Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userThe search for galaxy cluster members with deep learning of panchromatic HST imaging and extensive spectroscopy
172
0
0.0
(
0
)
Ask ChatGPT about the research
No Arabic abstract
The next generation of data-intensive surveys are bound to produce a vast amount of data, which can be dealt with using machine-learning methods to explore possible correlations within the multi-dimensional parameter space. We explore the classification capabilities of convolution neural networks (CNNs) to identify galaxy cluster members (CLMs) by using Hubble Space Telescope (HST) images of 15 galaxy clusters at redshift 0.19<z<0.60, observed as part of the CLASH and Hubble Frontier Field programmes. We used extensive spectroscopic information, based on the CLASH-VLT VIMOS programme combined with MUSE observations, to define the knowledge base. We performed various tests to quantify how well CNNs can identify cluster members on the basis of imaging information only. We investigated the CNN capability to predict source memberships outside the training coverage, by identifying CLMs at the faint end of the magnitude distributions. We find that the CNNs achieve a purity-completeness rate ~90%, demonstrating stable behaviour, along with a remarkable generalisation capability with respect to cluster redshifts. We concluded that if extensive spectroscopic information is available as a training base, the proposed approach is a valid alternative to catalogue-based methods because it has the advantage of avoiding photometric measurements, which are particularly challenging and time-consuming in crowded cluster cores. As a byproduct, we identified 372 photometric CLMs, with mag(F814)<25, to complete the sample of 812 spectroscopic CLMs in four galaxy clusters RX~J2248-4431, MACS~J0416-2403, MACS~J1206-0847 and MACS~J1149+2223. When this technique is applied to the data that are expected to become available from forthcoming surveys, it will be an efficient tool for a variety of studies requiring CLM selection, such as galaxy number densities, luminosity functions, and lensing mass reconstruction.
rate research
Read More
We present the data release paper for the Galaxy Zoo: Hubble (GZH) project. This is the third phase in a large effort to measure reliable, detailed morphologies of galaxies by using crowdsourced visual classifications of colour composite images. Images in GZH were selected from various publicly-released Hubble Space Telescope Legacy programs conducted with the Advanced Camera for Surveys, with filters that probe the rest-frame optical emission from galaxies out to $z sim 1$. The bulk of the sample is selected to have $m_{I814W} < 23.5$,but goes as faint as $m_{I814W} < 26.8$ for deep images combined over 5 epochs. The median redshift of the combined samples is $z = 0.9 pm 0.6$, with a tail extending out to $z sim 4$. The GZH morphological data include measurements of both bulge- and disk-dominated galaxies, details on spiral disk structure that relate to the Hubble type, bar identification, and numerous measurements of clump identification and geometry. This paper also describes a new method for calibrating morphologies for galaxies of different luminosities and at different redshifts by using artificially-redshifted galaxy images as a baseline. The GZH catalogue contains both raw and calibrated morphological vote fractions for 119,849 galaxies, providing the largest dataset to date suitable for large-scale studies of galaxy evolution out to $z sim 1$.
We use velocity dispersion measurements of 21 individual cluster members in the core of Abell 383, obtained with MMT Hectospec, to separate the galaxy and the smooth dark halo (DH) lensing contributions. While lensing usually constrains the overall, projected mass density, the innovative use of velocity dispersion measurements as a proxy for masses of individual cluster members breaks inherent degeneracies and allows us to (a) refine the constraints on single galaxy masses and on the galaxy mass-to-light scaling relation and, as a result, (b) refine the constraints on the DM-only map, a high-end goal of lens modelling. The knowledge of cluster member velocity dispersions improves the fit by 17% in terms of the image reproduction $chi^2$, or 20% in terms of the rms. The constraints on the mass parameters improve by ~10% for the DH, while for the galaxy component, they are refined correspondingly by ~50%, including the galaxy halo truncation radius. For an L$^*$ galaxy with M$^*_B$=-20.96, for example, we obtain best fitting truncation radius r$^*_{tr}=20.5^{+9.6}_{-6.7}$ kpc and velocity dispersion $sigma^*=324pm17 km/s$. Moreover, by performing the surface brightness reconstruction of the southern giant arc, we improve the constraints on r$_{tr}$ of two nearby cluster members, which have measured velocity dispersions, by more than ~30%. We estimate the stripped mass for these two galaxies, getting results that are consistent with numerical simulations. In the future, we plan to apply this analysis to other galaxy clusters for which velocity dispersions of member galaxies are available.
We present HST/ACS $g$ and $z$ photometry and half-light radii $R_{rm h}$ measurements of 360 globular cluster (GC) candidates around the nearby S0 galaxy NGC 3115. We also include Subaru/Suprime-Cam $g$, $r$, and $i$ photometry of 421 additional candidates. The well-established color bimodality of the GC system is obvious in the HST/ACS photometry. We find evidence for a blue tilt in the blue GCs, wherein the blue GCs get redder as luminosity increases, indicative of a mass-metallicity relationship. We find a color gradient in both the red and blue subpopulations, with each group of clusters becoming bluer at larger distances from NGC 3115. The gradient is of similar strength in both subpopulations, but is monotonic and more significant for the blue clusters. On average, the blue clusters have ~10% larger $R_{rm h}$ than the red clusters. This average difference is less than is typically observed for early-type galaxies but does match that measured in the literature for M104, suggesting that morphology and inclination may affect the measured size difference between the red and blue clusters. However, the scatter on the $R_{rm h}$ measurements is large. We also identify 31 clusters more extended than typical GCs, which we consider ultra-compact dwarf (UCD) candidates. Many of these objects are fainter than typical UCDs. While it is likely that a significant number will be background contaminants, six of these UCD candidates are spectroscopically confirmed. To explore low-mass X-ray binaries in the GC system, we match our ACS and Suprime-Cam detections to corresponding Chandra X-ray sources. We identify 45 X-ray - GC matches, 16 among the blue subpopulation and 29 among the red subpopulation. These X-ray/GC coincidence fractions are larger than is typical for most GC systems, probably due to the increased depth of the X-ray data compared to previous studies of GC systems.
We present the results of a proof-of-concept experiment which demonstrates that deep learning can successfully be used for production-scale classification of compact star clusters detected in HST UV-optical imaging of nearby spiral galaxies (D < 20 Mpc) in the PHANGS-HST survey. Given the relatively small nature of existing, human-labelled star cluster samples, we transfer the knowledge of state-of-the-art neural network models for real-object recognition to classify star clusters candidates into four morphological classes. We perform a series of experiments to determine the dependence of classification performance on: neural network architecture (ResNet18 and VGG19-BN); training data sets curated by either a single expert or three astronomers; and the size of the images used for training. We find that the overall classification accuracies are not significantly affected by these choices. The networks are used to classify star cluster candidates in the PHANGS-HST galaxy NGC 1559, which was not included in the training samples. The resulting prediction accuracies are 70%, 40%, 40-50%, 50-70% for class 1, 2, 3 star clusters, and class 4 non-clusters respectively. This performance is competitive with consistency achieved in previously published human and automated quantitative classification of star cluster candidate samples (70-80%, 40-50%, 40-50%, and 60-70%). The methods introduced herein lay the foundations to automate classification for star clusters at scale, and exhibit the need to prepare a standardized dataset of human-labelled star cluster classifications, agreed upon by a full range of experts in the field, to further improve the performance of the networks introduced in this study.
We present accurate photometric redshifts for galaxies observed by the Cluster Lensing and Supernova survey with Hubble (CLASH). CLASH observed 25 massive galaxy cluster cores with the Hubble Space Telescope in 16 filters spanning 0.2 - 1.7 $mu$m. Photometry in such crowded fields is challenging. Compared to our previously released catalogs, we make several improvements to the photometry, including smaller apertures, ICL subtraction, PSF matching, and empirically measured uncertainties. We further improve the Bayesian Photometric Redshift (BPZ) estimates by adding a redder elliptical template and by inflating the photometric uncertainties of the brightest galaxies. The resulting photometric redshift accuracies are dz/(1+z) $sim$ 0.8%, 1.0%, and 2.0% for galaxies with I-band F814W AB magnitudes $<$ 18, 20, and 23, respectively. These results are consistent with our expectations. They improve on our previously reported accuracies by a factor of 4 at the bright end and a factor of 2 at the faint end. Our new catalog includes 1257 spectroscopic redshifts, including 382 confirmed cluster members. We also provide stellar mass estimates. Finally, we include lensing magnification estimates of background galaxies based on our public lens models. Our new catalog of all 25 CLASH clusters is available via MAST. The analysis techniques developed here will be useful in other surveys of crowded fields, including the Frontier Fields and surveys carried out with J-PAS and JWST.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
