J/A+A/668/A73 Search of strong lens systems in DES (Rojas+, 2022)
Search of strong lens systems in the Dark Energy Survey using
convolutional neural networks.
Rojas K., Savary E., Clement B., Maus M., Courbin F., Lemon C., Chan J.H.H.,
Vernardos G., Joseph R., Canameras R., Galan A.
<Astron. Astrophys. 668, A73 (2022)>
=2022A&A...668A..73R 2022A&A...668A..73R (SIMBAD/NED BibCode)
ADC_Keywords: Surveys ; Gravitational lensing
Keywords: gravitational lensing: strong - techniques: image processing -
surveys - catalogs
Abstract:
We present our search for strong lens, galaxy-scale systems in the
first data release of the Dark Energy Survey (DES), based on a
color-selected parent sample of 18745029 luminous red galaxies
(LRGs). We used a convolutional neural network (CNN) to grade this LRG
sample with values between 0 (non-lens) and 1 (lens). Our training set
of mock lenses is data-driven, that is, it uses lensed sources taken
from HST-COSMOS images and lensing galaxies from DES images of our LRG
sample. A total of 76582 cutouts were obtained with a score above
0.9, which were then visually inspected and classified into two
catalogs. The first one contains 405 lens candidates, of which 90
present clear lensing features and counterparts, while the other 315
require more evidence, such as higher resolution imaging or spectra,
to be conclusive. A total of 186 candidates are newly identified by
our search, of which 16 are among the 90 most promising (best)
candidates. The second catalog includes 539 ring galaxy candidates.
This catalog will be a useful false positive sample for training
future CNNs. For the 90 best lens candidates we carry out color-based
deblending of the lens and source light without fitting any analytical
profile to the data. This method is shown to be very efficient in the
deblending, even for very compact objects and for objects with a
complex morphology. Finally, from the 90 best lens candidates, we
selected 52 systems with one single deflector to test an automated
modeling pipeline that has the capacity to successfully model 79% of
the sample within an acceptable computing runtime.
Description:
We used DES-DR1 to search for galaxy-scale strong lensing systems
using a CNN that carries out a binary classification of optical images
in the g, r; and i bands. In doing so, we targeted massive galaxies,
that is, LRGs, which were selected using a wide color-magnitude cut
accounting for realistic color contamination by the putative
background star-forming blue galaxies.
From our visual inspection, we created two main catalogs: a
lens candidates catalog and a ring galaxy candidates catalog, the
latter being our main source of contaminants. The first catalog
contains a total of 405 lens candidate systems: 90 SL and 315 ML. Out
of these, 186 were totally new systems and 219 were identified (but
not necessarily confirmed) in previous searches. We deblended the lens
and source light for our 90 SL systems using the MuSCADeT software,
which does not involve any profile fitting, but uses the color
contrast between the lens and source together with sparse
regularization. This was successful in deblending most of the cases,
where there were clear differences in the colors of the lens and
source. The second catalog contains 539 ring galaxy candidates. We
expect to use this ring catalog in the future to improve the training
of machine learning algorithms in the recognition between lenses and
ring galaxies. Still, 539 objects is not a sufficient sample for
training CNNs and, thus, furthers work with, for instance, generative
adversarial networks, is likely to be needed.
Finally, we selected, from the SL category, the 52 systems that
apparently had one well-defined galaxy as a deflector to test an
automated modeling pipeline.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table1.dat 101 90 Strong lens (SL) candidates
table2.dat 96 315 Maybe lens (ML) candidates
table3.dat 71 539 Ring galaxies candidates
table4.dat 98 61 Strong lensing model
refs.dat 65 30 References
--------------------------------------------------------------------------------
See also:
II/357 : The Dark Energy Survey (DES): Data Release 1 (Abbott+, 2018)
II/371 : The Dark Energy Survey (DES): Data Release 2 (Abbott+, 2021)
Byte-by-byte Description of file: table1.dat table2.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 A18 --- Cand Candidate name (DES JHHMMSS+DDMMSS)
20- 39 F20.16 deg RAdeg Right ascension (J2000)
41- 61 F21.17 deg DEdeg Declination (J2000)
63- 66 F4.2 --- SCnn [0.8/1.0] Cnn score (G1)
68- 71 F4.2 --- VISL [0/1] Visual inspection score (G1)
73-101 A29 --- Ref References (SL or ML) in refs.dat file
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table3.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 A18 --- Cand Candidate name (DES JHHMMSS+DDMMSS)
20- 39 F20.16 deg RAdeg Right ascension (J2000)
41- 61 F21.17 deg DEdeg Declination (J2000)
63- 66 F4.2 --- SCnn [0.9/1.0] Cnn score (G1)
68- 71 F4.2 --- VISR [0.28/1.0] Visual inspection score (G1)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: table4.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 18 A18 --- Cand Candidate name (DES JHHMMSS+DDMMSS)
19 A1 --- n_Cand [*] * for model result after manually
redoing the mask
21- 24 F4.2 --- rchi2 Reduced chi2 of the model
26- 30 F5.3 arcsec RE Einstein radius
31 A1 --- --- [+]
32- 36 F5.3 arcsec E_RE Error on RE (upper value)
37 A1 --- --- [-]
38- 42 F5.3 arcsec e_RE Error on RE (lower value)
44- 48 F5.3 --- qm Axis ratio
49 A1 --- --- [+]
50- 54 F5.3 --- E_qm Error on qm ratio (upper value)
55 A1 --- --- [-]
56- 60 F5.3 --- e_qm Error on qm ratio (lower value)
62- 64 I3 deg PAm Position angle of the lens mass SIE model
65 A1 --- --- [+]
66- 67 I2 deg E_PAm Error on PAm (upper value)
68 A1 --- --- [-]
69- 70 I2 deg e_PAm Error on PAm (lower value)
72- 76 F5.3 --- gammaext strength of external shear
77 A1 --- --- [+]
78- 82 F5.3 --- E_gammaext Error on gammaext (upper value)
83 A1 --- --- [-]
84- 88 F5.3 --- e_gammaext Error on gammaext (lower value)
90- 92 I3 deg phiext angle of external shear
93 A1 --- --- [+]
94- 95 I2 deg E_phiext Error on phiext (upper value)
96 A1 --- --- [-]
97- 98 I2 deg e_phiext Error on phiext (lower value)
--------------------------------------------------------------------------------
Byte-by-byte Description of file: refs.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 3 A3 --- Ref Reference code
5- 23 A19 --- BibCode BibCode
25- 42 A18 --- Aut Author's name
45- 65 A21 --- Com Comments
--------------------------------------------------------------------------------
Global notes:
Note (G1): images classified as lenses obtain ∼1 while non-lenses obtain ∼0.
--------------------------------------------------------------------------------
History:
From Karina Rojas, karina.rojasolate(at)gmail.com
Acknowledgements:
This work is supported by the Swiss National Science Foundation (SNSF)
and by the European Research Council (ERC) under the European
Union's Horizon 2020 research and innovation program (COSMICLENS:
grant agreement No 787886).
RC thanks the Max Planck Society for support through the Max Planck
Research Group of S.∼H.∼Suyu. This project has received funding from
the European Research Council (ERC) under the European Unions Horizon
2020 research and innovation programme (LENSNOVA: grant agreement No
771776). GV has received funding from the European Union's Horizon
2020 research and innovation programme under the Marie
Sklodovska-Curie grant agreement No 897124.
This research has made use of the VizieR catalogue access tool, CDS,
Strasbourg, France (DOI : 10.26093/cds/vizier). The original
description of the VizieR service was published in Ochsenbein et al.,
2000A&AS..143...23O 2000A&AS..143...23O .
This research has made use of the SIMBAD database, operated at CDS,
Strasbourg, France (Wenger et al., 2000A&AS..143....9W 2000A&AS..143....9W).
This material is based upon work supported by the National Science
Foundation under Cooperative Agreement 1258333 managed by the
Association of Universities for Research in Astronomy (AURA), and the
Department of Energy under Contract No. DE-AC02-76SF00515 with the
SLAC National Accelerator Laboratory. Additional funding for
Rubin Observatory comes from private donations, grants to
universities, and in-kind support from LSSTC Institutional Members.
This work rely on the following packages: numpy (van der Walt et al.,
2011, Computing in Science Engineering, 13, 22), scipy (Virtanen et
al., 2020, Nature Methods, 17, 261), matplotlib (Hunter, 2007,
Computing in Science & Engineering, 9, 90), astropy (Prince-Whelam et
al., 2018AJ....156..123A 2018AJ....156..123A) and aGithub scarlet_extensions
(https://github.com/herjy/scarlet_extensions/releases/tag/paper) and
csv. A notebook detailing the procedure of deblending can be found in
the following repository: aGithub Lens-Deblend
(https://github.com/herjy/Lens_deblend/releases/tag/prototype).
(End) Patricia Vannier [CDS] 10-Oct-2022