-
Photometry of high-redshift blended galaxies using deep learning
Authors:
Alexandre Boucaud,
Marc Huertas-Company,
Caroline Heneka,
Emille E. O. Ishida,
Nima Sedaghat,
Rafael S. de Souza,
Ben Moews,
Hervé Dole,
Marco Castellano,
Emiliano Merlin,
Valerio Roscani,
Andrea Tramacere,
Madhura Killedar,
Arlindo M. M. Trindade
Abstract:
The new generation of deep photometric surveys requires unprecedentedly precise shape and photometry measurements of billions of galaxies to achieve their main science goals. At such depths, one major limiting factor is the blending of galaxies due to line-of-sight projection, with an expected fraction of blended galaxies of up to 50%. Current deblending approaches are in most cases either too slo…
▽ More
The new generation of deep photometric surveys requires unprecedentedly precise shape and photometry measurements of billions of galaxies to achieve their main science goals. At such depths, one major limiting factor is the blending of galaxies due to line-of-sight projection, with an expected fraction of blended galaxies of up to 50%. Current deblending approaches are in most cases either too slow or not accurate enough to reach the level of requirements. This work explores the use of deep neural networks to estimate the photometry of blended pairs of galaxies in monochrome space images, similar to the ones that will be delivered by the Euclid space telescope. Using a clean sample of isolated galaxies from the CANDELS survey, we artificially blend them and train two different network models to recover the photometry of the two galaxies. We show that our approach can recover the original photometry of the galaxies before being blended with $\sim$7% accuracy without any human intervention and without any assumption on the galaxy shape. This represents an improvement of at least a factor of 4 compared to the classical SExtractor approach. We also show that forcing the network to simultaneously estimate a binary segmentation map results in a slightly improved photometry. All data products and codes will be made public to ease the comparison with other approaches on a common data set.
△ Less
Submitted 3 May, 2019;
originally announced May 2019.
-
A probabilistic approach to emission-line galaxy classification
Authors:
R. S. de Souza,
M. L. L. Dantas,
M. V. Costa-Duarte,
E. D. Feigelson,
M. Killedar,
P. -Y. Lablanche,
R. Vilalta,
A. Krone-Martins,
R. Beck,
F. Gieseke
Abstract:
We invoke a Gaussian mixture model (GMM) to jointly analyse two traditional emission-line classification schemes of galaxy ionization sources: the Baldwin-Phillips-Terlevich (BPT) and $\rm W_{Hα}$ vs. [NII]/H$α$ (WHAN) diagrams, using spectroscopic data from the Sloan Digital Sky Survey Data Release 7 and SEAGal/STARLIGHT datasets. We apply a GMM to empirically define classes of galaxies in a thre…
▽ More
We invoke a Gaussian mixture model (GMM) to jointly analyse two traditional emission-line classification schemes of galaxy ionization sources: the Baldwin-Phillips-Terlevich (BPT) and $\rm W_{Hα}$ vs. [NII]/H$α$ (WHAN) diagrams, using spectroscopic data from the Sloan Digital Sky Survey Data Release 7 and SEAGal/STARLIGHT datasets. We apply a GMM to empirically define classes of galaxies in a three-dimensional space spanned by the $\log$ [OIII]/H$β$, $\log$ [NII]/H$α$, and $\log$ EW(H$α$), optical parameters. The best-fit GMM based on several statistical criteria suggests a solution around four Gaussian components (GCs), which are capable to explain up to 97 per cent of the data variance. Using elements of information theory, we compare each GC to their respective astronomical counterpart. GC1 and GC4 are associated with star-forming galaxies, suggesting the need to define a new starburst subgroup. GC2 is associated with BPT's Active Galaxy Nuclei (AGN) class and WHAN's weak AGN class. GC3 is associated with BPT's composite class and WHAN's strong AGN class. Conversely, there is no statistical evidence -- based on four GCs -- for the existence of a Seyfert/LINER dichotomy in our sample. Notwithstanding, the inclusion of an additional GC5 unravels it. The GC5 appears associated to the LINER and Passive galaxies on the BPT and WHAN diagrams respectively. Subtleties aside, we demonstrate the potential of our methodology to recover/unravel different objects inside the wilderness of astronomical datasets, without lacking the ability to convey physically interpretable results. The probabilistic classifications from the GMM analysis are publicly available within the COINtoolbox (https://cointoolbox.github.io/GMM\_Catalogue/).
△ Less
Submitted 18 August, 2017; v1 submitted 22 March, 2017;
originally announced March 2017.
-
Simulation-based marginal likelihood for cluster strong lensing cosmology
Authors:
Madhura Killedar,
Stefano Borgani,
Dunja Fabjan,
Klaus Dolag,
Gian Luigi Granato,
Massimo Meneghetti,
Susana Planelles,
Cinthia Ragone-Figueroa
Abstract:
Comparisons between observed and predicted strong lensing properties of galaxy clusters have been routinely used to claim either tension or consistency with $Λ$CDM cosmology. However, standard approaches to such cosmological tests are unable to quantify the preference for one cosmology over another. We advocate approximating the relevant Bayes factor using a marginal likelihood that is based on th…
▽ More
Comparisons between observed and predicted strong lensing properties of galaxy clusters have been routinely used to claim either tension or consistency with $Λ$CDM cosmology. However, standard approaches to such cosmological tests are unable to quantify the preference for one cosmology over another. We advocate approximating the relevant Bayes factor using a marginal likelihood that is based on the following summary statistic: the posterior probability distribution function for the parameters of the scaling relation between Einstein radii and cluster mass, $α$ and $β$. We demonstrate, for the first time, a method of estimating the marginal likelihood using the X-ray selected $z>0.5$ MACS clusters as a case in point and employing both N-body and hydrodynamic simulations of clusters. We investigate the uncertainty in this estimate and consequential ability to compare competing cosmologies, that arises from incomplete descriptions of baryonic processes, discrepancies in cluster selection criteria, redshift distribution, and dynamical state. The relation between triaxial cluster masses at various overdensities provide a promising alternative to the strong lensing test.
△ Less
Submitted 1 September, 2017; v1 submitted 20 July, 2015;
originally announced July 2015.
-
The Overlooked Potential of Generalized Linear Models in Astronomy-III: Bayesian Negative Binomial Regression and Globular Cluster Populations
Authors:
R. S. de Souza,
J. M. Hilbe,
B. Buelens,
J. D. Riggs,
E. Cameron,
E. E. O. Ishida,
A. L. Chies-Santos,
M. Killedar
Abstract:
In this paper, the third in a series illustrating the power of generalized linear models (GLMs) for the astronomical community, we elucidate the potential of the class of GLMs which handles count data. The size of a galaxy's globular cluster population $N_{\rm GC}$ is a prolonged puzzle in the astronomical literature. It falls in the category of count data analysis, yet it is usually modelled as i…
▽ More
In this paper, the third in a series illustrating the power of generalized linear models (GLMs) for the astronomical community, we elucidate the potential of the class of GLMs which handles count data. The size of a galaxy's globular cluster population $N_{\rm GC}$ is a prolonged puzzle in the astronomical literature. It falls in the category of count data analysis, yet it is usually modelled as if it were a continuous response variable. We have developed a Bayesian negative binomial regression model to study the connection between $N_{\rm GC}$ and the following galaxy properties: central black hole mass, dynamical bulge mass, bulge velocity dispersion, and absolute visual magnitude. The methodology introduced herein naturally accounts for heteroscedasticity, intrinsic scatter, errors in measurements in both axes (either discrete or continuous), and allows modelling the population of globular clusters on their natural scale as a non-negative integer variable. Prediction intervals of 99% around the trend for expected $N_{\rm GC}$comfortably envelope the data, notably including the Milky Way, which has hitherto been considered a problematic outlier. Finally, we demonstrate how random intercept models can incorporate information of each particular galaxy morphological type. Bayesian variable selection methodology allows for automatically identifying galaxy types with different productions of GCs, suggesting that on average S0 galaxies have a GC population 35% smaller than other types with similar brightness.
△ Less
Submitted 13 August, 2015; v1 submitted 15 June, 2015;
originally announced June 2015.
-
The Overlooked Potential of Generalized Linear Models in Astronomy - I: Binomial Regression
Authors:
R. S. de Souza,
E. Cameron,
M. Killedar,
J. Hilbe,
R. Vilalta,
U. Maio,
V. Biffi,
B. Ciardi,
J. D. Riggs
Abstract:
Revealing hidden patterns in astronomical data is often the path to fundamental scientific breakthroughs; meanwhile the complexity of scientific inquiry increases as more subtle relationships are sought. Contemporary data analysis problems often elude the capabilities of classical statistical techniques, suggesting the use of cutting edge statistical methods. In this light, astronomers have overlo…
▽ More
Revealing hidden patterns in astronomical data is often the path to fundamental scientific breakthroughs; meanwhile the complexity of scientific inquiry increases as more subtle relationships are sought. Contemporary data analysis problems often elude the capabilities of classical statistical techniques, suggesting the use of cutting edge statistical methods. In this light, astronomers have overlooked a whole family of statistical techniques for exploratory data analysis and robust regression, the so-called Generalized Linear Models (GLMs). In this paper -- the first in a series aimed at illustrating the power of these methods in astronomical applications -- we elucidate the potential of a particular class of GLMs for handling binary/binomial data, the so-called logit and probit regression techniques, from both a maximum likelihood and a Bayesian perspective. As a case in point, we present the use of these GLMs to explore the conditions of star formation activity and metal enrichment in primordial minihaloes from cosmological hydro-simulations including detailed chemistry, gas physics, and stellar feedback. We predict that for a dark mini-halo with metallicity $\approx 1.3 \times 10^{-4} Z_{\bigodot}$, an increase of $1.2 \times 10^{-2}$ in the gas molecular fraction, increases the probability of star formation occurrence by a factor of 75%. Finally, we highlight the use of receiver operating characteristic curves as a diagnostic for binary classifiers, and ultimately we use these to demonstrate the competitive predictive performance of GLMs against the popular technique of artificial neural networks.
△ Less
Submitted 4 April, 2015; v1 submitted 26 September, 2014;
originally announced September 2014.
-
Gravitational lensing in WDM cosmologies: The cross section for giant arcs
Authors:
Hareth Mahdi,
Martijn van Beek,
Pascal Elahi,
Geraint Lewis,
Chris Power,
Madhura Killedar
Abstract:
The nature of the dark sector of the Universe remains one of the outstanding problems in modern cosmology, with the search for new observational probes guiding the development of the next generation of observational facilities. Clues come from tension between the predictions from ΛCDM and observations of gravitationally lensed galaxies. Previous studies showed that galaxy clusters in the ΛCDM are…
▽ More
The nature of the dark sector of the Universe remains one of the outstanding problems in modern cosmology, with the search for new observational probes guiding the development of the next generation of observational facilities. Clues come from tension between the predictions from ΛCDM and observations of gravitationally lensed galaxies. Previous studies showed that galaxy clusters in the ΛCDM are not strong enough to reproduce the observed number of lensed arcs. This work aims to constrain the warm dark matter cosmologies by means of the lensing efficiency of galaxy clusters drawn from these alternative models. The lensing characteristics of two samples of simulated clusters in the warm dark matter (ΛWDM) and cold dark matter (ΛCDM) cosmologies have been studied. The results show that even though the CDM clusters are more centrally concentrated and contain more substructures, the WDM clusters have slightly higher lensing efficiency than their CDM counterparts. The key difference is that WDM clusters have more extended and more massive subhaloes than CDM analogues. These massive substructures significantly stretch the critical lines and caustics and hence they boost the lensing efficiency of the host halo. Despite the increase in the lensing efficiency due to the contribution of massive substructures in the WDM clusters, this is not enough to resolve the arc statistics problem.
△ Less
Submitted 6 April, 2014;
originally announced April 2014.
-
On the role of AGN feedback on the thermal and chemodynamical properties of the hot intra-cluster medium
Authors:
S. Planelles,
S. Borgani,
D. Fabjan,
M. Killedar,
G. Murante,
G. L. Granato,
C. Ragone-Figueroa,
K. Dolag
Abstract:
We present an analysis of the properties of the ICM in an extended set of cosmological hydrodynamical simulations of galaxy clusters and groups performed with the TreePM+SPH GADGET-3 code. Besides a set of non-radiative simulations, we carried out two sets of simulations including radiative cooling, star formation, metal enrichment and feedback from supernovae, one of which also accounts for the e…
▽ More
We present an analysis of the properties of the ICM in an extended set of cosmological hydrodynamical simulations of galaxy clusters and groups performed with the TreePM+SPH GADGET-3 code. Besides a set of non-radiative simulations, we carried out two sets of simulations including radiative cooling, star formation, metal enrichment and feedback from supernovae, one of which also accounts for the effect of feedback from AGN resulting from gas accretion onto super-massive black holes. These simulations are analysed with the aim of studying the relative role played by SN and AGN feedback on the general properties of the diffuse hot baryons in galaxy clusters and groups: scaling relations, temperature, entropy and pressure radial profiles, and ICM chemical enrichment. We find that simulations including AGN feedback produce scaling relations that are in good agreement with X-ray observations at all mass scales. However, our simulations are not able to account for the observed diversity between CC and NCC clusters: unlike for observations, we find that temperature and entropy profiles of relaxed and unrelaxed clusters are quite similar and resemble more the observed behaviour of NCC clusters. As for the pattern of metal enrichment, we find that an enhanced level of iron abundance is produced by AGN feedback with respect to the case of purely SN feedback. As a result, while simulations including AGN produce values of iron abundance in groups in agreement with observations, they over-enrich the ICM in massive clusters. The efficiency of AGN feedback in displacing enriched gas from halos into the inter-galactic medium at high redshift also creates a widespread enrichment in the outskirts of clusters and produces profiles of iron abundance whose slope is in better agreement with observations.
△ Less
Submitted 4 November, 2013;
originally announced November 2013.
-
Characterizing Diffused Stellar Light in simulated galaxy clusters
Authors:
Weiguang Cui,
G. Murante,
P. Monaco,
S. Borgani,
G. L. Granato,
M. Killedar,
G. De Lucia,
V. Presotto,
K. Dolag
Abstract:
[Abridged] In this paper, we carry out a detailed analysis of the performance of two different methods to identify the diffuse stellar light in cosmological hydrodynamical simulations of galaxy clusters. One method is based on a dynamical analysis of the stellar component. The second method is closer to techniques commonly employed in observational studies. Both the dynamical method and the method…
▽ More
[Abridged] In this paper, we carry out a detailed analysis of the performance of two different methods to identify the diffuse stellar light in cosmological hydrodynamical simulations of galaxy clusters. One method is based on a dynamical analysis of the stellar component. The second method is closer to techniques commonly employed in observational studies. Both the dynamical method and the method based on the surface brightness limit criterion are applied to the same set of hydrodynamical simulations for a large sample about 80 galaxy clusters.
We find significant differences between the ICL and DSC fractions computed with the two corresponding methods, which amounts to about a factor of two for the AGN simulations, and a factor of four for the CSF set. We also find that the inclusion of AGN feedback boosts the DSC and ICL fractions by a factor of 1.5-2, respectively, while leaving the BCG+ICL and BCG+DSC mass fraction almost unchanged. The sum of the BCG and DSC mass stellar mass fraction is found to decrease from ~80 per cent in galaxy groups to ~60 per cent in rich clusters, thus in excess of what found from observational analysis.
We identify the average surface brightness limits that yields the ICL fraction from the SBL method close to the DSC fraction from the dynamical method. These surface brightness limits turn out to be brighter in the CSF than in the AGN simulations. This is consistent with the finding that AGN feedback makes BCGs to be less massive and with shallower density profiles than in the CSF simulations. The BCG stellar component, as identified by both methods, are slightly older and more metal-rich than the stars in the diffuse component.
△ Less
Submitted 11 November, 2013; v1 submitted 4 October, 2013;
originally announced October 2013.
-
Galactic Winds in Cosmological Simulations of the Circumgalactic Medium
Authors:
Paramita Barai,
Matteo Viel,
Stefano Borgani,
Edoardo Tescari,
Luca Tornatore,
Klaus Dolag,
Madhura Killedar,
Pierluigi Monaco,
Valentina D'Odorico,
Stefano Cristiani
Abstract:
(Abridged) We explore new observationally-constrained sub-resolution models of galactic outflows and investigate their impact on the circumgalactic medium (CGM) over redshifts z = 2 - 4. We perform cosmological hydrodynamic simulations, including star formation, chemical enrichment, and four cases of SNe-driven outflows: no wind (NW), an energy-driven constant velocity wind (CW), a radially varyin…
▽ More
(Abridged) We explore new observationally-constrained sub-resolution models of galactic outflows and investigate their impact on the circumgalactic medium (CGM) over redshifts z = 2 - 4. We perform cosmological hydrodynamic simulations, including star formation, chemical enrichment, and four cases of SNe-driven outflows: no wind (NW), an energy-driven constant velocity wind (CW), a radially varying wind (RVWa) where the outflow velocity has a positive correlation with galactocentric distance (r), and a RVW with additional dependence on halo mass (RVWb). Overall, we find that the outflows expel metal-enriched gas away from galaxies, significantly quench star formation, and enrich the CGM. At z = 2, the radial profiles of gas properties around galaxy centers are most sensitive to the choice of the wind model for halo masses (10^9 - 10^11) M_sun. We infer that the RVWb model is similar to the NW case, except that it substantially enriches the CGM: the carbon metallicity (Z_C) is 10 times higher in RVWb than in NW at r > R_200; and the warm gas of 10^4 - 10^5 K and delta < 100 is enriched to 50 times higher than in NW. We also find that the impact of models CW and RVWa are similar, with the following differences. RVWa causes a greater suppression of star formation rate at z < 5, and has a higher fraction of low-density (delta < 10), warm-hot (10^4 - 10^6 K) gas than in CW. Outflows in CW produce a higher and earlier enrichment of some IGM phases than RVWa. By visual inspection, we note that the RVWa model shows more pronounced bipolar outflows and galactic disks. We present fitting formulae for [Z_C-delta] and [Z_C-r], also for the abundance of CIV as a function of r. We predict observational diagnostics to distinguish between different outflow scenarios: Z_C of the CGM gas at r = (30 - 300) kpc/h comoving, and CIV fraction of the inner gas at r < (4 - 5) kpc/h comoving.
△ Less
Submitted 21 January, 2013; v1 submitted 12 October, 2012;
originally announced October 2012.
-
How Baryonic Processes affect Strong Lensing properties of Simulated Galaxy Clusters
Authors:
Madhura Killedar,
Stefano Borgani,
Massimo Meneghetti,
Klaus Dolag,
Dunja Fabjan,
Luca Tornatore
Abstract:
The observed abundance of giant arcs produced by galaxy cluster lenses and the measured Einstein radii have presented a source of tension for LCDM. Previous cosmological tests for high-redshift clusters (z>0.5) have suffered from small number statistics in the simulated sample and the implementation of baryonic physics is likely to affect the outcome. We analyse zoomed-in simulations of a fairly l…
▽ More
The observed abundance of giant arcs produced by galaxy cluster lenses and the measured Einstein radii have presented a source of tension for LCDM. Previous cosmological tests for high-redshift clusters (z>0.5) have suffered from small number statistics in the simulated sample and the implementation of baryonic physics is likely to affect the outcome. We analyse zoomed-in simulations of a fairly large sample of cluster-sized objects, with Mvir > 3x10^14 Msun/h, identified at z=0.25 and z=0.5, for a concordance LCDM cosmology. We start with dark matter only simulations, and then add gas hydrodynamics, with different treatments of baryonic processes: non-radiative cooling, radiative cooling with star formation and galactic winds powered by supernova explosions, and finally including the effect of AGN feedback. We find that the addition of gas in non-radiative simulations does not change the strong lensing predictions significantly, but gas cooling and star formation together significantly increase the number of expected giant arcs and the Einstein radii, particularly for lower redshift clusters and lower source redshifts. Further inclusion of AGN feedback reduces the predicted strong lensing efficiencies such that the lensing probability distributions becomes closer to those obtained for simulations including only dark matter. Our results indicate that the inclusion of baryonic physics in simulations will not solve the arc-statistics problem at low redshifts, when the physical processes included provide a realistic description of cooling in the central regions of galaxy clusters. [Abridged]
△ Less
Submitted 28 August, 2012;
originally announced August 2012.
-
Gravitational Lensing with Three-Dimensional Ray Tracing
Authors:
Madhura Killedar,
Paul D. Lasky,
Geraint F. Lewis,
Chris J. Fluke
Abstract:
High redshift sources suffer from magnification or demagnification due to weak gravitational lensing by large scale structure. One consequence of this is that the distance-redshift relation, in wide use for cosmological tests, suffers lensing-induced scatter which can be quantified by the magnification probability distribution. Predicting this distribution generally requires a method for ray-traci…
▽ More
High redshift sources suffer from magnification or demagnification due to weak gravitational lensing by large scale structure. One consequence of this is that the distance-redshift relation, in wide use for cosmological tests, suffers lensing-induced scatter which can be quantified by the magnification probability distribution. Predicting this distribution generally requires a method for ray-tracing through cosmological N-body simulations. However, standard methods tend to apply the multiple thin-lens approximation. In an effort to quantify the accuracy of these methods, we develop an innovative code that performs ray-tracing without the use of this approximation. The efficiency and accuracy of this computationally challenging approach can be improved by careful choices of numerical parameters; therefore, the results are analysed for the behaviour of the ray-tracing code in the vicinity of Schwarzschild and Navarro-Frenk-White lenses. Preliminary comparisons are drawn with the multiple lens-plane ray-bundle method in the context of cosmological mass distributions for a source redshift of $z_{s}=0.5$.
△ Less
Submitted 21 October, 2011;
originally announced October 2011.
-
Lyman alpha absorbers in motion: consequences of gravitational lensing for the cosmological redshift drift experiment
Authors:
Madhura Killedar,
Geraint F. Lewis
Abstract:
The evolution of the expansion rate of the Universe results in a drift in the redshift of distant sources over time. A measurement of this drift would provide us with a direct probe of expansion history. The Lyman alpha forest has been recognized as the best candidate for this experiment, but the signal would be weak and it will take next generation large telescopes coupled with ultra-stable hig…
▽ More
The evolution of the expansion rate of the Universe results in a drift in the redshift of distant sources over time. A measurement of this drift would provide us with a direct probe of expansion history. The Lyman alpha forest has been recognized as the best candidate for this experiment, but the signal would be weak and it will take next generation large telescopes coupled with ultra-stable high resolution spectrographs to reach the cm/s resolution required. One source of noise that has not yet been assessed is the transverse motion of Lyman alpha absorbers, which varies the gravitational potential in the line of sight and subsequently shifts the positions of background absorption lines. We examine the relationship between the pure cosmic signal and the observed redshift drift in the presence of moving Lyman alpha clouds, particularly the collapsed structures associated with Lyman limit systems (LLSs) and damped Lyman alpha systems (DLAs). Surprisingly, the peculiar velocities and peculiar accelerations both enter the expression, although the acceleration term stands alone as an absolute error, whilst the velocity term appears as a fractional noise component. An estimate of the magnitude of the noise reassures us that the motion of the Lyman alpha absorbers will not pose a threat to the detection of the signal.
△ Less
Submitted 23 October, 2009;
originally announced October 2009.
