Engineering Papers⌕ Search

SEARCH · Engineering Papers

Results for “hydrodynamics modeling”

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 1,765 records · Page 98
All sourcesNASA NTRSDOE OSTI
NASA NTRS1994

X-ray clusters in a cold dark matter + lambda universe: A direct, large-scale, high-resolution, hydrodynamic simulation

A new, three-dimensional, shock-capturing, hydrodynamic code is utilized to determine the distribution of hot gas in a cold dark matter (CDM) + lambda model universe. Periodic boundary conditions are assumed: a box with size 85/h Mpc, having cell size 0.31/h Mpc, is followed in a simulation with 270(exp 3) = 10(exp 7.3) cells. We adopt omega = 0.45, lambda = 0.55, h identically equal to H/100 km/s/Mpc = 0.6, and then, from the cosmic background explorer (COBE) and light element nucleosynthesis, sigma(sub 8) = 0.77, omega(sub b) = 0.043. We identify the X-ray emitting clusters in the simulation box, compute the luminosity function at several wavelength bands, the temperature function and estimated sizes, as well as the evolution of these quantities with redshift. This open model succeeds in matching local observations of clusters in contrast to the standard omega = 1, CDM model, which fails. It predicts an order of magnitude decline in the number density of bright (h nu = 2-10 keV) clusters from z = 0 to z = 2 in contrast to a slight increase in the number density for standard omega = 1, CDM model. This COBE-normalized CDM + lambda model produces approximately the same number of X-ray clusters having L(sub x) greater than 10(exp 43) erg/s as observed. The background radiation field at 1 keV due to clusters is approximately the observed background which, after correction for numerical effects, again indicates that the model is consistent with observations.

Cen, Renyue↗
NASA NTRS1994

Hot gas in the cold dark matter scenario: X-ray clusters from a high-resolution numerical simulation

A new, three-dimensional, shock-capturing hydrodynamic code is utilized to determine the distribution of hot gas in a standard cold dark matter (CDM) model of the universe. Periodic boundary conditions are assumed: a box with size 85 h(exp -1) Mpc having cell size 0.31 h(exp -1) Mpc is followed in a simulation with 270(exp 3) = 10(exp 7.3) cells. Adopting standard parameters determined from COBE and light-element nucleosynthesis, sigma(sub 8) = 1.05, omega(sub b) = 0.06, and assuming h = 0.5, we find the X-ray-emitting clusters and compute the luminosity function at several wavelengths, the temperature distribution, and estimated sizes, as well as the evolution of these quantities with redshift. We find that most of the total X-ray emissivity in our box originates in a relatively small number of identifiable clusters which occupy approximately 10(exp -3) of the box volume. This standard CDM model, normalized to COBE, produces approximately 5 times too much emission from clusters having L(sub x) is greater than 10(exp 43) ergs/s, a not-unexpected result. If all other parameters were unchanged, we would expect adequate agreement for sigma(sub 8) = 0.6. This provides a new and independent argument for lower small-scale power than standard CDM at the 8 h(exp -1) Mpc scale. The background radiation field at 1 keV due to clusters in this model is approximately one-third of the observed background, which, after correction for numerical effects, again indicates approximately 5 times too much emission and the appropriateness of sigma(sub 8) = 0.6. If we have used the observed ratio of gas to total mass in clusters, rather than basing the mean density on light-element nucleosynthesis, then the computed luminosity of each cluster would have increased still further, by a factor of approximately 10. The number density of clusters increases to z approximately 1, but the luminosity per typical cluster decreases, with the result that evolution in the number density of bright clusters is moderate in this redshift range, showing a broad peak near z = 0.7, and then a rapid decline above redshift z = 3. Detailed computations of the luminosity functions in the range L(sub x) = 10(exp 40) - 10(exp 44) ergs/s in various energy bands are presented for both cluster central regions and total luminosities to be used in comparison with ROSAT and other observational data sets. The quantitative results found disagree significantly with those found by other investigators using semianalytic techniques. We find little dependence of core radius on cluster luminosity and a dependence of temperature on luminosity given by log kT(sub x) = A + B log L(sub x), which is slightly steeper (B = 0.38) than is indicated by observations. Computed temperatures are somewhat higher than observed, as expected, in that COBE-normalized CDM has too much power on the relevant scales. A modest average temperature gradient is found, with temperatures dropping to 90% of central values at 0.4 h(exp -1) Mpc and 70% of central values at 0.9 h(exp -1) Mpc. Examining the ratio of gas to total mass in the clusters normalized to Omega(sub B) h(exp 2) = 0.015, and comparing with observations, we conclude, in agreement with White (1991), that the cluster observations argue for an open universe.

Kang, Hyesung↗
NASA NTRS2005

Physics-Based Spectra of Accretion Disks around Black Holes

The purpose of this grant was to begin the process of deriving the light output of accretion disks around black holes directly from the actual processes that inject heat into the accreting matter, rather than from guessed dependences of heating rate on physical parameters. At JHU, the effort has focussed so far on models of accretion onto "intermediate mass black holes", a possible class of black holes, examples of which may have recently been discovered in nearby galaxies. There, Krolik and his student (Yawei Hui) have computed stellar atmospheres for uniformly-heated disks around this class of black holes. Their models serve two purposes: they are the very first serious attempts to compute the spectrum from accreting black holes in this mass range; and a library of such models can be used later in this program as contrasts for those computed on the basis of real disk dynamics. The output from these local disk calculations has also been successfully coupled to a program that applies the appropriate relativistic transformations and computes photon trajectories in order to predict the spectrum received by observers located at different polar angles. The principal new result of these calculations is the discovery of potentially observable ionization edges of H-like C and O at frequencies near the peak in flux from these objects. Most of the grant money at UCSB was spent on supporting graduate student Shane Davis. In addition. some money was spent on supporting two other students: Ari Socrates (now a Hubble Fellow at Princeton), and Laura Melling. Davis spent the year constructing stellar atmosphere models of accretion disks appropriate for the high/soft (thermal) state of black hole X-ray binaries. As with AGN models published previously by our collaboration with NASA support. our models include a complete general relativistic treatment of both the disk structure and the propagation of photons from the disk to a distant observer. They also include all important continuum opacity sources, including Compton scattering and bound-free opacity from abundant metal species. The principal new result is that bound-free opacity is very significant in altering the continuum spectral shape, resulting for example in quite different "color correction factors" compared to those predicted previously. In addition, the models predict a relationship between luminosity and inner disk temperature that is, for the first time, in accord with that observed. The primary purpose of the grant was to incorporate more realistic accretion disk physics, learned largely from simulations, into such spectral models. The Davis et al. paper includes consideration of a vertical dissipation profile computed from radiation magneto-hydrodynamic simulations of MRI turbulence by N. J. Turner (2004). So long as the disk is effectively thick, such dissipation profiles do not affect the predicted spectrum significantly. (More work needs to be done on these simulations, however.) A potentially more serious issue is that MRI turbulence produces substantial inhomogeneities, as do photon bubble instabilities. These inhomogeneities can affect the spectra by enhancing the effects of absorption opacity over scattering opacity. We have done some preliminary Monte Carlo calculations to explore these effects.

Krolik, Julian H.↗
DOE OSTI2019

Role of mineral flotation technology in improving bitumen extraction from mined Athabasca oil sands—II. Flotation hydrodynamics of water‐based oil sand extraction

Abstract Bitumen flotation hydrodynamics in water‐based oil sand extraction is critically reviewed by comparing aeration of oil sand slurries with mineral flotation. The role of the two‐stage particle‐bubble attachment model in flotation is emphasized as a means to accelerate bitumen flotation recovery. It involves the generation of micro/nanobubbles and their frosting on hydrophobic bitumen droplets, followed by their attachment to a flotation‐size bubble via its coalescence with the nanobubbles frosted on the bitumen. Nanobubble generation by hydrodynamic cavitation demonstrates that the size of nanobubbles can be reduced, and the number of nanobubbles increased by fast liquid flow, intensified agitation, high dissolved gas content and surfactant concentration. The mechanism of pre‐existing gas nuclei in enhancing nanobubble generation by cavitation is utilized to produce a large volume of stabilized nanobubbles for practical flotation, by continuously recirculating the stream through a gas saturation tank or a cavitation tube. The aeration of oil sand slurries in hydrotransport pipelines is analyzed based on its similarity to dissolved air flotation. Bitumen extraction recovery increased significantly with the presence of nanobubbles in the system. The role of improved flotation hydrodynamics in bitumen recovery is briefly discussed in terms of the Suncor operation using flotation columns to process oil sand middling streams. Future research should be directed at understanding bitumen flotation kinetics, optimizing size ranges of nanobubbles for maximized flotation recovery, minimizing wearing of cavitation tubes in industrial operations, and intensifying the role of in‐situ nanobubble nucleation on hydrophobic particles/bitumen droplets in flotation, especially for bitumen extraction from underperforming oil sands.

Zhou, Joe Z.↗
DOE OSTI2024

Dynamics of the O ( 4 ) critical point in QCD: Critical pions and diffusion in model G

We present a detailed study of the finite momentum dynamics of the O ( 4 ) critical point of QCD, which lies in the dynamic universality class of “model G.” The critical scaling of the model is analyzed in multiple dynamical channels. For instance, the finite momentum analysis allows us to precisely extract the pion dispersion curve below the critical point. The pion velocity is in striking agreement with the predictions relation and static universality. The pion damping rate and velocity are both consistent with the dynamical critical exponent ζ = 3 / 2 of model G. Similarly, although the critical amplitude for the diffusion coefficient of the conserved O ( 4 ) charges is small, it is clearly visible both in the restored phase and with finite explicit symmetry breaking, and its dynamical scaling is again consistent with ζ = 3 / 2 . We determine a new set of universal dynamical critical amplitude ratios relating the diffusion coefficient to a suitably defined order parameter relaxation time. We also show that in a finite volume simulation, the chiral condensate diffuses on the coset manifold in a manner consistent with dynamical scaling, and with a diffusion coefficient that is determined by the transport coefficients of hydrodynamic pions. Finally, the amplitude ratios (together with other nonuniversal amplitudes also reported here) compile all relevant information for further studies of model G both in and out of equilibrium. Published by the American Physical Society 2024

73 NUCLEAR PHYSICS AND RADIATION PHYSICS↗
NASA NTRS1984

Theory and modeling of atmospheric turbulence, part 2

Two dimensional geostrophic turbulence driven by a random force is investigated. Based on the Liouville equation, which simulates the primitive hydrodynamical equations, a group-kinetic theory of turbulence is developed and the kinetic equation of the scaled singlet distribution is derived. The kinetic equation is transformed into an equation of spectral balance in the equilibrium and non-equilibrium states. Comparison is made between the propagators and the Green's functions in the case of the non-asymptotic quasi-linear equation to prove the equivalence of both kinds of approximations used to describe perturbed trajectories of plasma turbulence. The microdynamical state of fluid turbulence is described by a hydrodynamical system and transformed into a master equation analogous to the Vlasov equation for plasma turbulence. The spectral balance for the velocity fluctuations of individual components shows that the scaled pressure strain correlation and the cascade transfer are two transport functions that play the most important roles.

Chen, C. M.↗
NASA NTRS1989

On the ability of rough surface scattering approximations to predict hydrodynamic modulation of the ocean radar cross section - A numerical study

Scattering from the ocean surface at microwave frequencies is often described by the Bragg scattering model or the two-scale model. Here, the ability of these models to predict cross section modulation is investigated by comparing Bragg and two-scale cross-sections with cross sections calculated by a numerical electromagnetic approach. The numerical method is used as the standard of comparison. For slightly rough surfaces, the Bragg model is found to be very accurate, but for much rougher surfaces, the two-scale model is superior and should be used for calculating both the absolute cross section and the change in cross section due to spatial or temporal modulation of the ocean wave height spectrum. These results are relevant to the interpretation of SAR images of ocean surface phenomena which perturb the wave height spectrum at ripple wavelengths.

Durden, Stephen L.↗
NASA NTRS2016

Migration of Accreting Planets in Radiative Discs From Dynamical Torques

We present the results of hydrodynamical simulations of the orbital evolution of planets undergoing runaway gas accretion in radiative discs. We consider accreting disc models with constant mass flux through the disc, and where radiative cooling balances the effect of viscous heating and stellar irradiation. We assume that 20–30 M(sub ⨁) giant planet cores are formed in the region where viscous heating dominates and migrate outward under the action of a strong entropy-related corotation torque. In the case where gas accretion is neglected and for an α viscous stress parameter α = 2 × 10(exp -3), we find evidence for strong dynamical torques in accreting discs with accretion rates M ̇ ≳ 7 × 10(exsp -8)M(sub ⨀) yr(exp -1). Their main effect is to increase outward migration rates by a factor of ∼2 typically. In the presence of gas accretion, however, runaway outward migration is observed with the planet passing through the zero-torque radius and the transition between the viscous heating and stellar heating dominated regimes. The ability for an accreting planet to enter a fast migration regime is found to depend strongly on the planet growth rate but can occur for values of the mass flux through the disc of M ̇ ≳ 5 × 10(exp -8)M(sub ⨀) yr(exp -1).We find that an episode of runaway outward migration can cause an accreting planet formed in the 5–10 au region to temporarily orbit at star–planet separations as large as∼60–70 au. However, increase in the amplitude of the Lindblad torque associated with planet growth plus change in the streamline topology near the planet systematically cause the direction of migration to be reversed. Subsequent evolution corresponds to the planet migrating inward rapidly until it becomes massive enough to open a gap in the disc and migrate in the type II regime. Our results indicate that a planet can reach large orbital distances under the combined effect of dynamical torques and gas accretion, but an alternative mechanism is required to explain the presence of massive planets on wide orbits.

Accretion↗
NASA NTRS1995

Features and applications of the Groove Analysis Program (GAP)

An IBM Personal Computer (PC) version of the Groove Analysis program (GAP) was developed to predict the steady state heat transport capability of an axially grooved heat pipe for a specified groove geometry and working fluid. In the model, the capillary limit is determined by the numerical solution of the differential equation for momentum conservation with the appropriate boundary conditions. This governing equation accounts for the hydrodynamic losses due to friction in liquid and vapor flows and due to liquid/vapor shear interaction. Back-pumping in both 0-g and 1-g is accounted for in the boundary condition at the condenser end. Slug formation in 0-g and puddle flow in 1-g are also considered in the model. At the user's discretion, the code will perform the analysis for various fluid inventories (undercharge, nominal charge, overcharge, or a fixed fluid charge) and heat pipe elevations. GAP will also calculate the minimum required heat pipe wall thickness for pressure containment at design temperatures that are greater than or lower than the critical temperature of the working fluid. This paper discusses the theory behind the development of the GAP model. It also presents the many useful and powerful capabilities of the model. Furthermore, a correlation of flight test performance data and the predictions using GAP are presented and discussed.

Ku, Jentung↗
NASA NTRS1992

The formation of coronal regions in accretion disks

A mechanism for heating the hot low-density gas or winds above accretion disks, which is similar to that used to model chromospheric heating in the sun and other stars, is proposed. Sound waves propagating through an accretion disk are refracted away from the central plane by the strong density gradient. As they move into regions of lower density, the sound waves accelerate to form shocks, which heat the gas, leading to the formation of a hot low-density region. The steeper density gradients present in disks make the process more efficient than in stellar atmospheres. Results of hydrodynamical simulations show that waves with frequencies similar to the local Keplerian frequency lead to the most efficient heating. For the optically thin region modeled in the present study, the result is the formation of coronal regions with densities less than about 10 exp 10/cu cm, and the temperatures ranging from 10 exp 4 to 6 K over a few scale heights. It is argued that the hot low-density gas which results from the shock heating is responsible for the observed UV lines from cataclysmic variables, as well as the spatially coincidental lines of H and He I.

Murray, Stephen D.↗
NASA NTRS1983

Initialization of mesoscale models - The possible impact of remotely sensed data

Little or no improvement appears to have been achieved in mesoscale numerical prediction. This has been blamed on inherent unpredictability, insufficient spatial resolution, inadequate parameterization of important physical processes, and errors in the numerical discretization of the governing hydrodynamic equations. Attention is presently given to the more fundamental difficulty of forecasting when present conditions are not known with sufficient accuracy, and to the matching of remotely sensed sounding-capability satellite and all-weather Doppler radar data with mesoscale numerical models. This matching is not straightforward; simple and variational forms of four-dimensional assimilation, and Kalman filtering, are suggested approaches but their practical testing remains to be accomplished. The most severe problem faced is that of the temperature and humidity profiles' vertical resolution.

Gal-Chen, T.↗
DOE OSTI2025

Droplet breakup by multimodal nonlinear Rayleigh Taylor instability

A droplet impacted by a shock wave will undergo a process of fragmentation due to the development of interfacial hydrodynamic instabilities. The interface experiences variable acceleration and shear that result in the development of both inertial (Rayleigh–Taylor) and shear (Kelvin–Helmholtz) instabilities. These perturbations grow in time and drive the fragmentation and breakup of the deformed droplet. Experiments are performed on nominally 0.86 mm water droplet subjected to a Mach 7.6 detonation wave, resulting in a high Weber number (~36,000) breakup event. Perturbation growth is measured from a series of high-speed (> MHz) shadowgraph images. It is proposed that, given the size of the large-scale perturbations observed in experiments, these instabilities are growing in the non-linear regime and can be described by bubble-merger models for nonlinear mixing. Calculations are performed for the growth rates and size of these instabilities using deformation and external flow models to establish the time-dependent boundary conditions. The concurrence of the measured perturbation widths and the predictions of the simple model lend credence to the theory. This novel approach serves to open a new avenue in the characterization of droplet breakup via hydrodynamic instabilities.

Droplet breakup↗
DOE OSTI2024

Data-driven prediction of scaling and ignition of inertial confinement fusion experiments

Recent advances in inertial confinement fusion (ICF) at the National Ignition Facility (NIF), including ignition and energy gain, are enabled by a close coupling between experiments and high-fidelity simulations. Neither simulations nor experiments can fully constrain the behavior of ICF implosions on their own, meaning pre- and postshot simulation studies must incorporate experimental data to be reliable. Linking past data with simulations to make predictions for upcoming designs and quantifying the uncertainty in those predictions has been an ongoing challenge in ICF research. We have developed a data-driven approach to prediction and uncertainty quantification that combines large ensembles of simulations with Bayesian inference and deep learning. The approach builds a predictive model for the statistical distribution of key performance parameters, which is jointly informed by past experiments and physics simulations. The prediction distribution captures the impact of experimental uncertainty, expert priors, design changes, and shot-to-shot variations. We have used this new capability to predict a 10× increase in ignition probability between Hybrid-E shots driven with 2.05 MJ compared to 1.9 MJ, and validated our predictions against subsequent experiments. We describe our new Bayesian postshot and prediction capabilities, discuss their application to NIF ignition and validate the results, and finally investigate the impact of data sparsity on our prediction results.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗
NASA NTRS2005

Hydrodynamics of Packed Bed Reactor in Low Gravity

Packed bed reactors are well known for their vast and diverse applications in the chemical industry; from gas absorption, to stripping, to catalytic conversion. Use of this type of reactor in terrestrial applications has been rather extensive because of its simplicity and relative ease of operation. Developing similar reactors for use in microgravity is critical to many space-based advanced life support systems. However, the hydrodynamics of two-phase flow packed bed reactors in this new environment and the effects of one physiochemical process on another has not been adequately assessed. Surface tension or capillary forces play a much greater role which results in a shifting in flow regime transitions and pressure drop. Results from low gravity experiments related to flow regimes and two-phase pressure drop models are presented in this paper along with a description of plans for a flight experiment on the International Space Station (ISS). Understanding the packed bed hydrodynamics and its effects on mass transfer processes in microgravity is crucial for the design of packed bed chemical or biological reactors to be used for water reclamation and other life support processes involving water purification.

Motil, Brian J.↗
DOE OSTI2023

Compressible pairwise interaction extended point-particle model for force prediction of shock-particle bed interaction

We propose a pairwise influence framework for the complex unsteady compressible particle-laden flow problem by accounting for the scattered hydrodynamic waves emitting from neighboring particles in a Euler-Lagrange simulation. It has been observed from particle-resolved (PR) simulations of randomly dispersed particle beds under a loading shock that the compressible pseudoturbulence dominates the flow system even after the primary shock has passed, which causes fluctuations observed in the forces experienced by the particle. Moreover, the fact that each particle exists in the vicinity of a random arrangement of other particles modifies the time history of the drag force experienced by each particle during and after the passage of the shock. First, the scattering flow field due to an incoming shock interacting with a single sphere is constructed using an analysis of the flow in the acoustic limit. Then we examine the validity of the compressible Maxey-Riley-Gatignol force model by comparing the force prediction against a PR simulation of two interacting particles for various particle arrangements and incoming shock strength. Subsequently, the neighboring influences are stored as a library of maps that can be used readily in the calculation of the perturbation force. Lastly, the pairwise interaction assumption is evaluated by comparing the force predicted with the model with PR simulations of a randomly packed particle bed of 10% volume fraction for both water and air as the fluid medium for an incoming shock Mach number 1.22. With a considerably lower cost for the implementation of the model compared to PR simulations, it is verified that the model is reasonably accurate in pinpointing particles whose peak force is significantly larger or smaller than the mean drag but also to capture the prolonged fluctuations after the initial shock.

72 PHYSICS OF ELEMENTARY PARTICLES AND FIELDS↗
DOE OSTI2023

Domain Adaptive Graph Neural Networks for Constraining Cosmological Parameters Across Multiple Data Sets

Deep learning models have been shown to outperform methods that rely on summary statistics, like the power spectrum, in extracting information from complex cosmological data sets. However, due to differences in the subgrid physics implementation and numerical approximations across different simulation suites, models trained on data from one cosmological simulation show a drop in performance when tested on another. Similarly, models trained on any of the simulations would also likely experience a drop in performance when applied to observational data. Training on data from two different suites of the CAMELS hydrodynamic cosmological simulations, we examine the generalization capabilities of Domain Adaptive Graph Neural Networks (DA-GNNs). By utilizing GNNs, we capitalize on their capacity to capture structured scale-free cosmological information from galaxy distributions. Moreover, by including unsupervised domain adaptation via Maximum Mean Discrepancy (MMD), we enable our models to extract domain-invariant features. We demonstrate that DA-GNN achieves higher accuracy and robustness on cross-dataset tasks. Using data visualizations, we show the effects of domain adaptation on proper latent space data alignment. This shows that DA-GNNs are a promising method for extracting domain-independent cosmological information, a vital step toward robust deep learning for real cosmic survey data.

79 ASTRONOMY AND ASTROPHYSICS↗
DOE OSTI2023

Bridging the Gap between Cosmological Simulations with Graph Neural Networks and Domain Adaptation

Deep learning models have been shown to outperform methods that rely on summary statistics, like the power spectrum, in extracting information from complex cosmological data sets. However, due to differences in the subgrid physics implementation and numerical approximations across different simulation suites, models trained on data from one cosmological simulation show a drop in performance when tested on another. Similarly, models trained on any of the simulations would also likely experience a drop in performance when applied to observational data. Training on data from two different suites of the CAMELS hydrodynamic cosmological simulations, we examine the generalization capabilities of Domain Adaptive Graph Neural Networks (DA-GNNs). By utilizing GNNs, we capitalize on their capacity to capture structured scale-free cosmological information from galaxy distributions. Moreover, by including unsupervised domain adaptation via Maximum Mean Discrepancy (MMD), we enable our models to extract domain-invariant features. We demonstrate that DA-GNN achieves higher accuracy and robustness on cross dataset tasks (up to 28% better relative error and up to almost an order of magnitude better χ 2 ). Using data visualizations, we show the effects of domain adaptation on proper latent space data alignment. This shows that DA-GNNs are a promising method for extracting domain-independent cosmological information, a vital step toward robust deep learning for real cosmic survey data.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2023

Domain Adaptive Graph Neural Networks for Constraining Cosmological Parameters Across Multiple Data Sets

Deep learning models have been shown to outperform methods that rely on summary statistics, like the power spectrum, in extracting information from complex cosmological data sets. However, due to differences in the subgrid physics implementation and numerical approximations across different simulation suites, models trained on data from one cosmological simulation show a drop in performance when tested on another. Similarly, models trained on any of the simulations would also likely experience a drop in performance when applied to observational data. Training on data from two different suites of the CAMELS hydrodynamic cosmological simulations, we examine the generalization capabilities of Domain Adaptive Graph Neural Networks (DA-GNNs). By utilizing GNNs, we capitalize on their capacity to capture structured scale-free cosmological information from galaxy distributions. Moreover, by including unsupervised domain adaptation via Maximum Mean Discrepancy (MMD), we enable our models to extract domain-invariant features. We demonstrate that DA-GNN achieves higher accuracy and robustness on cross-dataset tasks (up to $28\%$ better relative error and up to almost an order of magnitude better $\chi^2$). Using data visualizations, we show the effects of domain adaptation on proper latent space data alignment. This shows that DA-GNNs are a promising method for extracting domain-independent cosmological information, a vital step toward robust deep learning for real cosmic survey data.

79 ASTRONOMY AND ASTROPHYSICS↗