewer to obtain a qualitative overview of the flare. However, AIA data are not presented in this paper. In summary, spectroscopic data from EIS are presented that can be used for predictive tests of models of chromospheric evaporation as envisaged in the Standard Flare Model.

79 ASTRONOMY AND ASTROPHYSICS↗

Land-Use Change Impact on Soil Sustainability in a Climate and Vegetation Transition Zone

A growing world population and climate change are expected to influence future agricultural productivity and land use. This study determined the impact of land-use change on soil sustainability and discussed the factors contributing to these changes. South Dakota was selected as a model system because corn ( Zea mays L.) grain prices tripled between 2006 and 2012 and it is located in a climate and grassland/cropland transition zone. High resolution imagery was used to visually determine land uses (cropland, grassland, nonagricultural, habitat, and water) at 14,400 points in 2006 and 2012. At each point, land-use change and the USDA land capability class (LCC) were determined. Over the 6-yr study period, 6.87% of the grasslands (730,000 ha) were converted to cropland, with 93% occurring on lands generally considered suitable for crop production (LCC ≤ IV) if appropriate practices are followed. Converted grasslands, however, had higher LCC values than existing croplands and lower LCC values than remaining grasslands. In addition, 4.2% of the croplands (250,000 ha) were converted to grasslands, and statewide, 20,000 ha of croplands were converted to grasslands in areas limited by excess water (LCC V). The conversion of grasslands could not be linked to one specific factor and may be related to: (i) a desire to increase financial returns, (ii) changes in the land ownership structure, (iii) technology improvements, (iv) governmental policies, (v) climate change, and (vi) an aging workforce. Here, research and outreach programs that balance the goods and services of different land uses are needed to maintain sustainable agroecosystems.

54 ENVIRONMENTAL SCIENCES↗

The Spitzer survey of interstellar clouds in the gould belt. VI. The Auriga-California molecular cloud observed with IRAC and MIPS

We present observations of the Auriga-California Molecular Cloud (AMC) at 3.6, 4.5, 5.8, 8.0, 24, 70, and 160 μm observed with the IRAC and MIPS detectors as part of the Spitzer Gould Belt Legacy Survey. The total mapped areas are 2.5 deg{sup 2} with IRAC and 10.47 deg{sup 2} with MIPS. This giant molecular cloud is one of two in the nearby Gould Belt of star-forming regions, the other being the Orion A Molecular Cloud (OMC). We compare source counts, colors, and magnitudes in our observed region to a subset of the SWIRE data that was processed through our pipeline. Using color-magnitude and color-color diagrams, we find evidence for a substantial population of 166 young stellar objects (YSOs) in the cloud, many of which were previously unknown. Most of this population is concentrated around the LkHα 101 cluster and the filament extending from it. We present a quantitative description of the degree of clustering and discuss the relative fraction of YSOs in earlier (Class I and F) and later (Class II) classes compared to other clouds. We perform simple SED modeling of the YSOs with disks to compare the mid-IR properties to disks in other clouds and identify 14 classical transition disk candidates. Although the AMC is similar in mass, size, and distance to the OMC, it is forming about 15-20 times fewer stars.

79 ASTRONOMY AND ASTROPHYSICS↗

Nasa Space Flight Instruments: Cost Time Trends

Are NASA’s space flight instruments becoming cheaper or more expensive as time marches forward? After analyzing the costs of hundreds of instruments launched over the last 30 years, the short answer to this question is no… and yes. This paper gives a visual analysis of the cost time trends for various NASA space flight instrument types, such as optical, particles detectors, fields detectors and microwave instruments. In addition to the statistical approaches utilized, such as significance tests, cluster analysis and principle components analysis (PCA), we will also discuss the intangibles which are likely at play, including technological progress, NASA policy and the luck of the draw associated with mission manifests. This analysis was performed as the main driver for the NASA Instrument Cost Model (NICM) recent cost estimating model redesign. Started in 2004, the first version of NICM was based off of instruments launched from 1985-2005, or 20 years’ worth of data. As NICM hit its 10-year anniversary, we wanted to know: should NICM continue to only use the most recent 20 years’ worth of data (1995-2015)? Are instruments becoming cheaper or more expensive as time marches forward? There is evidence in favor of a drop in the median dollar-per-kg value across some instrument types, but little in others. Whereas further research is needed to substantiate, Particles and Optical-Planetary instrument types show moderate to strong evidence of a downward trend in dollar-per-kg. Further research is required to study the nature of this trend (shift, taper, cyclic, etc.). Little evidence for a similar downward trend was detected for Fields or Microwave instruments, or Optical instruments on Earth Orbiting spacecraft. We presented evidence in favor of a drop in the median dollar-per-kg value for Particles and Optical-Planetary instrument types. While similar evidence was weak at best for Fields and Microwave instruments. We can speculate as to the causes for this effect, but we are also equipped to begin to rule out, or at least prioritize, some of the suspected drivers. We observed, for Particles and Optical-Planetary instruments, that perhaps a launch manifest effect was playing part of the role in the observed decrease in dollar-per-kg over the years, noting that the more flagship class missions, which have more money to spend on their instruments, were seen in the earlier years in our data, versus the later years which were dominated by less expensive class missions. However, if this were a dominating driver, would we not have seen the downward trend in the Fields and Microwave instruments as well, which were drawn from that same launch manifest? The fact that we did not observe this helps us rule out the launch manifest effect, and other drivers, such as advances in technology, that seem to be more likely suspects. In that case, however, why would technology advances be helping the Particles and Optical-Planetary instruments only? Why would it not be impacting Optical-Earth Orbiting instruments? Further suspects were looked at as well and ruled out, such as the “Faster, Better, Cheaper”era of NASA development which did not seem to actually impact trends by instrument type on a dollar-perkg scale. VI. Future Work A. Time Series Detailed Statistical Assessment The analysis discussed above sets the foundation for a more rigorous time series analysis of the data. Time series analysis will further explore evidence to-date of time trends for the instrument types which showed the strongest indicators for a decrease in dollar-per-kg: Optical (Planetary) and Particles instruments. More than providing evidence and top-level significance tests, time series analysis would help elucidate what kind of trend that exists in the data, their significance and allow statistically based forecasting (see Figure 10)

Mrozinksi, Joseph↗

Tungsten Biscorroles: New Chiral Sandwich Compounds

Abstract The oxidative metalation method, involving the interaction of free‐base meso ‐triarylcorroles and W(CO) 6 in refluxing decalin, led to a set of three tungsten(VI) biscorroles, the first homoleptic sandwich compounds involving corroles. Single‐crystal X‐ray structures of two of the complexes revealed square‐antiprismatic coordination and strongly domed corroles with long W−N distances of 2.15–2.22 Å and a substantial displacement of ∼1.17 Å of the metal relative to the mean N 4 planes of the ligands. The structures correspond to approximate C 2 symmetry and are thus chiral. DFT calculations strongly indicate that the enantiomers are configurationally stable and hence amenable to chiral resolution. Their other notable properties include a strongly blueshifted Soret band at (357±2) nm, a relatively intense π→W(d ) near‐IR feature at (781±3) nm, and a low electrochemical HOMO–LUMO gap of approximately 1.3 V. The results obtained herein suggest that metallobiscorroles may emerge as a new class of inherently chiral chromophores with novel optical and electrochemical properties.

Alemayehu, Abraham B.↗

A generalized family of anisotropic compact object in general relativity

Highlights: • We present model for anisotropic compact star under general relativity. • Here a 4-dimensional spacetime is embedded into the 5-dimensional flat metric. • The spherically symmetric metric has class 1 under which solution set is obtained. • Interesting and physically viable results are found out. • Investigations on the physical features of the model are performed. - Abstract: We present model for anisotropic compact star under the general theory of relativity of Einstein. In the study a 4-dimensional spacetime has been considered which is embedded into the 5-dimensional flat metric so that the spherically symmetric metric has class 1. A set of solutions for the field equations are found depending on the index n involved in the physical parameters. The interior solutions have been matched smoothly at the boundary of the spherical distribution to the exterior Schwarzschild solution which necessarily provides values of the unknown constants. We have chosen the values of n as n=2 and n=10 to 20000 for which interesting and physically viable results can be found out. The investigations on the physical features of the model include several astrophysical issues, like (i) regularity behaviour of stars at the centre, (ii) well behaved condition for velocity of sound, (iii) energy conditions, (iv) stability of the system via the following three techniques — adiabatic index, Herrera cracking concept and TOV equation, (v) total mass, effective mass and compactification factor and (vi) surface redshift. Specific numerical values of the compact star candidates LMC X-4 and SMC X-1 are calculated for central and surface densities as well as central pressure to compare the model value with actual observational data.

79 ASTRONOMY AND ASTROPHYSICS↗

Multi-fidelity Bayesian neural networks: Algorithms and applications

Here we propose a new class of Bayesian neural networks (BNNs) that can be trained using noisy data of variable fidelity, and we apply them to learn function approximations as well as to solve inverse problems based on partial differential equations (PDEs). These multi-fidelity BNNs consist of three neural networks: The first is a fully connected neural network, which is trained following the maximum a posteriori probability (MAP) method to fit the low-fidelity data; the second is a Bayesian neural network employed to capture the cross-correlation with uncertainty quantification between the low- and high-fidelity data; and the last one is the physics-informed neural network, which encodes the physical laws described by PDEs. For the training of the last two neural networks, we first employ the mean-field variational inference (VI) to maximize the evidence lower bound (ELBO) to obtain informative prior distributions for the hyperparameters in the BNNs, and subsequently we use the Hamiltonian Monte Carlo (HMC) method to estimate accurately the posterior distributions for the corresponding hyperparameters. We demonstrate the accuracy of the present method using synthetic data as well as real measurements. Specifically, we first approximate a one- and four-dimensional function, and then infer the reaction rates in one- and two-dimensional diffusion-reaction systems. Moreover, we infer the sea surface temperature (SST) in the Massachusetts and Cape Cod Bays using satellite images and in-situ measurements. Taken together, our results demonstrate that the present method can capture both linear and nonlinear correlation between the low- and high-fidelity data adaptively, identify unknown parameters in PDEs, and quantify uncertainties in predictions, given a few scattered noisy high-fidelity data. Finally, we demonstrate that we can effectively and efficiently reduce the uncertainties and hence enhance the prediction accuracy with an active learning approach, using as examples a specific one-dimensional function approximation and an inverse PDE problem.

97 MATHEMATICS AND COMPUTING↗