The PCA Lens-Finder: application to CFHTLS

Abstract

We present the results of a new search for galaxy-scale strong lensing systems in CFHTLS Wide. Our lens-finding technique involves a preselection of potential lens galaxies, applying simple cuts in size and magnitude. We then perform a Principal Component Analysis of the galaxy images, ensuring a clean removal of the light profile. Lensed features are searched for in the residual images using the clustering topometric algorithm DBSCAN. We find 1098 lens candidates that we inspect visually, leading to a cleaned sample of 109 new lens candidates. Using realistic image simulations we estimate the completeness of our sample and show that it is independent of source surface brightness, Einstein ring size (image separation) or lens redshift. We compare the properties of our sample to previous lens searches in CFHTLS. Including the present search, the total number of lenses found in CFHTLS amounts to 678, which corresponds to ∼4 lenses per square degree down to i(AB) = 24.8. This is equivalent to ∼60.000 lenses in total in a survey as wide as Euclid, but at the CFHTLS resolution and depth.

The DES Bright Arcs Survey: Hundreds of Candidate Strongly Lensed Galaxy Systems from the Dark Energy Survey Science Verification and Year 1 Observations

“…These include a search for elongated objects (Alard 2006), an ARCFINDER (Seidel & Bartelmann 2007) that identifies SL candidates associated with galaxy clusters or groups, analysis of third-order moments of galaxy shapes (Kubo & Dell'Antonio 2008) to find systems with arcs in the Deep Lens Survey (Wittman et al 2006), principal component analysis (PCA) to identify (Joseph et al 2014;Paraficz et al 2016) SL systems with complete or nearly complete Einstein rings, and Deep Learning (Lanusse et al 2017) and neural network (de Bom et al 2017;Petrillo et al 2017) analysis of galaxy shapes. Another new method, YATTALENS (Sonnenfeld et al 2017), identifies galaxy-galaxy lens candidates with arc-like features by modeling the source and lens galaxies and subtracting the lens galaxy from the image.…”

A neural network gravitational arc finder based on the Mediatrix filamentation method

“…Another possibility is to use the growing number of strong lensing systems detected in wide-field surveys and HST images to perform the training on real data sets. For example, over 600 candidate systems have been detected in recent studies using CFHTLS data (Maturi et al 2014;Gavazzi et al 2014;Brault & Gavazzi 2015;More et al 2016;Paraficz et al 2016), which could be used to train and better characterize the AMA and other arc finders. By training and validating the ANN with more realistic simulated arcs or with real data, we expect to reach a better agreement for c in comparison to applications to other data sets (and therefore to achieve a higher completeness), which is different from what we found when applying our trained ANN to the HST data.…”

“…This includes searches in Hubble Space Telescope (HST) mosaics, such as the Hubble Deep Field (HDF; Hogg et al 1996), HST Medium Deep Survey (Ratnatunga et al 1999), Great Observatories Origins Deep Survey (GOODS; Fassnacht et al 2004), Extended Groth Strip (EGS; Marshall et al 2009), HST Cosmic Evolution survey (COSMOS; Faure et al 2008;Jackson 2008) and in targeted observations of galaxies Brownstein et al 2012) and clusters Sand et al 2005;Horesh et al 2010;Xu et al 2016). Investigations from the ground include follow-ups of clusters (Luppino et al 1999;Zaritsky & Gonzalez 2003;Hennawi et al 2008;Kausch et al 2010;Furlanetto et al 2013a) and galaxies (Willis et al 2006), and searches in wide-field surveys, such as the Red-Sequence Cluster Survey (RCS; Gladders et al 2003;Bayliss 2012), Sloan Digital Sky Survey (SDSS; Estrada et al 2007;Belokurov et al 2009;Kubo et al 2010;Wen et al 2011;Bayliss 2012), Deep Lens Survey (DLS; Kubo & Dell'Antonio 2008), Canada-France-Hawaii Telescope A&A 597, A135 (2017) (CFHT) Legacy Survey (CFHTLS; Cabanac et al 2007;More et al 2012;Maturi et al 2014;Gavazzi et al 2014;More et al 2016;Paraficz et al 2016), CFHT Stripe 82 Survey (CS82; Caminha, More et al, in prep. ), and Dark Energy Survey 1 (DES; The Dark Energy Survey Collaboration 2005) Science Verification data (Nord et al 2016).…”

The Euclid Data Processing Challenges

“…3) is ongoing to tests different algorithms of source (including strong lens) detection and deblending (see Paraficz et al 2016 andTramacere et al 2016). …”