UWCScholar :: Browsing by Author "Vaccari, Mattia"

Browsing by Author "Vaccari, Mattia"

Now showing 1 - 18 of 18

A Comparison of deep learning architectures for optical galaxy morphology classification
(International Conference on Electrical, Computer and Energy Technologies (ICECET), 2021) Fielding, Ezra; Nyirenda, Clement N.; Vaccari, Mattia
The classification of galaxy morphology plays a crucial role in understanding galaxy formation and evolution. Traditionally, this process is done manually. The emergence of deep learning techniques has given room for the automation of this process. As such, this paper offers a comparison of deep learning architectures to determine which is best suited for optical galaxy morphology classification. Adapting the model training method proposed by Walmsley et al in 2021, the Zoobot Python library is used to train models to predict Galaxy Zoo DECaLS decision tree responses, made by volunteers, using EfficientNet B0, DenseNet121 and ResNet50 as core model architectures. The predicted results are then used to generate accuracy metrics per decision tree question to determine architecture performance. DenseNet121 was found to produce the best results, in terms of accuracy, with a reasonable training time. In future, further testing with more deep learning architectures could prove beneficial.
Cosmos2020: A panchromatic view of the universe to z∼10 from two complementary catalogs
(IOP Publishing, 2022) Weaver, John R.; Kauffmann, Olivier; Vaccari, Mattia
The Cosmic Evolution Survey (COSMOS) has become a cornerstone of extragalactic astronomy. Since the last public catalog in 2015, a wealth of new imaging and spectroscopic data have been collected in the COSMOS field. This paper describes the collection, processing, and analysis of these new imaging data to produce a new reference photometric redshift catalog. Source detection and multiwavelength photometry are performed for 1.7 million sources across the 2 deg2 of the COSMOS field, ∼966,000 of which are measured with all available broadband data using both traditional aperture photometric methods and a new profile-fitting photometric extraction tool, THE FARMER, which we have developed. A detailed comparison of the two resulting photometric catalogs is presented. Photometric redshifts are computed for all sources in each catalog utilizing two independent photometric redshift codes.
Data intensive research and data carpentries at UWC
(University of the Western Cape, 2020) Vaccari, Mattia; Schäfer, Sarah
Galaxy–galaxy lensing in the voice deep survey
(EDP Sciences, 2022) Luo, Ruibiao; Fu, Liping; Vaccari, Mattia
The multi-band photometry of the VOICE imaging data, overlapping with 4.9 deg2 of the Chandra Deep Field South (CDFS) area, enables both shape measurement and photometric redshift estimation to be the two essential quantities for weak lensing analysis. The depth of magAB is up to 26.1 (5σ limiting) in r-band. We estimate the excess surface density (ESD; ∆Σ) based on galaxy–galaxy measurements around galaxies at lower redshift (0.10 < zl < 0.35) while we select the background sources as those at higher redshift ranging from 0.3 to 1.5. The foreground galaxies are divided into two major categories according to their colour (blue and red), each of which has been further divided into high- and low-stellar-mass bins. The halo masses of the samples are then estimated by modelling the signals, and the posterior of the parameters are sampled using a Monte Carlo Markov chain process. We compare our results with the existing stellar-to-halo mass relation (SHMR) and find that the blue low-stellar-mass bin (median M∗ = 108.31 M ) deviates from the SHMR relation whereas the other three samples agree well with empirical curves. We interpret this discrepancy as the effect of the low star-formation efficiency of the low-mass blue dwarf galaxy population dominated in the VOICE-CDFS area.
Hydra I: An extensible multi-source-finder comparison and cataloguing tool
(Cambridge University Press, 2023) Boyce, M. M.; Hopkins, A. M.; Vaccari, Mattia
The latest generation of radio surveys are now producing sky survey images containing many millions of radio sources. In this context it is highly desirable to understand the performance of radio image source finder (SF) software and to identify an approach that optimises source detection capabilities. We have created Hydra to be an extensible multi-SF and cataloguing tool that can be used to compare and evaluate different SFs. Hydra, which currently includes the SFs Aegean, Caesar, ProFound, PyBDSF, and Selavy, provides for the addition of new SFs through containerisation and configuration files. The SF input RMS noise and island parameters are optimised to a 90% “percentage real detections” threshold (calculated from the difference between detections in the real and inverted images), to enable comparison between SFs.
Hydra ii: Characterisation of Aegean, Caesar, profound, pybdsf, and selavy source finders
(Cambridge University Press, 2023) Boyce, M. M.; Hopkins, A. M.; Vaccari, Mattia
We present a comparison between the performance of a selection of source finders using a new software tool called Hydra. The companion paper, Paper I, introduced the Hydra tool and demonstrated its performance using simulated data. Here we apply Hydra to assess the performance of different source finders by analysing real observational data taken from the Evolutionary Map of the Universe (EMU) Pilot Survey. EMU is a wide-field radio continuum survey whose primary goal is to make a deep (20μJy/beam RMS noise), intermediate angular resolution (15′′), 1 GHz survey of the entire sky south of +30◦ declination, and expecting to detect and catalogue up to 40 million sources. With the main EMU survey expected to begin in 2022 it is highly desirable to understand the performance of radio image source finder software and to identify an approach that optimises source detection capabilities. Hydra has been developed to refine this process, as well as to deliver a range of metrics and source finding data products from multiple source finders. We present the performance of the five source finders tested here in terms of their completeness and reliability statistics, their flux density and source size measurements, and an exploration of case studies to highlight finder-specific limitations.
Lenses In VoicE (LIVE): Searching for strong gravitational lenses in the VOICE@VST survey using Convolutional Neural Networks
(Monthly notices of the Royal Astronomical Society, 2021) Vaccari, Mattia
We present a sample of 16 likely strong gravitational lenses identified in the VST Optical Imaging of the CDFS and ES1 fields (VOICE survey) using Convolutional Neural Networks (CNNs). We train two different CNNs on composite images produced by superimposing simulated gravitational arcs on real Luminous Red Galaxies observed in VOICE. Specifically, the first CNN is trained on single-band images and more easily identifies systems with large Einstein radii, while the second one, trained on composite RGB images, is more accurate in retrieving systems with smaller Einstein radii. We apply both networks to real data from the VOICE survey, taking advantage of the high limiting magnitude (26.1 in the r-band) and low PSF FWHM (0.8" in the r-band) of this deep survey. We analyse ∼ 21, 200 images with 𝑚𝑎𝑔𝑟 < 21.5, identifying 257 lens candidates. To retrieve a high-confidence sample and to assess the accuracy of our technique, nine of the authors perform a visual inspection. Roughly 75% of the systems are classified as likely lenses by at least one of the authors. Finally, we assemble the LIVE sample (Lenses In VoicE) composed by the 16 systems passing the chosen grading threshold. Three of these candidates show likely lensing features when observed by the Hubble Space Telescope. This work represents a further confirmation of the ability of CNNs to inspect large samples of galaxies searching for gravitational lenses. These algorithms will be crucial to exploit the full scientific potential of forthcoming surveys with the Euclid satellite and the Vera Rubin Observatory.
Machine learning approaches to study star formation and black hole accretion in the Meerkat/MIGHTEE survey
(University of the Western Cape, 2023) Silima, Walter; Vaccari, Mattia
Galaxy formation and evolution are driven by two main physical processes: star formation and black hole accretion. Both processes can be traced via the synchrotron emission at radio wavelengths. However, a reliable classification of radio sources as star-formation-dominated sources (or Star-Forming Galaxies, SFGs) and blackhole- accretion-dominated sources (or Active Galactic Nuclei, AGN) is non-trivial and often requires extensive use of multi-wavelength data. Although significant effort has been put into classifying radio sources as SFGs or AGN over the decades, the rapid growth of radio data available from facilities such as the South African MeerKAT telescope, the Australian Square Kilometre Array Pathfinder (ASKAP), and eventually the Square Kilometre Array (SKA) requires the development of efficient and reliable automated classification techniques.
Mightee-hi: Evolution of hi scaling relations of star-forming galaxies at z < 0.5*
(IOP Publishing, 2022) Sinigaglia, Francesco; Rodighiero, Giulia; Vaccari, Mattia
We present the first measurements of H I galaxy scaling relations from a blind survey at z > 0.15. We perform spectral stacking of 9023 spectra of star-forming galaxies undetected in H I at 0.23 < z < 0.49, extracted from MIGHTEE-H I Early Science data cubes, acquired with the MeerKAT radio telescope. We stack galaxies in bins of galaxy properties (stellar mass M*, star formation rateSFR, and specific star formation rate sSFR, with sSFR ≡ M*/SFR), obtaining 5σ detections in most cases, the strongest H I-stacking detections to date in this redshift range. With these detections, we are able to measure scaling relations in the probed redshift interval, finding evidence for a moderate evolution from the median redshift of our sample zmed ∼ 0.37 to z ∼ 0. In particular, low-M* galaxies (  ~ * log 9 10( ) M M ) experience a strong H I depletion (∼0.5 dex in log10( ) M M H I  ), while massive galaxies (  ~ * log 11 10( ) M M ) keep their H I mass nearly unchanged. When looking at the star formation activity, highly star-forming galaxies evolve significantly in MH I ( fH I, where fH I ≡ MH I/M*) at fixed SFR (sSFR), while at the lowest probed SFR (sSFR) the scaling relations show no evolution.
The nature of the microjy source population
(University of the Western Cape, 2015) Ocran Emmanuel Francis; Taylor, Russ; Vaccari, Mattia
The study of the faint radio universe and of its properties has recently become a very active field of research not only because of the much improved capabilities of the SKA pathfinders but also because of the need to better plan for SKA surveys. These new facilities will map large areas of the sky to unprecedented depths and transform radio astronomy into the leading technique for investigating the complex processes which govern the formation and evolution of galaxies. This thesis combines multi-wavelength techniques, highly relevant to future deep radio surveys, to study the properties of faint radio sources. The nature of the faint radio sources is presented, over a large GMRT survey area of an area of 1.2 deg2 comprising 2800 sources. Utilising multi-wavelength data we have matched 85% of the radio population to Spitzer/IRAC and obtained a redshift estimate for 63%. The redshift associations are a combination of photometric and spectroscopic redshift estimates. This study investigates several multi-wavelength diagnostics used to identify AGN, using radio, infrared, optical and x-ray data . This analysis shows that various diagnostics (from the radio through the X-ray ones) select fairly different types of AGNs, with an evidence of a disagreement in the number of AGNs selected by each individual diagnostics. For the sources with redshift we use a classification scheme based on radio luminosity, x-ray emission, BOSS/SDSS spectroscopy, IRAC colors satisfying the Donley criterion, and MIPS 24ɥm radio-loud AGN criteria to separate them into AGNs and SFGs. On the basis of this classification, we find that at least 12.5% of the sources with redshifts are AGNs while the remaining 87.5% are adopted as SFGs. We explore the nature of the classified sources through the far-infrared radio correlation. We measure a median qIR value of 2:45± 0:01 for the SFGs and qIR value of 2:27 ± 0:05 for the AGNs. The decrease in the median value of qIR for the AGNs is a result of the additional AGN component to radio emission for the AGN-powered sources and find tentative evidence of an evolution of the qIR with redshift.
An overview of the Dwarf Galaxy Survey
(The Astronomical Society of the Pacific, 2013) Remy-Ruyer, A.; Galametz, M.; Vaccari, Mattia; Madden, S.C.
The Dwarf Galaxy Survey (DGS) program is studying low-metallicity galax- ies using 230h of far-infrared (FIR) and submillimetre (submm) photometric and spectroscopic observations of the Herschel Space Observatory and draws to this a rich database of a wide range of wavelengths tracing the dust, gas and stars. This sample of 50 galaxies includes the largest metallicity range achievable in the local Universe including the lowest metallicity (Z) galaxies, 1/50 Z⊙, and spans 4 orders of magnitude in star formation rates. The survey is designed to get a handle on the physics of the interstellar medium (ISM) of low metallicity dwarf galaxies, especially on their dust and gas properties and the ISM heating and cooling processes. The DGS produces PACS and SPIRE maps of low-metallicity galaxies observed at 70, 100, 160, 250, 350, and 500 μm with the highest sensi- tivity achievable to date in the FIR and submm. The FIR fine-structure lines, [CII] 158μm, [OI] 63μm, [OI] 145μm, [OIII] 88μm, [NIII] 57μm and [NII] 122 and 205 μm have also been observed with the aim of studying the gas cooling in the neutral and ionized phases. The SPIRE FTS observations include many CO lines (J=4-3 to J=13-12), [NII] 205 μm and [CI] lines at 370 and 609 μm. This paper describes the sample selection and global properties of the galaxies, the observing strategy as well as the vast ancillary database available to comple- ment the Herschel observations. The scientific potential of the full DGS survey is described with some example results included.
Revised SWIRE photometric redshifts
(Oxford University Press, 2013) Rowan-Robinson, Michael; Gonzalez-Solares, Eduardo A.; Vaccari, Mattia; Marchetti, L.
We have revised the Spitzer Wide-Area Infrared Extragalactic survey (SWIRE) Photometric Redshift Catalogue to take account of new optical photometry in several of the SWIRE areas, and incorporating Two Micron All Sky Survey (2MASS) and UKIRT Infrared Deep Sky Survey (UKIDSS) near-infrared data. Aperture matching is an important issue for combining near-infrared and optical data, and we have explored a number of methods of doing this. The increased number of photometric bands available for the redshift solution results in improvements both in the rms error and, especially, in the outlier rate. We have also found that incorporating the dust torus emission into the quasi-stellar object (QSO) templates improves the performance for QSO redshift estimation. Our revised redshift catalogue contains over 1 million extragalactic objects, of which 26 288 are QSOs.
The role of environment in galaxy evolution in the servs survey. I. density maps and cluster candidates
(The American Astronomical Society, 2020) Vaccari, Mattia; Krefting, Nick; Sajina, Anna
We use photometric redshifts derived from new u-band through 4.5 μm Spitzer IRAC photometry in the 4.8 deg2 of the XMM-LSS field to construct surface density maps in the redshift range of 0.1–1.5. Our density maps show evidence for large-scale structure in the form of filaments spanning several tens of megaparsecs. Using these maps, we identify 339 overdensities that our simulated light-cone analysis suggests are likely associated with dark matter halos with masses, Mhalo, log(Mhalo/Me) > 13.7. From this list of overdensities we recover 43 of 70 known X-raydetected and spectroscopically confirmed clusters. The missing X-ray clusters are largely at lower redshifts and lower masses than our target log(Mhalo/Me) > 13.7. The bulk of the overdensities are compact, but a quarter show extended morphologies that include likely projection effects, clusters embedded in apparent filaments, and at least one potential cluster merger (at z ∼ 1.28). The strongest overdensity in our highest-redshift slice (at z ∼ 1.5) shows a compact red galaxy core, potentially implying a massive evolved cluster.
The roles of star formation and AGN activity of IRS sources in the HerMES fields
(Oxford University Press, 2013) Feltre, Anna; Hatziminaoglou, Evanthia; Vaccari, Mattia; Hernán-Caballero, Antonio; Fritz, Jacopo; Franceschini, Alberto; Bock, J.; Cooray, Asantha; Farrah, Duncan; Gonzalez-Solares, Eduardo A.; Ibar, Edo; Isaak, Kate G.; Lo Faro, Barbara; Marchetti, L.; Oliver, Seb J.; Page, Matthew J.; Rigopoulou, Dimitra; Roseboom, Isaac G.; Symeonidis, Myrto
In this work we explore the impact of the presence of an active galactic nucleus (AGN) on the mid- and far-infrared (IR) properties of galaxies as well as the effects of simultaneous AGN and starburst activity in these same galaxies. To do this we apply a multi-component, multi-band spectral synthesis technique to a sample of 250 μm selected galaxies of the HerschelMulti- tiered Extragalactic Survey (HerMES), with IRS spectra available for all galaxies. Our results confirm that the inclusion of the IRS spectra plays a crucial role in the spectral analysis of galaxies with an AGN component improving the selection of the best-fit hot dust (torus) model. We find a correlation between the obscured star formation rate (SFR) derived from the IR luminosity of the starburst component, SFRIR and SFRPAH, derived from the luminosity of the PAH features, LPAH, with SFRFIR taking higher values than SFRPAH. The correlation is different for AGN- and starburst-dominated objects. The ratio of LPAH to that of the starburst component, LPAH/LSB, is almost constant for AGN-dominated objects but decreases with increasing LSB for starburst-dominated objects. SFRFIR increases with the accretion luminosity, Lacc, with the increase less prominent for the very brightest, unobscured AGN-dominated sources. We find no correlation between the masses of the hot (AGN-heated) and cold (starburstheated) dust components. We interpret this as a non-constant fraction of gas driven by the gravitational effects to the AGN while the starburst is ongoing. We also find no evidence of the AGN affecting the temperature of the cold dust component, though this conclusion is mostly based on objects with a non-dominantAGN component.We conclude that our findings do not provide evidence that the presence of AGN affects the star formation process in the host galaxy, but rather that the two phenomena occur simultaneously over a wide range of luminosities.
A structure function analysis of vst-cosmos agn
(EDP Sciences, 2022) de Cicco, Demetra; Bauer, Franz E.; Vaccari, Mattia
We present our sixth work in a series dedicated to variability studies of active galactic nuclei (AGN), based on the survey of the COSMOS field by the VLT Survey Telescope (VST). Its 54 r-band visits over 3.3 yr and single-visit depth of 24.6 r-band mag make this dataset a valuable scaled-down version that can help forecast the performance of the Rubin Observatory Legacy Survey of Space and Time (LSST).This work is centered on the analysis of the structure function (SF) of VST-COSMOS AGN, investigating possible differences in its shape and slope related to how the AGN were selected, and explores possible connections between the AGN ensemble variability and the black-hole mass, accretion rate, bolometric luminosity, redshift, and obscuration of the source.
Supervised machine learning techniques for radio source classification
(University of the Western Cape, 2023) Mofokeng, Chaka; Vaccari, Mattia
Classification is one of the most fundamental aspects of scientific investigation. Astronomers have thus developed several classification schemes to try and make sense of the evolving properties of planets, stars and galaxies. One of the most popular ways to classify galaxies is according to their shape, or morphology, which has long been performed visually to produce annotated galaxy catalogues. However, visual inspection and manual annotation by astronomers will not be able to keep up with the expected data flow from next-generation sky surveys. In this context, the main objective of our study was to use deep learning to automate radio source characterization (that is detection, classification and identification) from image data efficiently. We adopted a pre-trained deep learning model called CLARAN (Classifying Radio Sources Automatically with Neural Networks) based on the Radio Galaxy Zoo Citizen Science Classification Project and applied it to a GMRT 610 MHz survey in the ELAIS-N1 region covering an area of 12.8 square degrees at a resolution of approximately 6 arcsec at a rootmean-square noise of about 40 µJy/beam.
Variability and transient search in the SUDARE–VOICE field: A new method to extract the light curves
(Oxford University Press, 2020) Vaccari, Mattia; Liu, Dezi; Deng, Wenqiang
The VLT Survey Telescope (VST) Optical Imaging of the CDFS and ES1 Fields Survey, in synergy with the SUDARE survey, is a deep optical ugri imaging of the CDFS and ES1 fields using the VST. The observations for the CDFS field comprise about 4.38 deg2 down to r ∼ 26 mag. The total on-sky time spans over 4 yr in this field, distributed over four adjacent subfields. In this paper, we use the multiepoch r-band imaging data to measure the variability of the detected objects and search for transients. We perform careful astrometric and photometric calibrations and point spread function modelling. A new method, referring to as differential running-average photometry, is proposed to measure the light curves of the detected objects. With the method, the difference of PSFs between different epochs can be reduced, and the background fluctuations are also suppressed.
The VLA-COSMOS 3 GHz Large Project: Evolution of specific star formation rates out to z∼5
(IOP Publishing, 2020) Vaccari, Mattia; Leslie, Sarah K.; Schinnerer, Eva
We provide a coherent, uniform measurement of the evolution of the logarithmic star formation rate (SFR) – stellar mass (M∗) relation, called the main sequence of star-forming galaxies (MS), for starforming and all galaxies out to z ∼ 5. We measure the MS using mean stacks of 3 GHz radio continuum images to derive average SFRs for ∼ 200,000 mass-selected galaxies at z > 0.3 in the COSMOS field. We describe the MS relation adopting a new model that incorporates a linear relation at low stellar mass (log(M∗/M )<10) and a flattening at high stellar mass that becomes more prominent at low redshift (z < 1.5). We find that the SFR density peaks at 1.5 < z < 2 and at each epoch there is a characteristic stellar mass (M∗ = 1−4×1010M ) that contributes the most to the overall SFR density. This characteristic mass increases with redshift, at least to z ∼ 2.5. We find no significant evidence for variations in the MS relation for galaxies in different environments traced by the galaxy number density at 0.3 < z < 3, nor for galaxies in X-ray groups at z ∼ 0.75. We confirm that massive bulgedominated galaxies have lower SFRs than disk-dominated galaxies at a fixed stellar mass at z < 1.2. As a consequence, the increase in bulge-dominated galaxies in the local star-forming population leads to a flattening of the MS at high stellar masses. This indicates that “mass-quenching” is linked with changes in the morphological composition of galaxies at a fixed stellar mass.

Browsing by Author "Vaccari, Mattia"

Results Per Page

Sort Options