arXiv subjects

Kevin Schawinski

Publications and source records attributed to Kevin Schawinski.

At least 37 records · Page 2Linked to original sources

arXiv2019

The CO(3-2)/CO(1-0) luminosity line ratio in nearby star-forming galaxies and AGN from xCOLD GASS, BASS and SLUGS

We study the r31=L'CO(3-2)/L'CO(1-0) luminosity line ratio in a sample of nearby (z < 0.05) galaxies: 25 star-forming galaxies (SFGs) from the xCOLD GASS survey, 36 hard X-ray selected AGN host galaxies from BASS and 37 infrared luminous galaxies from SLUGS. We find a trend for r31 to increase with star-formation efficiency (SFE). We model r31 using the UCL-PDR code and find that the gas density is the main parameter responsible for variation of r31, while the interstellar radiation field and cosmic ray ionization rate play only a minor role. We interpret these results to indicate a relation between SFE and gas density. We do not find a difference in the r31 value of SFGs and AGN host galaxies, when the galaxies are matched in SSFR ( = 0.52 +/- 0.04 for SFGs and = 0.53 +/- 0.06 for AGN hosts). According to the results of UCL-PDR models, the X-rays can contribute to the enhancement of the CO line ratio, but only for strong X-ray fluxes and for high gas density (nH > 10$^4$ cm-3). We find a mild tightening of the Kennicutt-Schmidt relation when we use the molecular gas mass surface density traced by CO(3-2) (Pearson correlation coefficient R=0.83), instead of the molecular gas mass surface density traced by CO(1-0) (R=0.78), but the increase in correlation is not statistically significant (p-value=0.06). This suggests that the CO(3-2) line can be reliably used to study the relation between SFR and molecular gas for normal SFGs at high redshift, and to compare it with studies of low-redshift galaxies, as is common practice.

astro-ph.GA↗

arXiv2019

A forward modelling approach to AGN variability -- Method description and early applications

We present a numerical framework for the variability of active galactic nuclei (AGN), which links the variability of AGN over a broad range of timescales and luminosities to the observed properties of the AGN population as a whole, and particularly the Eddington ratio distribution function (ERDF). We have implemented our framework on GPU architecture, relying on previously published time series generating algorithms. After extensive tests that characterise several intrinsic and numerical aspects of the simulations, we describe some applications used for current and future time domain surveys and for the study of extremely variable sources (e.g., "changing look" or flaring AGN). Specifically, we define a simulation setup which reproduces the AGN variability observed in the PTF/iPTF survey, and use it to forward model longer light curves of the kind that may be observed within the LSST main survey. Thanks to our effcient implementations, these simulations are able to cover for example over 1 Myr with a roughly weekly cadence. We envision that this framework will become highly valuable to prepare for, and best exploit, data from upcoming time domain surveys, such as for example LSST.

astro-ph.GA↗

arXiv2019

BAT AGN Spectroscopic Survey -- XIII. The nature of the most luminous obscured AGN in the low-redshift universe

We present a multi wavelength analysis of 28 of the most luminous low-redshift narrow-line, ultra-hard X-ray selected active galactic nuclei (AGN) drawn from the 70 month Swift/BAT all-sky survey, with bolometric luminosities of log(L_bol/erg/s) > 45.25. The broad goal of our study is to determine whether these objects have any distinctive properties, potentially setting them aside from lower-luminosity obscured AGN in the local Universe. Our analysis relies on the first data release of the BAT AGN Spectroscopic Survey (BASS/DR1) and on dedicated observations with the VLT, Palomar, and Keck observatories. We find that the vast majority of our sources agree with commonly used AGN selection criteria which are based on emission line ratios and on mid-infrared colours. Our AGN are predominantly hosted in massive galaxies (9.8 < log(M_*/M_sun) < 11.7); based on visual inspection of archival optical images, they appear to be mostly ellipticals. Otherwise, they do not have distinctive properties. Their radio luminosities, determined from publicly available survey data, show a large spread of almost 4 orders of magnitude - much broader than what is found for lower X-ray luminosity obscured AGN in BASS. Moreover, our sample shows no preferred combination of black hole masses (M_BH) and/or Eddington ratio (lambda_Edd), covering 7.5 < log(M_BH/M_sun) < 10.3 and 0.01 < lambda_Edd < 1. Based on the distribution of our sources in the lambda_Edd-N_H plane, we conclude that our sample is consistent with a scenario where the amount of obscuring material along the line of sight is determined by radiation pressure exerted by the AGN on the dusty circumnuclear gas.

astro-ph.GA↗

arXiv2019

BAT AGN Spectroscopic Survey -- XVII: The Parsec-scale Jet Properties of the Ultra Hard X-ray Selected Local AGN

We have performed a very long baseline interferometry (VLBI) survey of local (z < 0.05) ultra hard X-ray (14-195 keV) selected active galactic nuclei (AGN) from the Swift Burst Alert Telescope (BAT) using KVN, KaVA, and VLBA. We first executed fringe surveys of 142 BAT-detected AGN at 15 or 22 GHz. Based on the fringe surveys and archival data, we find 10/279 nearby AGN (~4%) VLBI have 22 GHz flux above 30 mJy. This implies that the X-ray AGN with a bright nuclear jet are not common. Among these 10 radio-bright AGN, we obtained 22 GHz VLBI imaging data of our own for four targets and reprocessed archival data for six targets. We find that, although our 10 AGN observed with VLBI span a wide range of pc-scale morphological types, they lie on a tight linear relation between accretion luminosity and nuclear jet luminosity. Our result suggests that a powerful nuclear radio jet correlates with the accretion disc luminosity. We also probed the fundamental plane of black hole activity at VLBI scales (e.g., few milli-arcsecond). The jet luminosity and size distribution among our sample roughly fit into the proposed AGN evolutionary scenario, finding powerful jets after the blow-out phase based on the Eddington ratio (λ_{Edd})-hydrogen column density (N_{H}) relation. In addition, we find some hints of gas inflow or galaxy-galaxy merger in the majority of our sample. This implies that gas supply via tidal interactions in galactic scale may help the central AGN to launch a powerful parsec-scale jet.

astro-ph.GA↗

arXiv2019

RadioGAN - Translations between different radio surveys with generative adversarial networks

Radio surveys are widely used to study active galactic nuclei. Radio interferometric observations typically trade-off surface brightness sensitivity for angular resolution. Hence, observations using a wide range of baseline lengths are required to recover both bright small-scale structures and diffuse extended emission. We investigate if generative adversarial networks (GANs) can extract additional information from radio data and might ultimately recover extended flux from a survey with a high angular resolution and vice versa. We use a GAN for the image-to-image translation between two different data sets, namely the Faint Images of the Radio Sky at Twenty-Centimeters (FIRST) and the NRAO VLA Sky Survey (NVSS) radio surveys. The GAN is trained to generate the corresponding image cutout from the other survey for a given input. The results are analyzed with a variety of metrics, including structural similarity as well as flux and size comparison of the extracted sources. RadioGAN is able to recover extended flux density within a $20\%$ margin for almost half of the sources and learns more complex relations between sources in the two surveys than simply convolving them with a different synthesized beam. RadioGAN is also able to achieve subbeam resolution by recognizing complicated underlying structures from unresolved sources. RadioGAN generates over a third of the sources within a $20\%$ deviation from both original size and flux for the FIRST to NVSS translation, while for the NVSS to FIRST mapping it achieves almost $30\%$ within this range.

astro-ph.GA↗

arXiv2019

The composite nature of Dust-Obscured Galaxies (DOGs) at z~2-3 in the COSMOS field: II. The AGN fraction

We present the X-ray properties of 108 Dust-Obscured Galaxies (DOGs; F$_{24 μm}$/F$_{R} >$ 1000) in the COSMOS field, all of which detected in at least three far-infrared bands with the Herschel Observatory. Out of the entire sample, 22 are individually detected in the hard 2-8 keV X-ray band by the Chandra COSMOS Legacy survey, allowing us to classify them as AGN. Of them, 6 (27%) are Compton Thick AGN candidates with column densities N$_{H}$$>$10$^{24}$ cm$^{-2}$ while 15 are moderately obscured AGNs with 10$^{22}$ $<$ N$_{H}$ $<$ 10$^{24}$ cm$^{-2}$. Additionally, we estimate AGN contributions to the IR luminosity (8-1000$μ$m rest-frame) greater than 20% for 19 DOGs based on SED decomposition using Spitzer/MIPS 24$μ$m and the five Herschel bands (100-500 $μ$m). Only 7 of these are detected in X-rays individually. We performed a X-ray stacking analysis for the 86 undetected DOGs. We find that the AGN fraction in DOGs increases with 24$μ$m flux and that it is higher than that of the general 24$μ$m population. However, no significant difference is found when considering only X-ray detections. This strongly motivates the combined use of X-ray and far-IR surveys to successfully probe a wider population of AGNs, particularly for the most obscured ones.

astro-ph.GA↗

arXiv2019

Continuous Integration of Machine Learning Models with ease.ml/ci: Towards a Rigorous Yet Practical Treatment

Continuous integration is an indispensable step of modern software engineering practices to systematically manage the life cycles of system development. Developing a machine learning model is no difference - it is an engineering process with a life cycle, including design, implementation, tuning, testing, and deployment. However, most, if not all, existing continuous integration engines do not support machine learning as first-class citizens. In this paper, we present ease.ml/ci, to our best knowledge, the first continuous integration system for machine learning. The challenge of building ease.ml/ci is to provide rigorous guarantees, e.g., single accuracy point error tolerance with 0.999 reliability, with a practical amount of labeling effort, e.g., 2K labels per test. We design a domain specific language that allows users to specify integration conditions with reliability constraints, and develop simple novel optimizations that can lower the number of labels required by up to two orders of magnitude for test conditions popularly used in real production systems.

cs.LG↗

arXiv2019

On the Prevalence of Super-Massive Black Holes over Cosmic Time

We investigate the abundance of Super-Massive Black Hole (SMBH) seeds in primordial galaxy halos. We explore the assumption that dark matter halos outgrowing a critical halo mass M_c have some probability p of having spawned a SMBH seed. Current observations of local, intermediate-mass galaxies constrain these parameters: For $M_c=10^{11}M_\odot$, all halos must be seeded, but when adopting smaller M_c masses the seeding can be much less efficient. The constraints also put lower limits on the number density of black holes in the local and high-redshift Universe. Reproducing z~6 quasar space densities depends on their typical halo mass, which can be constrained by counting nearby Lyman Break Galaxies and Lyman Alpha Emitters. For both observables, our simulations demonstrate that single-field predictions are too diverse to make definitive statements, in agreement with mixed claims in the literature. If quasars are not limited to the most massive host halos, they may represent a tiny fraction (~10^-5) of the SMBH population. Finally, we produce a wide range of predictions for gravitational events from SMBH mergers. We define a new diagnostic diagram for LISA to measure both SMBH space density and the typical delay between halo merger and black hole merger. While previous works have explored specific scenarios, our results hold independent of the seed mechanism, seed mass, obscuration, fueling methods and duty cycle.

astro-ph.GA↗

arXiv2018

AGN photoionization of gas in companion galaxies as a probe of AGN radiation in time and direction

We consider AGN photoionization of gas in companion galaxies (cross-ionization) as a way to sample the intensity of AGN radiation in both direction and time, independent of the gas properties of the AGN host galaxies. From an initial set of 212 AGN+companion systems, identified with the help of Galaxy Zoo participants, we obtained long-slit optical spectra of 32 pairs which were a priori likely to show cross-ionization based on projected separation or angular extent of the companion. From emission-line ratios, 10 of these systems are candidates for cross-ionization, roughly the fraction expected if most AGN have ionization cones with 70-degree opening angles. Among these, Was 49 remains the strongest nearby candidate. NGC 5278/9 and UGC 6081 are dual-AGN systems with tidal debris, complicating identification of cross-ionization. The two weak AGN in the NGC 5278/9 system ionize gas filaments to a projected radius 14 kpc from each galaxy. In UGC 6081, an irregular high-ionization emission region encompasses both AGN, extending more than 15 kpc from each. The observed AGN companion galaxies with and without signs of external AGN photoionization have similar distributions in estimated incident AGN flux, suggesting that geometry of escaping radiation or long-term variability control this facet of the AGN environment. This parallels conclusions for luminous QSOs based on the proximity effect among Lyman-alpha absorbers. In some galaxies, mismatch between spectroscopic classifications in the common BPT diagram and the intensity of weaker He II and [Ne V] emission lines highlights the limits of common classifications in low-metallicity environments.

astro-ph.GA↗

arXiv2018

Exploring galaxy evolution with generative models

Context. Generative models open up the possibility to interrogate scientific data in a more data-driven way. Aims: We propose a method that uses generative models to explore hypotheses in astrophysics and other areas. We use a neural network to show how we can independently manipulate physical attributes by encoding objects in latent space. Methods: By learning a latent space representation of the data, we can use this network to forward model and explore hypotheses in a data-driven way. We train a neural network to generate artificial data to test hypotheses for the underlying physical processes. Results: We demonstrate this process using a well-studied process in astrophysics, the quenching of star formation in galaxies as they move from low-to high-density environments. This approach can help explore astrophysical and other phenomena in a way that is different from current methods based on simulations and observations.

astro-ph.GA↗

arXiv2018

A population of luminous accreting black holes with hidden mergers

Major galaxy mergers are thought to play an important part in fuelling the growth of supermassive black holes. However, observational support for this hypothesis is mixed, with some studies showing a correlation between merging galaxies and luminous quasars and others showing no such association. Recent observations have shown that a black hole is likely to become heavily obscured behind merger-driven gas and dust, even in the early stages of the merger, when the galaxies are well separated (5 to 40 kiloparsecs). Merger simulations further suggest that such obscuration and black-hole accretion peaks in the final merger stage, when the two galactic nuclei are closely separated (less than 3 kiloparsecs). Resolving this final stage requires a combination of high-spatial-resolution infrared imaging and high-sensitivity hard-X-ray observations to detect highly obscured sources. However, large numbers of obscured luminous accreting supermassive black holes have been recently detected nearby (distances below 250 megaparsecs) in X-ray observations. Here we report high-resolution infrared observations of hard-X-ray-selected black holes and the discovery of obscured nuclear mergers, the parent populations of supermassive-black-hole mergers. We find that obscured luminous black holes (bolometric luminosity higher than 2x10^44 ergs per second) show a significant (P<0.001) excess of late-stage nuclear mergers (17.6 per cent) compared to a sample of inactive galaxies with matching stellar masses and star formation rates (1.1 per cent), in agreement with theoretical predictions. Using hydrodynamic simulations, we confirm that the excess of nuclear mergers is indeed strongest for gas-rich major-merger hosts of obscured luminous black holes in this final stage.

astro-ph.GA↗

arXiv2018

Radio Galaxy Zoo: ClaRAN - A Deep Learning Classifier for Radio Morphologies

The upcoming next-generation large area radio continuum surveys can expect tens of millions of radio sources, rendering the traditional method for radio morphology classification through visual inspection unfeasible. We present ClaRAN - Classifying Radio sources Automatically with Neural networks - a proof-of-concept radio source morphology classifier based upon the Faster Region-based Convolutional Neutral Networks (Faster R-CNN) method. Specifically, we train and test ClaRAN on the FIRST and WISE images from the Radio Galaxy Zoo Data Release 1 catalogue. ClaRAN provides end users with automated identification of radio source morphology classifications from a simple input of a radio image and a counterpart infrared image of the same region. ClaRAN is the first open-source, end-to-end radio source morphology classifier that is capable of locating and associating discrete and extended components of radio sources in a fast (< 200 milliseconds per image) and accurate (>= 90 %) fashion. Future work will improve ClaRAN's relatively lower success rates in dealing with multi-source fields and will enable ClaRAN to identify sources on much larger fields without loss in classification accuracy.

astro-ph.IM↗

arXiv2018

ALMA [CI]$^{3}P_{1}-^{3}P_{0}$ observations of NGC6240: a puzzling molecular outflow, and the role of outflows in the global $α_{\rm CO}$ factor of (U)LIRGs

We present ALMA and ACA [CI]$^{3}P_{1}-^{3}P_{0}$ ([CI](1-0)) observations of NGC6240, which we combine with ALMA CO(2-1) and IRAM PdBI CO(1-0) data to study the physical properties of the massive molecular (H$_2$) outflow. We discover that the receding and approaching sides of the H$_2$ outflow, aligned east-west, exceed 10 kpc in their total extent. High resolution ($0.24"$) [CI](1-0) line images surprisingly reveal that the outflow emission peaks between the two AGNs, rather than on either of the two, and that it dominates the velocity field in this nuclear region. We combine the [CI](1-0) and CO(1-0) data to constrain the CO-to-H$_2$ conversion factor ($α_{\rm CO}$) in the outflow, which is on average $2.1\pm1.2~\rm M_{\odot} (K~km~s^{-1}~pc^2)^{-1}$. We estimate that $60\pm20$ % of the total H$_2$ gas reservoir of NGC6240 is entrained in the outflow, for a resulting mass-loss rate of $\dot{M}_{\rm out}=2500\pm1200~M_{\odot}~yr^{-1}$ $\equiv 50\pm30$ SFR. This energetics rules out a solely star formation-driven wind, but the puzzling morphology challenges a classic radiative-mode AGN feedback scenario. For the quiescent gas we compute $\langleα_{\rm CO}\rangle = 3.2\pm1.8~\rm M_{\odot} (K~km~s^{-1}~pc^2)^{-1}$, which is at least twice the value commonly employed for (U)LIRGs. We observe a tentative trend of increasing $r_{21}\equiv L^{\prime}_{\rm CO(2-1)}/L^{\prime}_{\rm CO(1-0)}$ ratios with velocity dispersion and measure $r_{21}>1$ in the outflow, whereas $r_{21}\simeq1$ in the quiescent gas. We propose that molecular outflows are the location of the warmer, strongly unbound phase that partially reduces the opacity of the CO lines in (U)LIRGs, hence driving down their global $α_{\rm CO}$ and increasing their $r_{21}$ values.

astro-ph.GA↗

arXiv2018

Using transfer learning to detect galaxy mergers

We investigate the use of deep convolutional neural networks (deep CNNs) for automatic visual detection of galaxy mergers. Moreover, we investigate the use of transfer learning in conjunction with CNNs, by retraining networks first trained on pictures of everyday objects. We test the hypothesis that transfer learning is useful for improving classification performance for small training sets. This would make transfer learning useful for finding rare objects in astronomical imaging datasets. We find that these deep learning methods perform significantly better than current state-of-the-art merger detection methods based on nonparametric systems like CAS and GM$_{20}$. Our method is end-to-end and robust to image noise and distortions; it can be applied directly without image preprocessing. We also find that transfer learning can act as a regulariser in some cases, leading to better overall classification accuracy ($p = 0.02$). Transfer learning on our full training set leads to a lowered error rate from 0.038 $\pm$ 1 down to 0.032 $\pm$ 1, a relative improvement of 15%. Finally, we perform a basic sanity-check by creating a merger sample with our method, and comparing with an already existing, manually created merger catalogue in terms of colour-mass distribution and stellar mass function.

astro-ph.IM↗

arXiv2018

PSFGAN: a generative adversarial network system for separating quasar point sources and host galaxy light

The study of unobscured active galactic nuclei (AGN) and quasars depends on the reliable decomposition of the light from the AGN point source and the extended host galaxy light. The problem is typically approached using parametric fitting routines using separate models for the host galaxy and the point spread function (PSF). We present a new approach using a Generative Adversarial Network (GAN) trained on galaxy images. We test the method using Sloan Digital Sky Survey (SDSS) r-band images with artificial AGN point sources added which are then removed using the GAN and with parametric methods using GALFIT. When the AGN point source PS is more than twice as bright as the host galaxy, we find that our method, PSFGAN, can recover PS and host galaxy magnitudes with smaller systematic error and a lower average scatter ($49\%$). PSFGAN is more tolerant to poor knowledge of the PSF than parametric methods. Our tests show that PSFGAN is robust against a broadening in the PSF width of $\pm 50\%$ if it is trained on multiple PSF's. We demonstrate that while a matched training set does improve performance, we can still subtract point sources using a PSFGAN trained on non-astronomical images. While initial training is computationally expensive, evaluating PSFGAN on data is more than $40$ times faster than GALFIT fitting two components. Finally, PSFGAN it is more robust and easy to use than parametric methods as it requires no input parameters.

astro-ph.GA↗

arXiv2018

A model for AGN variability on multiple timescales

We present a framework to link and describe AGN variability on a wide range of timescales, from days to billions of years. In particular, we concentrate on the AGN variability features related to changes in black hole fuelling and accretion rate. In our framework, the variability features observed in different AGN at different timescales may be explained as realisations of the same underlying statistical properties. In this context, we propose a model to simulate the evolution of AGN light curves with time based on the probability density function (PDF) and power spectral density (PSD) of the Eddington ratio ($L/L_{\rm Edd}$) distribution. Motivated by general galaxy population properties, we propose that the PDF may be inspired by the $L/L_{\rm Edd}$ distribution function (ERDF), and that a single (or limited number of) ERDF+PSD set may explain all observed variability features. After outlining the framework and the model, we compile a set of variability measurements in terms of structure function (SF) and magnitude difference. We then combine the variability measurements on a SF plot ranging from days to Gyr. The proposed framework enables constraints on the underlying PSD and the ability to link AGN variability on different timescales, therefore providing new insights into AGN variability and black hole growth phenomena.

astro-ph.GA↗

arXiv2018

The fraction of AGN in major merger galaxies and its luminosity dependence

We use a phenomenological model which connects the galaxy and AGN populations to investigate the process of AGN triggering through major galaxy mergers at z~0. The model uses stellar mass functions as input and allows the prediction of AGN luminosity functions based on assumed Eddington ratio distribution functions (ERDFs). We show that the number of AGN hosted by merger galaxies relative to the total number of AGN increases as a function of AGN luminosity. This is due to more massive galaxies being more likely to undergo a merger and does not require the assumption that mergers lead to higher Eddington ratios than secular processes. Our qualitative analysis also shows that to match the observations, the probability of a merger galaxy hosting an AGN and accreting at a given Eddington value has to be increased by a factor ~10 relative to the general AGN population. An additional significant increase of the fraction of high Eddington ratio AGN among merger host galaxies leads to inconsistency with the observed X-ray luminosity function. Physically our results imply that, compared to the general galaxy population, the AGN fraction among merger galaxies is ~10 times higher. On average, merger triggering does however not lead to significantly higher Eddington ratios.

astro-ph.GA↗

arXiv2018

The 105 month Swift-BAT all-sky hard X-ray survey

We present a catalog of hard X-ray sources detected in the first 105 months of observations with the Burst Alert Telescope (BAT) coded mask imager on board the Swift observatory. The 105 month Swift-BAT survey is a uniform hard X-ray all-sky survey with a sensitivity of $8.40\times 10^{-12}\ {\rm erg\ s^{-1}\ cm^{-2}}$ over 90% of the sky and $7.24\times 10^{-12}\ {\rm erg\ s^{-1}\ cm^{-2}}$ over 50% of the sky in the 14-195 keV band. The Swift-BAT 105 month catalog provides 1632 (422 new detections) hard X-ray sources in the 14-195 keV band above the 4.8σ significance level. Adding to the previously known hard X-ray sources, 34% (144/422) of the new detections are identified as Seyfert AGN in nearby galaxies (z<0.2). The majority of the remaining identified sources are X-ray binaries (7%, 31) and blazars/BL Lac objects (10%, 43). As part of this new edition of the Swift-BAT catalog, we release eight-channel spectra and monthly sampled light curves for each object in the online journal and at the Swift-BAT 105 month Web site.

astro-ph.HE↗