Searching for galaxy-scale strong-lenses in galaxy clusters with deep networks -- I: methodology and network performance

Abstract

Galaxy-scale strong lenses in galaxy clusters provide a unique tool to investigate their inner mass distribution and the sub-halo density profiles in the low-mass regime, which can be compared with the predictions from cosmological simulations. We search for galaxy-galaxy strong-lensing systems in HST multi-band imaging of galaxy cluster cores from the CLASH and HFF programs by exploring the classification capabilities of deep learning techniques. Convolutional neural networks are trained utilising highly-realistic simulations of galaxy-scale strong lenses injected into the HST cluster fields around cluster members. To this aim, we take advantage of extensive spectroscopic information on member galaxies in 16 clusters and the accurate knowledge of the deflection fields in half of these from high-precision strong lensing models. Using observationally-based distributions, we sample magnitudes, redshifts and sizes of the background galaxy population. By placing these sources within the secondary caustics associated with cluster galaxies, we build a sample of ~3000 galaxy-galaxy strong lenses which preserve the full complexity of real multi-colour data and produce a wide diversity of strong lensing configurations. We study two deep learning networks processing a large sample of image cutouts in three HST/ACS bands, and we quantify their classification performance using several standard metrics. We find that both networks achieve a very good trade-off between purity and completeness (85%-95%), as well as good stability with fluctuations within 2%-4%. We characterise the limited number of false negatives and false positives in terms of the physical properties of the background sources and cluster members. We also demonstrate the neural networks' high degree of generalisation by applying our method to HST observations of 12 clusters with previously known galaxy-scale lensing systems.

17 pages, 13 figures, to be published on A&A

References in corpus (23)

• The Cosmic Evolution Survey (COSMOS) -- Overview
• The James Webb Space Telescope
• A Bayesian approach to strong lensing modelling of galaxy clusters
• The GLASS James Webb Space Telescope Early Release Science Program. I. Survey Design and Release Plans
• Discovery and properties of ultra-high redshift galaxies ($9<z<12$) in the JWST ERO SMACS 0723 Field
• The MUSE 3D view of the Hubble Deep Field South
• Multi-scale cluster lens mass mapping I. Strong Lensing modelling
• Cosmological Constraints from Strong Gravitational Lensing in Galaxy Clusters
• An excess of small-scale gravitational lenses observed in galaxy clusters
• The SL2S Galaxy-scale Lens Sample. V. Dark Matter Halos and Stellar IMF of Massive Early-type Galaxies out to Redshift 0.8
• RingFinder: automated detection of galaxy-scale gravitational lenses in ground-based multi-filter imaging data
• LinKS: Discovering galaxy-scale strong lenses in the Kilo-Degree Survey using Convolutional Neural Networks
• Mass distribution in the core of MACS J1206: robust modeling from an exceptionally large sample of central multiple images
• The MUSE Deep Lensed Field on the Hubble Frontier Field MACS~J0416
• Realistic simulations of gravitational lensing by galaxy clusters: extracting arc parameters from mock DUNE images
• Gravitational lenses and lens candidates identified from the COSMOS field
• A new high-precision strong lensing model of the galaxy cluster MACS J0416.1-2403
• HOLISMOKES. VI. New galaxy-scale strong lens candidates from the HSC-SSP imaging survey
• Deep Learning for Strong Lensing Search: Tests of the Convolutional Neural Networks and New Candidates from KiDS DR3
• The probability of galaxy-galaxy strong lensing events in hydrodynamical simulations of galaxy clusters
• Pilot-WINGS: An extended MUSE view of the structure of Abell 370
• The Kormendy relation of early-type galaxies as a function of wavelength in Abell S1063, MACS J0416.1-2403, and MACS J1149.5+2223
• The Orthogonally Aligned Dark Halo of an Edge-on Lensing Galaxy in the Hubble Frontier Fields: A Challenge for Modified Gravity

Cited by in corpus (6)

• Euclid Preparation XXXIII. Characterization of convolutional neural networks for the identification of galaxy-galaxy strong lensing events
• A model for galaxy-galaxy strong lensing statistics in surveys
• HOLISMOKES XIII: Strong-lens candidates at all mass scales and their environments from the Hyper-Suprime Cam and deep learning
• YOLO-CL cluster detection in the Rubin/LSST DC2 simulation
• Navigating AGN variability with self-organizing maps
• Most Strong Lensing Deflectors in the AGEL Survey Are in Group and Cluster Environments