Searcharxiv⌕ Search

arXiv subjects

Nico Cappelluti

Publications and source records attributed to Nico Cappelluti.

At least 37 records · Page 2Linked to original sources
arXiv2024

Morphological Parameters and Associated Uncertainties for 8 Million Galaxies in the Hyper Suprime-Cam Wide Survey

We use the Galaxy Morphology Posterior Estimation Network (GaMPEN) to estimate morphological parameters and associated uncertainties for $\sim 8$ million galaxies in the Hyper Suprime-Cam (HSC) Wide survey with $z \leq 0.75$ and $m \leq 23$. GaMPEN is a machine learning framework that estimates Bayesian posteriors for a galaxy's bulge-to-total light ratio ($L_B/L_T$), effective radius ($R_e$), and flux ($F$). By first training on simulations of galaxies and then applying transfer learning using real data, we trained GaMPEN with $<1\%$ of our dataset. This two-step process will be critical for applying machine learning algorithms to future large imaging surveys, such as the Rubin-Legacy Survey of Space and Time (LSST), the Nancy Grace Roman Space Telescope (NGRST), and Euclid. By comparing our results to those obtained using light-profile fitting, we demonstrate that GaMPEN's predicted posterior distributions are well-calibrated ($\lesssim 5\%$ deviation) and accurate. This represents a significant improvement over light profile fitting algorithms which underestimate uncertainties by as much as $\sim60\%$. For an overlapping sub-sample, we also compare the derived morphological parameters with values in two external catalogs and find that the results agree within the limits of uncertainties predicted by GaMPEN. This step also permits us to define an empirical relationship between the Sérsic index and $L_B/L_T$ that can be used to convert between these two parameters. The catalog presented here represents a significant improvement in size ($\sim10 \times $), depth ($\sim4$ magnitudes), and uncertainty quantification over previous state-of-the-art bulge+disk decomposition catalogs. With this work, we also release GaMPEN's source code and trained models, which can be adapted to other datasets.

astro-ph.GA↗
arXiv2023

First Detection of an Over-Massive Black Hole Galaxy UHZ1: Evidence for Heavy Black Hole Seed Formation from Direct Collapse

The recent Chandra-JWST discovery of a quasar in the z = 10.1 galaxy UHZ1 reveals that accreting supermassive black holes (SMBHs) were already in place 470 million years after the Big Bang. The Chandra X-ray source detected in UHZ1 is a Compton-thick quasar with a bolometric luminosity of $L_{\rm bol}\sim5\times10^{45}\ \rm{erg\ s^{-1}},$ which corresponds to an estimated BH mass of $\sim4\times10^7 \ \rm{M_{\odot}}$ assuming accretion at the Eddington rate. JWST NIRCAM and NIRSpec data yield a stellar mass estimate for UHZ1 comparable to its BH mass. These characteristics are in excellent agreement with prior theoretical predictions for a unique class of transient, high-redshift objects, Over-massive Black Hole Galaxies [OBGs] by Natarajan et al. that harbor a heavy initial black hole seed that likely formed from the direct collapse of the gas. Based on the excellent agreement between the observed multi-wavelength properties of UHZ1 with theoretical model template predictions, suggests that UHZ1 is the first detected OBG candidate. Our assertion rests on multiple lines of concordant evidence between model predictions and the following observed properties of UHZ1: its X-ray detection and the estimated ratio of the X-ray flux to the IR flux that is consistent with theoretical expectations for a heavy initial BH seed; its high measured redshift of z = 10.1, as predicted for the transient OBG stage (9 < z< 12); the amplitude and shape of the detected JWST Spectral Energy Distribution (SED) between 1 - 5 microns, which is in very good agreement with simulated template SEDs for OBGs; and the extended JWST morphology of UHZ1 that is suggestive of a recent merger, also expected for the formation of transient OBGs. As the first OBG candidate, UHZ1 provides compelling evidence for the formation of heavy initial seeds from direct collapse in the early Universe.

astro-ph.HE↗
arXiv2023

Detecting Wandering Intermediate-Mass Black Holes with AXIS in the Milky Way and Local Massive Galaxies

This white paper explores the detectability of intermediate-mass black holes (IMBHs) wandering in the Milky Way (MW) and massive local galaxies, with a particular emphasis on the role of AXIS. IMBHs, ranging within $10^{3-6} \, M_\odot$, are commonly found at the centers of dwarf galaxies and may exist, yet undiscovered, in the MW. By using model spectra for advection-dominated accretion flows (ADAFs), we calculated the expected fluxes emitted by a population of wandering IMBHs with a mass of $10^5 \, M_\odot$ in various MW environments and extrapolated our results to massive local galaxies. Around $40\%$ of the potential population of wandering IMBHs in the MW can be detected in an AXIS deep field. We proposed criteria to aid in selecting IMBH candidates using already available optical surveys. We also showed that IMBHs wandering in $>200$ galaxies within $10$ Mpc can be easily detected with AXIS when passing within dense galactic environments (e.g., molecular clouds and cold neutral medium). In summary, we highlighted the potential X-ray detectability of wandering IMBHs in local galaxies and provided insights for guiding future surveys. Detecting wandering IMBHs is crucial for understanding their demographics, evolution, and the merging history of galaxies.

astro-ph.GA↗
arXiv2023

Uncovering the First AGN Jets with AXIS

Jets powered by AGN in the early Universe ($z \gtrsim 6$) have the potential to not only define the trajectories of the first-forming massive galaxies but to enable the accelerated growth of their associated SMBHs. Under typical assumptions, jets could even rectify observed quasars with light seed formation scenarios; however, not only are constraints on the parameters of the first jets lacking, observations of these objects are scarce. Owing to the significant energy density of the CMB at these epochs capable of quenching radio emission, observations will require powerful, high angular resolution X-ray imaging to map and characterize these jets. As such, \textit{AXIS} will be necessary to understand early SMBH growth and feedback.

astro-ph.GA↗
arXiv2023

AGN in overdense environments at high-$z$ with AXIS

Overdense regions at high redshift ($z \gtrsim 2$) are perfect laboratories to study the relations between environment and SMBH growth, and the AGN feedback processes on the surrounding galaxies and diffuse gas. In this white paper, we discuss how AXIS will 1) constrain the AGN incidence in protoclusters, as a function of parameters such as redshift, overdensity, mass of the structure; 2) search for low-luminosity and obscured AGN in the satellite galaxies of luminous QSOs at $z>6$, exploiting the large galaxy density around such biased objects; 3) probe the AGN feedback on the proto-ICM via the measurement of the AGN contribution to the gas ionization and excitation, and the detection of extended X-ray emission from the ionized gas and from radio jets; 4) discover new large-scale structures in the wide and deep AXIS surveys as spikes in the redshift distribution of X-ray sources. These goals can be achieved only with an X-ray mission with the capabilities of AXIS, ensuring a strong synergy with current and future state-of-the-art facilities in other wavelengths. This White Paper is part of a series commissioned for the AXIS Probe Concept Mission; additional AXIS White Papers can be found at http://axis.astro.umd.edu/ with a mission overview at https://arxiv.org/abs/2311.00780.

astro-ph.IM↗
arXiv2023

Overview of the Advanced X-ray Imaging Satellite (AXIS)

The Advanced X-ray Imaging Satellite (AXIS) is a Probe-class concept that will build on the legacy of the Chandra X-ray Observatory by providing low-background, arcsecond-resolution imaging in the 0.3-10 keV band across a 450 arcminute$^2$ field of view, with an order of magnitude improvement in sensitivity. AXIS utilizes breakthroughs in the construction of lightweight segmented X-ray optics using single-crystal silicon, and developments in the fabrication of large-format, small-pixel, high readout rate CCD detectors with good spectral resolution, allowing a robust and cost-effective design. Further, AXIS will be responsive to target-of-opportunity alerts and, with onboard transient detection, will be a powerful facility for studying the time-varying X-ray universe, following on from the legacy of the Neil Gehrels (Swift) X-ray observatory that revolutionized studies of the transient X-ray Universe. In this paper, we present an overview of AXIS, highlighting the prime science objectives driving the AXIS concept and how the observatory design will achieve these objectives.

astro-ph.IM↗
arXiv2023

The Accretion History of AGN: The Spectral Energy Distributions of X-ray Luminous AGN

Spectral energy distributions (SEDs) from X-ray to far-infrared (FIR) wavelengths are presented for a sample of 1246 X-ray luminous active galactic nuclei (AGN; $L_{0.5-10\rm{keV}}>10^{43}$ erg s$^{-1}$), with $z_{\rm{spec}}<1.2$, selected from Stripe 82X, COSMOS, and GOODS-N/S. The rest-frame SEDs show a wide spread ($\sim2.5$ dex) in the relative strengths of broad continuum features at X-ray, ultraviolet (UV), mid-infrared (MIR), and FIR wavelengths. A linear correlation (log-log slope of 0.7$\pm0.04$) is found between $L_{\rm{MIR}}$ and $L_{\rm{X}}$. There is significant scatter in the relation between the $L_{\rm{UV}}$ and $L_{\rm{X}}$ due to heavy obscuration, however the most luminous and unobscured AGN show a linear correlation (log-log slope of 0.8$\pm0.06$) in the relation above this scatter. The relation between $L_{\rm{FIR}}$ and $L_{\rm{X}}$ is predominantly flat, but with decreasing dispersion at $L_{\rm{X}}>10^{44}$ erg s$^{-1}$. The ratio between the "galaxy subtracted" bolometric luminosity and the intrinsic $L_{\rm{X}}$ increases from a factor of $\sim$$10-70$ from log $L_{\rm{bol}}/{\rm(erg\; s}^{-1})=44.5-46.5$. Characteristic SED shapes have been determined by grouping AGN based on relative strengths of the UV and MIR emission. The average $L_{1μ\rm{m}}$ is constant for the majority of these SED shapes, while AGN with the strongest UV and MIR emission have elevated $L_{1μ\rm{m}}$, consistent with the AGN emission dominating their SEDs at optical and NIR wavelengths. A strong correlation is found between the SED shape and both the $L_{\rm{X}}$ and $L_{\rm{bol}}$, such that $L_{\rm{bol}}/L_{\rm{X}}=20.4\pm1.8$, independent of the SED shape. This is consistent with an evolutionary scenario of increasing $L_{\rm{bol}}$ with decreasing obscuration as the AGN blows away circumnuclear gas.

astro-ph.GA↗
arXiv2023

Evidence of X-ray emission from the Warm Hot Intergalactic Medium

The Universe has evolved from an initial diffuse, uniform gas to a complex structure that includes both voids and high-density galaxy clusters connected by gaseous filaments, known as the Cosmic Web, and traced by 3D surveys of galaxies. The filamentary structure contains a significant fraction of the baryonic matter and is predicted to be mostly in the form of a moderately high temperature plasma, the Warm Hot Intergalactic Medium. Plasma at this temperature and ionization level emits mostly in soft X-rays. The filamentary structure, however, is hard to detect because the other sources contributing to the Diffuse X-ray Background are much brighter and, currently, there are very few reported detections of emission from the filaments. We report the first high-confidence level indirect detection of X-ray emission from the Warm Hot Intergalactic Medium. Applying the Power Spectrum Analysis to XMM-Newton and eROSITA data, we separated its contribution from other sources modeled in previous studies. Our result is in good agreement with numerical simulations and fills a critical gap in the picture of the large-scale structure of the Universe, in which filamentary gas, galaxies and dark matter interact and co-evolve.

astro-ph.CO↗
arXiv2023

Point source contribution to the Diffuse X-ray Background below 1 keV and its effect on our understanding of the circum-galactic medium

We studied the spectral signature of different components of the Diffuse X-ray Background (DXB), including Local Hot Bubble (LHB), Solar Wind Charge Exchange (SWCX), Galactic Halo, and typically unresolved point sources (galaxies and AGN), in the direction of the Chandra Deep Field South (CDFS) using the 4 Ms XMM-Newton survey and Chandra 4 Ms Source Catalog. In this paper, we present our results showing how the different components contribute to the DXB below 1 keV. In particular, we have found that ~6% of the emission at 3/4 keV (all-sky average value ~ 3$\times10^{-3}$ cm$^{-6}$pc), which is typically associated with Galactic Halo (GH) and Circum-galactic medium (CGM) is, in fact, due to emission from typically unresolved galaxies. We will discuss the effect that this has on our understanding of GH and CGM, and to our understanding of the missing CGM baryons.

astro-ph.HE↗
arXiv2023

Measuring the Cosmic X-ray Background in 3-20keV with Straylight from NuSTAR

By characterizing the contribution of stray light to large datasets from the NuSTAR X-ray observatory collected over 2012--2017, we report a measurement of the cosmic X-ray background in the 3--20 keV energy range. These data represent $\sim20\%$ sky coverage while avoiding Galactic Ridge X-ray emission and are less weighted by deep, survey fields than previous measurements with NuSTAR. Images in narrow energy bands are stacked in detector space and spatially fit with a model representing the stray light and uniform pattern expected from the cosmic X-ray background and the instrumental background, respectively. We establish baseline flux values from Earth-occulted data and validate the fitting method on stray light observations of the Crab, which further serve to calibrate the resulting spectra. We present independent spectra of the cosmic X-ray background with the FPMA and FPMB detector arrays, which are in excellent agreement with the canonical characterization by HEAO 1 and are $10\%$ lower than most subsequent measurements; $F_{\rm{3-20~keV}}^{FPMA} = 2.63 \times 10^{-11}~\rm{erg~s^{-1}~cm^{-2}~deg^{-2}}$ and $F_{\rm{3-20~keV}}^{FPMB} = 2.58 \times 10^{-11}~\rm{erg~s^{-1}~cm^{-2}~deg^{-2}}$. We discuss these results in light of previous measurements of the cosmic X-ray background and consider the impact of systematic uncertainties on our spectra.

astro-ph.HE↗
arXiv2023

QUOTAS: A new research platform for the data-driven investigation of black holes

We present QUOTAS, a novel research platform for the data-driven investigation of super-massive black hole (SMBH) populations. While SMBH data sets -- observations and simulations -- have grown rapidly in complexity and abundance, our computational environments and analysis tools have not matured commensurately to exhaust opportunities for discovery. Motivated to explore BH host galaxy and the parent dark matter halo connection, in this pilot version of QUOTAS, we assemble and co-locate the high-redshift, luminous quasar population at $z \geq 3$ alongside simulated data of the same epochs. Leveraging machine learning algorithms (ML) we expand simulation volumes that successfully replicate halo populations beyond the training set. Training ML on the Illustris-TNG300 simulation that includes baryonic physics, we populate the larger LEGACY Expanse dark matter-only box with quasars. Our first science results comparing observational and ML simulated quasars at $z \sim 3$, reveal that while the recovered Black Hole Mass Functions and clustering are in good agreement, simulated SMBHs fail to accrete, shine and grow at high enough rates to match observed quasars. We conclude that sub-grid models of mass accretion and SMBH feedback implemented in Illustris-TNG300 do not reproduce their observed mass growth. QUOTAS, demonstrates the power of ML, both for analyzing large complex datasets, and offering a unique opportunity to interrogate our theoretical model assumptions. We deploy ML again to derive and devise an optimal survey strategy for bringing the undetected lower luminosity quasar population into view. QUOTAS, and all related materials are publicly available at the Google Kaggle platform.

astro-ph.CO↗
arXiv2023

New metrics to probe the dynamical state of galaxy clusters

We present new diagnostic metrics to probe the dynamical state of galaxy clusters. These novel metrics rely on the computation of the power spectra of the matter and gas distributions and their cross-correlation derived from cluster observations. This analysis permits us to cross-correlate the fluctuations in the matter distribution, inferred from high-resolution lensing mass maps derived from Hubble Space Telescope (HST) data, with those derived from the emitted X-ray surface brightness distribution of the hot Intra-Cluster medium (ICM) from the Chandra X-ray Observatory (CXO). These methodological tools allow us to quantify with unprecedented resolution the coherence with which the gas traces the mass and interrogate the assumption that the gas is in hydro-static equilibrium with the underlying gravitational potential. We characterize departures from equilibrium as a function of scale with a new gas-mass coherence parameter. The efficacy of these metrics is demonstrated by applying them to the analysis of two representative clusters known to be in different dynamical states: the massive merging cluster Abell 2744, from the HST Frontier Fields (HSTFF) sample, and the dynamically relaxed cluster Abell 383, from the Cluster Lensing And Supernova Survey with Hubble (CLASH) sample. Using lensing mass maps in combination with archival Chandra data, and simulated cluster analogs available from the OMEGA500 suite, we quantify the fluctuations in the mass and X-ray surface brightness and show that new insights into the dynamical state of the clusters can be obtained from our gas-mass coherence analysis.

astro-ph.CO↗
arXiv2022

Using Machine Learning to Determine Morphologies of $z<1$ AGN Host Galaxies in the Hyper Suprime-Cam Wide Survey

We present a machine-learning framework to accurately characterize morphologies of Active Galactic Nucleus (AGN) host galaxies within $z<1$. We first use PSFGAN to decouple host galaxy light from the central point source, then we invoke the Galaxy Morphology Network (GaMorNet) to estimate whether the host galaxy is disk-dominated, bulge-dominated, or indeterminate. Using optical images from five bands of the HSC Wide Survey, we build models independently in three redshift bins: low $(0<z<0.25)$, medium $(0.25<z<0.5)$, and high $(0.5<z<1.0)$. By first training on a large number of simulated galaxies, then fine-tuning using far fewer classified real galaxies, our framework predicts the actual morphology for $\sim$ $60\%-70\%$ host galaxies from test sets, with a classification precision of $\sim$ $80\%-95\%$, depending on redshift bin. Specifically, our models achieve disk precision of $96\%/82\%/79\%$ and bulge precision of $90\%/90\%/80\%$ (for the 3 redshift bins), at thresholds corresponding to indeterminate fractions of $30\%/43\%/42\%$. The classification precision of our models has a noticeable dependency on host galaxy radius and magnitude. No strong dependency is observed on contrast ratio. Comparing classifications of real AGNs, our models agree well with traditional 2D fitting with GALFIT. The PSFGAN+GaMorNet framework does not depend on the choice of fitting functions or galaxy-related input parameters, runs orders of magnitude faster than GALFIT, and is easily generalizable via transfer learning, making it an ideal tool for studying AGN host galaxy morphology in forthcoming large imaging survey.

astro-ph.GA↗
arXiv2022

On the cosmic evolution of AGN obscuration and the X-ray luminosity function: XMM-Newton and Chandra spectral analysis of the 31.3 deg$^2$ Stripe 82X

We present X-ray spectral analysis of XMM and Chandra observations in the 31.3 deg$^2$ Stripe-82X (S82X) field. Of the 6181 X-ray sources in this field, we analyze a sample of 2937 active galactic nuclei (AGN) with solid redshifts and sufficient counts determined by simulations. Our results show a population with median values of spectral index $Γ=1.94_{-0.39}^{+0.31}$, column density log$\,N_{\mathrm{H}}/\mathrm{cm}^{-2}=20.7_{-0.5}^{+1.2}$ and intrinsic, de-absorbed, 2-10 keV luminosity log$\,L_{\mathrm{X}}/\mathrm{erg\,s}^{-1}=44.0_{-1.0}^{+0.7}$, in the redshift range 0-4. We derive the intrinsic fraction of AGN that are obscured ($22\leq\mathrm{log}\,N_{\mathrm{H}}/\mathrm{cm}^{-2}<24$), finding a significant increase in the obscured AGN fraction with redshift and a decline with increasing luminosity. The average obscured AGN fraction is $57\pm4\%$ for log$\,L_{\mathrm{X}}/\mathrm{erg\,s}^{-1}>43$. This work constrains the AGN obscuration and spectral shape of the still uncertain high-luminosity and high-redshift regimes (log$\,L_{\mathrm{X}}/\mathrm{erg\,s}^{-1}>45.5$, $z>3$), where the obscured AGN fraction rises to $64\pm12\%$. We report a luminosity and density evolution of the X-ray luminosity function, with obscured AGN dominating at all luminosities at $z>2$ and unobscured sources prevailing at log$\,L_{\mathrm{X}}/\mathrm{erg\,s}^{-1}>45$ at lower redshifts. Our results agree with evolutionary models in which the bulk of AGN activity is triggered by gas-rich environments and in a downsizing scenario. Also, the black hole accretion density (BHAD) is found to evolve similarly to the star formation rate density, confirming the co-evolution between AGN and host-galaxy, but suggesting different time scales in their growing history. The derived BHAD evolution shows that Compton-thick AGN contribute to the accretion history of AGN as much as all other AGN populations combined.

astro-ph.GA↗
arXiv2022

The X-ray Angular Power Spectrum of Extended Sources in the eROSITA Final Equatorial Depth Survey

The eROSITA Final Equatorial Depth Survey (eFEDS), with a sky area of 140 square degrees with depth equivalent to the equatorial patch of the final eROSITA all-sky survey, represents the largest continuous non-full-sky X-ray fields to-date, making it the premier data set for measuring the angular power spectrum. In this work, we measure the X-ray angular power spectrum of galaxy clusters and groups in the eFEDS field. We show that the measured power spectrum is consistent with past observations, including the ROSAT All Sky Survey, and the Chandra COSMOS and Bootes fields. The predictions of cluster gas halo model that is calibrated from Chandra observations is also consistent with the eFEDS power spectrum. While the eFEDS does not have large enough sky coverage to provide meaningful cosmological constraints, we predict that the X-ray power spectrum from the cycle 4 of the eROSITA All-Sky Survey (eRASS4) will provide constraints on $Ω_M$ and $σ_8$ at the 10% level.

astro-ph.CO↗
arXiv2022

Constraining Sterile Neutrino Dark Matter in the Milky Way Halo with Swift-XRT

We present a search for sterile neutrino dark matter decay signals in the Milky Way's dark matter halo by considering the entirety of the Swift-XRT data archive. After filtering the raw archive, we analyze a $\sim$77 Ms data set containing the full field of view, as well as a $\sim$41 Ms data set with point-sources excised using the Swift-XRT Point Source catalog. We report non-detections of emission lines across the 3--6 keV continuum in both data sets, including at and around 3.5 keV. The point-sources excised data set is found to have higher sensitivity to faint dark matter decay signals due to its freedom from point-source contamination and is thus used to set constraints. Non-detections across the total data set's continuum are used to constrain the sterile neutrino dark matter parameter space, marginally strengthening existing X-ray constraints. Non-detections at $\sim$3.5 keV in data subsets grouped by angular distance from the galactic center are used to constrain the 3.5 keV line's galactic intensity profile, providing the strongest constraints to date across $\sim$1/4 of the galaxy.

astro-ph.HE↗
arXiv2022

X-Ray Redshifts of Obscured Chandra Source Catalog AGN

We have computed obscured AGN redshifts using the XZ method, adopting a broad treatment in which we employed a wide-ranging data set and worked primarily at the XZ counts sensitivity threshold, culminating with a redshift catalog containing 121 sources that lack documented redshifts. We considered 363 obscured AGN from the Chandra Source Catalog Release 2.0, 59 of which were selected using multiwavelength criteria while 304 were X-ray selected. One-third of the data set had cross-matched spectroscopic or photometric redshifts. These sources, dominated by low-$z$ and low-$N_H$ AGN, were supplemented by 1000 simulations to form a data set for testing the XZ method. We used a multi-layer perceptron neural network to examine and predict cases in which XZ fails to reproduce the known redshift, yielding a classifier that can identify and discard poor redshift estimates. This classifier demonstrated a statistically significant $\sim$3$σ$ improvement over the existing XZ redshift information gain filter. We applied the machine learning model to sources with no documented redshifts, resulting in the 121-source new redshift catalog, all of which were X-ray selected. Our neural network's performance suggests that nearly 90% of these redshift estimates are consistent with hypothetical spectroscopic or photometric measurements, strengthening the notion that redshifts can be reliably estimated using only X-rays, which is valuable to current and future missions such as Athena. We have also identified a possible Compton-thick candidate that warrants further investigation.

astro-ph.HE↗
arXiv2022

GaMPEN: A Machine Learning Framework for Estimating Bayesian Posteriors of Galaxy Morphological Parameters

We introduce a novel machine learning framework for estimating the Bayesian posteriors of morphological parameters for arbitrarily large numbers of galaxies. The Galaxy Morphology Posterior Estimation Network (GaMPEN) estimates values and uncertainties for a galaxy's bulge-to-total light ratio ($L_B/L_T$), effective radius ($R_e$), and flux ($F$). To estimate posteriors, GaMPEN uses the Monte Carlo Dropout technique and incorporates the full covariance matrix between the output parameters in its loss function. GaMPEN also uses a Spatial Transformer Network (STN) to automatically crop input galaxy frames to an optimal size before determining their morphology. This will allow it to be applied to new data without prior knowledge of galaxy size. Training and testing GaMPEN on galaxies simulated to match $z < 0.25$ galaxies in Hyper Suprime-Cam Wide $g$-band images, we demonstrate that GaMPEN achieves typical errors of $0.1$ in $L_B/L_T$, $0.17$ arcsec ($\sim 7\%$) in $R_e$, and $6.3\times10^4$ nJy ($\sim 1\%$) in $F$. GaMPEN's predicted uncertainties are well-calibrated and accurate ($<5\%$ deviation) -- for regions of the parameter space with high residuals, GaMPEN correctly predicts correspondingly large uncertainties. We also demonstrate that we can apply categorical labels (i.e., classifications such as "highly bulge-dominated") to predictions in regions with high residuals and verify that those labels are $\gtrsim 97\%$ accurate. To the best of our knowledge, GaMPEN is the first machine learning framework for determining joint posterior distributions of multiple morphological parameters and is also the first application of an STN to optical imaging in astronomy.

astro-ph.GA↗