J/A+A/644/A163 Pan-STARRS lens candidates from neural networks (Canameras+ 2020)
HOLISMOKES. II. Identifying galaxy-scale strong gravitational lenses in
Pan-STARRS using convolutional neural networks.
Canameras R., Schuldt S., Suyu S. H., Taubenberger S., Meinhardt T.,
Leal-Taixe L., Lemon C., Rojas K., Savary E.
<Astron. Astrophys. 644, A163 (2020)>
=2020A&A...644A.163C 2020A&A...644A.163C (SIMBAD/NED BibCode)
ADC_Keywords: Surveys ; Gravitational lensing ; Redshifts ; Photometry, ugriz
Keywords: gravitational lensing: strong - methods: data analysis -
galaxies: distances and redshifts - surveys
Abstract:
We present a systematic search for wide-separation (with Einstein
radius ∼1.5"), galaxy-scale strong lenses in the 30000 sq.deg of the
Pan-STARRS 3pi survey on the Northern sky. With long time delays of a
few days to weeks, these types of systems are particularly well-suited
for catching strongly lensed supernovae with spatially-resolved
multiple images and offer new insights on early-phase supernova
spectroscopy and cosmography. We produced a set of realistic
simulations by painting lensed COSMOS sources on Pan-STARRS image
cutouts of lens luminous red galaxies (LRGs) with redshift and
velocity dispersion known from the sloan digital sky survey (SDSS).
First, we computed the photometry of mock lenses in gri bands and
applied a simple catalog-level neural network to identify a sample of
1050207 galaxies with similar colors and magnitudes as the mocks.
Second, we trained a convolutional neural network (CNN) on Pan-STARRS
gri image cutouts to classify this sample and obtain sets of 105760
and 12382 lens candidates with scores of pCNN>0.5 and >0.9,
respectively. Extensive tests showed that CNN performances rely
heavily on the design of lens simulations and the choice of negative
examples for training, but little on the network architecture. The CNN
correctly classified 14 out of 16 test lenses, which are previously
confirmed lens systems above the detection limit of Pan-STARRS.
Finally, we visually inspected all galaxies with pCNN>0.9 to assemble
a final set of 330 high-quality newly-discovered lens candidates while
recovering 23 published systems. For a subset, SDSS spectroscopy on
the lens central regions proves that our method correctly identifies
lens LRGs at z∼0.1-0.7. Five spectra also show robust signatures of
high-redshift background sources, and Pan-STARRS imaging confirms one
of them as a quadruply-imaged red source at zs=1.185, which is likely
a recently quenched galaxy strongly lensed by a foreground LRG at
zd=0.3155. In the future, high-resolution imaging and spectroscopic
follow-up will be required to validate Pan-STARRS lens candidates and
derive strong lensing models. We also expect that the efficient and
automated two-step classification method presented in this paper will
be applicable to the ∼4 mag deeper gri stacks from the Rubin
Observatory Legacy Survey of Space and Time (LSST) with minor
adjustments.
Description:
Complete list of galaxy-scale strong lens candidates with lens
luminous red galaxies (LRGs) from a systematic search in Pan-STARRS.
These systems were selected with a neural network, as high confidence
candidates with CNN scores >0.9 and average grades >2.0 from visual
inspection (i.e. corresponding to definite or probable lenses). For
each candidate, the CNN score, average visual grade, photometry, and
SDSS redshift (where available) are given.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table1.dat 103 358 Complete list of strong lens candidates from visual
inspection of the highest neural network scores
--------------------------------------------------------------------------------
See also:
II/349 : The Pan-STARRS release 1 (PS1) Survey - DR1 (Chambers+, 2016)
Byte-by-byte Description of file: table1.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 13 A13 --- Name Candidate identifier (PS1JHHMM+DDMM)
16- 17 I2 h RAh Right Ascension (J2000)
18- 19 I2 min RAm Right Ascension (J2000)
20- 21 I2 s RAs Right Ascension (J2000)
24 A1 --- DE- Declination sign (J2000)
25- 26 I2 deg DEd Declination (J2000)
27- 28 I2 arcmin DEm Declination (J2000)
29- 30 I2 arcsec DEs Declination (J2000)
33- 37 F5.3 --- pCNN Output score from the neural network
40- 43 F4.2 --- Grade Average of the visual grades from four authors
46- 50 F5.2 mag gKron ? PS1 catalog g band Kron magnitude (1)
53- 57 F5.2 mag rKron ? PS1 catalog r band Kron magnitude
60- 64 F5.2 mag iKron ? PS1 catalog i band Kron magnitude
67- 71 F5.2 mag gAper PS1 catalog g band Aper magnitude (2)
74- 78 F5.2 mag rAper PS1 catalog r band Aper magnitude
81- 85 F5.2 mag iAper PS1 catalog i band Aper magnitude
88- 93 F6.4 --- z ? Redshift (3)
96 A1 --- f_z [*] * for spectroscopic redshift,
else photometric
99-103 A5 --- Ref Reference(s) for candidates in the
literature (4)
--------------------------------------------------------------------------------
Note (1): Kron magnitudes of the lens and source blends.
Note (2): Aperture magnitudes of 1.04" radii covering the lens central regions.
Note (3): SDSS photometric redshifts, with spectroscopic redshifts substituted
for sources marked with f_z.
Note (4): References for sources previously published as confirmed or
candidate lens systems (grades A and B or equivalent) as follows:
a = Jacobs et al., 2019ApJS..243...17J 2019ApJS..243...17J, Cat. J/ApJS/243/17
b = Diehl et al., 2017ApJS..232...15D 2017ApJS..232...15D, Cat. J/ApJS/232/15
c = Sonnenfeld et al., 2018PASJ...70S..29S 2018PASJ...70S..29S
d = Huang et al., 2020ApJ...894...78H 2020ApJ...894...78H
e = Wong et al., 2018ApJ...867..107W 2018ApJ...867..107W
f = Petrillo et al., 2019MNRAS.484.3879P 2019MNRAS.484.3879P
g = Stark et al., 2013MNRAS.436.1040S 2013MNRAS.436.1040S
h = Auger et al., 2009ApJ...705.1099A 2009ApJ...705.1099A, Cat. J/ApJ/705/1099
i = Jacobs et al., 2019MNRAS.484.5330J 2019MNRAS.484.5330J
j = Lemon et al., 2019MNRAS.483.4242L 2019MNRAS.483.4242L
k = Wang et al., 2017MNRAS.468.3757W 2017MNRAS.468.3757W
l = Jaelani et al., 2020MNRAS.495.1291J 2020MNRAS.495.1291J
m = Schirmer et al., 2010A&A...514A..60S 2010A&A...514A..60S, Cat. J/A+A/514/A60
--------------------------------------------------------------------------------
Acknowledgements:
Raoul Canameras, rcanameras(at)mpa-garching.mpg.de
References:
Suyu et al., Paper I 2020A&A...644A.162S 2020A&A...644A.162S
(End) Raoul Canameras [MPA, Germany], Patricia Vannier [CDS] 08-Sep-2020