[Paper Review] EasyCritics II. Testing its efficiency: new gravitational lens candidates in CFHTLenS

M. Carrasco, S. Stapelberg+4 more|arXiv (Cornell University)|Jul 10, 2018

Galaxies: Formation, Evolution, Phenomena2 references4 citations

TL;DR

EasyCritics II presents a fully automated algorithm that uses photometric data of bright elliptical galaxies (0.2 ≤ z ≤ 0.9) to model strong-lensing potential and generate critical curves, reducing the inspection area in CFHTLenS from 154 sq. deg to just 0.623 sq. deg (0.4%) while identifying 32 of 44 known lenses and discovering 9 new candidates with only 9% spurious detections.

ABSTRACT

We report the results of $EasyCritics$, a fully automated algorithm for the efficient search of strong-lensing (SL) regions in wide-field surveys, applied to the Canada-France-Hawaii Telescope Lensing Survey (CFHTLenS). By using only the photometric information of the brightest elliptical galaxies distributed over a wide redshift range ($\smash{0.2 \lesssim z \lesssim 0.9}$) and without requiring the identification of arcs, our algorithm produces lensing potential models and catalogs of critical curves of the entire survey area. We explore several parameter set configurations in order to test the efficiency of our approach. In a specific configuration, $EasyCritics$ generates only $\sim1200$ possibly super-critical regions in the CFHTLS area, drastically reducing the effective area for inspection from $154$ sq. deg to $\sim0.623$ sq. deg, $i.e.$ by more than two orders of magnitude. Among the pre-selected SL regions, we identify 32 of the 44 previously known lenses on the group and cluster scale, and discover 9 new promising lens candidates. The detection rate can be easily improved to $\sim82\%$ by a simple modification in the parameter set, but at the expense of increasing the total number of possible SL candidates. Note that $EasyCritics$ is fully complementary to other arc-finders since we characterize lenses instead of directly identifying arcs. Although future comparisons against numerical simulations are required for fully assessing the efficiency of $EasyCritics$, the algorithm seems very promising for upcoming surveys covering $\smash{10^{4}}$ sq. deg, such as the $Euclid$ mission and $LSST$, where the pre-selection of candidates for any kind of SL analysis will be indispensable due to the expected enormous data volume.

Motivation & Objective

• To develop an automated, efficient method for pre-selecting strong-lensing regions in wide-field surveys without relying on arc detection.
• To test the efficiency and reliability of EasyCritics in identifying known and new strong-lensing systems in the CFHTLenS survey.
• To reduce the effective area requiring manual inspection by orders of magnitude, enabling scalability for upcoming large surveys like Euclid and LSST.
• To validate the algorithm’s lens characterization by comparing Einstein radii with independent weak-lensing and luminosity measurements.
• To provide a complementary approach to arc-finding methods by focusing on lens potential modeling rather than visual arc identification.

Proposed method

• The algorithm uses photometric redshifts and flux measurements of the brightest elliptical galaxies (0.2 ≤ z ≤ 0.9) as lens plane inputs.
• It constructs gravitational lensing potential models using a singular isothermal sphere (SIS) approximation and computes critical curves and Einstein radii.
• Multiple parameter configurations are tested, varying the cluster velocity dispersion threshold (σ_clus) to optimize detection efficiency.
• The method generates pre-selected candidate regions (SL regions) based on super-critical curves, minimizing reliance on visual inspection.
• RGB FITS cutouts of 82'' × 82'' are created for all candidates to enable visual follow-up and feature assessment.
• Results are validated by cross-matching with known lenses (KLs and KCs), weak-lensing velocity dispersions, and total luminosities.

Experimental results

Research questions

• RQ1 Can an automated photometric-only method efficiently pre-select strong-lensing regions in wide-field surveys without relying on arc detection?
• RQ2 How does the detection efficiency of EasyCritics compare to known lens samples in CFHTLenS, and what is its false-positive rate?
• RQ3 To what extent do the Einstein radii derived by EasyCritics correlate with independent mass estimates (e.g., weak-lensing velocity dispersion and total luminosity)?
• RQ4 Can EasyCritics reduce the inspection area for strong-lensing searches by more than two orders of magnitude while preserving high completeness?
• RQ5 How does the algorithm perform in identifying both known and new lens candidates in a well-studied survey like CFHTLenS?

Key findings

• EasyCritics reduced the effective inspection area in CFHTLenS from 154 sq. deg to 0.623 sq. deg, a reduction of over 99.6%.
• The algorithm identified 32 of 44 known cluster-scale lenses (KLs) and 12 of 44 known cusp-scale lenses (KCs), achieving 73% completeness for KLs.
• Only 9% of the 1200 pre-selected SL regions were spurious, primarily due to photometric catalog issues or line-of-sight structures.
• Einstein radii derived by EasyCritics showed a strong correlation with weak-lensing velocity dispersions (scatter ~37%) and total luminosities (scatter ~31%).
• Nine new promising lens candidates were discovered after visual inspection of the pre-selected 1200 regions, indicating high discovery potential.
• A modified parameter set increased detection completeness to ~82%, though at the cost of increasing the number of candidates.