Search

We aim to identify QSO2 candidates in the redshift desert using optical and infrared photometry. At this intermediate redshift range, most of the prominent optical emission lines in QSO2 sources (e.g. CIV1549; [OIII]4959,5008) fall either outside the wavelength range of the SDSS optical spectra or in particularly noisy wavelength ranges, making QSO2 identification challenging. Therefore, we adopted a semi-supervised machine learning approach to select candidates in the SDSS galaxy sample. Recent applications of machine learning in astronomy focus on problems involving large data sets, with small data sets often being overlooked. We developed a few-shot learning approach for the identification and classification of rare-object classes using limited training data (200 sources). The new AMELIA pipeline uses a transfer-learning based approach with decision trees, distance-based, and deep learning methods to build a classifier capable of identifying rare objects on the basis of an observational training data set. We validated the performance of AMELIA by addressing the problem of identifying QSO2s at 1 $\leq$ z $\leq$ 2 using SDSS and WISE photometry, obtaining an F1-score above 0.8 in a supervised approach. We then used AMELIA to select new QSO2 candidates in the redshift desert and examined the nature of the candidates using SDSS spectra, when available. In particular, we identified a sub-population of [NeV]3426 emitters at z $\sim$ 1.1, which are highly likely to contain obscured AGNs. We used X-ray and radio cross-matching to validate our classification and investigated the performance of photometric criteria from the literature showing that our candidates have an inherent dusty nature. Finally, we derived physical properties for our QSO2 sample using photoionisation models and verified the AGN classification using an SED fitting., Comment: 20 pages, 9 figures, Accepted for publication in A&A

We developed and trained a pipeline of three machine learning (ML) models than can predict which sources are more likely to be an AGN and to be detected in specific radio surveys. Also, it can estimate redshift values for predicted radio-detectable AGNs. These models, which combine predictions from tree-based and gradient-boosting algorithms, have been trained with multi-wavelength data from near-infrared-selected sources in the Hobby-Eberly Telescope Dark Energy Experiment (HETDEX) Spring field. Training, testing, calibration, and validation were carried out in the HETDEX field. Further validation was performed on near-infrared-selected sources in the Stripe 82 field. In the HETDEX validation subset, our pipeline recovers 96% of the initially labelled AGNs and, from AGNs candidates, we recover 50% of previously detected radio sources. For Stripe 82, these numbers are 94% and 55%. Compared to random selection, these rates are two and four times better for HETDEX, and 1.2 and 12 times better for Stripe 82. The pipeline can also recover the redshift distribution of these sources with $\sigma_{\mathrm{NMAD}}$ = 0.07 for HETDEX ($\sigma_{\mathrm{NMAD}}$ = 0.09 for Stripe 82) and an outlier fraction of 19% (25% for Stripe 82), compatible with previous results based on broad-band photometry. Feature importance analysis stresses the relevance of near- and mid-infrared colours to select AGNs and identify their radio and redshift nature. Combining different algorithms in ML models shows an improvement in the prediction power of our pipeline over a random selection of sources. Tree-based ML models (in contrast to deep learning techniques) facilitate the analysis of the impact that features have on the predictions. This prediction can give insight into the potential physical interplay between the properties of radio AGNs (e.g. mass of black hole and accretion rate)., Comment: Accepted for publication in A&A, 24 pages and 21 figures. Updated to include information on retrieval of data and models, which can be obtained from https://zenodo.org/doi/10.5281/zenodo.10220008

In the era of huge astronomical surveys, machine learning offers promising solutions for the efficient estimation of galaxy properties. The traditional, `supervised' paradigm for the application of machine learning involves training a model on labelled data, and using this model to predict the labels of previously unlabelled data. The semi-supervised `pseudo-labelling' technique offers an alternative paradigm, allowing the model training algorithm to learn from both labelled data and as-yet unlabelled data. We test the pseudo-labelling method on the problems of estimating redshift, stellar mass, and star formation rate, using COSMOS2015 broad band photometry and one of several publicly available machine learning algorithms, and we obtain significant improvements compared to purely supervised learning. We find that the gradient-boosting tree methods CatBoost, XGBoost, and LightGBM benefit the most, with reductions of up to ~15% in metrics of absolute error. We also find similar improvements in the photometric redshift catastrophic outlier fraction. We argue that the pseudo-labellng technique will be useful for the estimation of redshift and physical properties of galaxies in upcoming large imaging surveys such as Euclid and LSST, which will provide photometric data for billions of sources., Comment: 10 pages, 6 figures, accepted for publication in MNRAS

The Hubble Frontier Fields offer an exceptionally deep window into the high-redshift universe, covering a substantially larger area than the Hubble Ultra-Deep field at low magnification and probing 1--2 mags deeper in exceptional high-magnification regions. We aim to leverage Atacama Large Millimetre Array (ALMA) band 6 ($\approx$263\,GHz) mosaics in the central portions of five Frontier Fields to characterize the infrared (IR) properties of $1582$ ultraviolet (UV)-selected Lyman-Break Galaxies (LBGs) at redshifts of $z {\sim}$2--8. We investigated individual and stacked fluxes and IR excess (IRX) values of the LBG sample as functions of stellar mass ($\mathrm{M}_{\bigstar}$), redshift, UV luminosity and slope $\beta$, and lensing magnification. Two (2) LBG candidates were individually detected above a significance of $4.1{-}\sigma$, while stacked samples of the remaining LBG candidates yielded no significant detections. We investigated our detections and upper limits in the context of the IRX-$\mathrm{M}_{\bigstar}$ and IRX-$\beta$ relations, probing at least one dex lower in stellar mass than past studies have done. Our upper limits exclude substantial portions of parameter space and they are sufficiently deep in a handful of cases to create mild tension with the typically assumed attenuation and consensus relations. We observe a clear and smooth trend between $\mathrm{M}_{\bigstar}$ and $\beta$, which extends to low masses and blue (low) $\beta$ values, consistent with expectations from previous works., Comment: Accepted for publication in A&A, 28 pages and 20 figures. Full versions of Tables C.1, C.2, C.3, C.4, and C.5 are available for download at http://astro.puc.cl/~rcarvaja/docs/tables_ALMA_FF_V_RCP.tar.gz

[abridged] The Hubble and Spitzer Space Telescope surveys of the Frontier Fields (FF) provide extremely deep images around six massive, strong-lensing clusters of galaxies. The ALMA FF survey aims to cover the same fields at 1.1mm, with maps reaching (unlensed) sensitivities of $<$70$\mu$Jy, in order to explore the properties of background dusty star-forming galaxies. We report on the multi-wavelength photometric analysis of all 12 significantly detected ($>$5$\sigma$) sources in the first three FF clusters observed by ALMA, based on data from Hubble and Spitzer, the VLT and Herschel. We measure the total photometry in all available bands and determine the photometric redshifts and the physical properties of the counterparts via SED-fitting. In particular, we carefully estimate the FIR photometry using 1.1mm priors to limit the misidentification of blended FIR counterparts, which strongly affect some flux estimates in previous FIR catalogs. We identify robust near-infrared (NIR) counterparts for all 11 sources with K$_s$ detection, the majority of which are quite red, with eight having $F814W-K_s\gtrsim 4$ and five having $F160W-[4.5]\gtrsim3$. From the FIR point of view, all our objects have $z_{phot}$$\sim$1--3, whereas based on the optical SED one object prefers a high-$z$ solution ($z\geq\ $7). Five objects among our sample have spectroscopic redshifts from the GLASS survey for which we can reproduce their SEDs with existing templates. This verification confirms the validity of our photometric redshift methodology. The mean redshift of our sample is $z_{phot}$=1.99$\pm$0.27. All 1.1mm selected objects are massive (10.0$<\log[M_{\star}(M_{\odot})]<$ 11.5), with high star formation rates ($<\log[SFR(M_{\odot}/yr)]> \approx$1.6) and high dust contents (8.1 $<\log[M_{dust} (M_{\odot})]<$8.8), consistent with previous ALMA surveys., Comment: 18 pages, 11 figures, accepted for publication in A&A

Most sub-mm emission line studies of galaxies to date have targeted sources with known redshifts where the frequencies of the lines are well constrained. Recent blind line scans circumvent the spectroscopic redshift requirement, which could represent a selection bias. Our aim is to detect emission lines present in continuum oriented observations. The detection of such lines provides spectroscopic redshift and yields properties of the galaxies. We perform a search for emission lines in the ALMA observations of five Frontier Fields clusters and assess the reliability of our detection by associating line candidates with detected galaxies in deep near-infrared imaging. We find 26 significant emission lines candidates, with observed line fluxes between 0.2-4.6 Jy km s$^{-1}$ and velocity dispersions (FWHM) of 25-600 km s$^{-1}$. Nine of these candidates lie nearby to near-infrared sources, boosting their reliability; in six cases the observed line frequency and strength are consistent with expectations given the photometric redshift and properties of the galaxy counterparts. We present redshift identifications, magnifications and molecular gas estimates for the galaxies with identified lines. We show that two of these candidates likely originate from starburst galaxies, one of which is a jellyfish galaxy, while another two are consistent with being main sequence galaxies based in their depletion times. This work highlights the degree to which serendipitous emission lines can be discovered in large mosaic continuum observations when deep ancillary data are available. The low number of high-significance line detections, however, confirms that such surveys are not as optimal as blind line scans. We stress that Monte Carlo simulations should be used to assess the line detections significances, since using the negative noise suffers from stochasticity and incurs larger uncertainties., Comment: new version accepted for publication in A&A

Dusty star-forming galaxies are among the most prodigious systems at high redshift (z>1), characterized by high star formation rates and huge dust reservoirs. The bright end of this population has been well characterized in recent years, but considerable uncertainties remain for fainter dusty star-forming galaxies, which are responsible for the bulk of star formation at high redshift and thus play a key role in galaxy growth and evolution. In this first paper of our series, we describe our methods for finding high redshift faint dusty galaxies using millimeter observations with ALMA. We obtained ALMA 1.1 mm mosaic images for three strong-lensing galaxy clusters from the Frontier Fields survey. The 2'x2' mosaics overlap with the deep HST WFC3/IR footprints and encompass the high magnification regions of each cluster. The combination of extremely high ALMA sensitivity and the magnification power of these clusters allows us to systematically probe the sub-mJy population of dusty star-forming galaxies over a large surveyed area. We present a description of the reduction and analysis of the ALMA continuum observations for the galaxy clusters Abell 2744 (z=0.308), MACSJ0416.1-2403 (z=0.396) and MACSJ1149.5+2223 (z=0.543), for which we reach observed rms sensitivities of 55, 59 and 71 $\mu$Jy beam$^{-1}$ respectively. We detect 12 dusty star-forming galaxies at S/N>5.0 across the three clusters, all of them presenting coincidence with NIR detected counterparts in the HST images. None of the sources fall close to the lensing caustics, thus they are not strongly lensed. The observed 1.1 mm flux densities for the total sample of galaxies range from 0.41 to 2.82 mJy, with observed effective radii spanning <0.05" to 0.37"$\pm$0.21". The lensing-corrected sizes of the detected sources appear to be in the same range as those measured in brighter samples, albeit with possibly larger dispersion., Comment: Accepted for publication in A&A. New version addressing last referee comments, 17 pages and 12 figures