J/A+A/685/A107 Machine learning applications for AGN studies (Mechbal+, 2024)
Machine learning applications in studies of the physical properties of active
galactic nuclei based on photometric observations.
Mechbal S., Ackermann M., Kowalski M.
<Astron. Astrophys. 685, A107 (2024)>
=2024A&A...685A.107M 2024A&A...685A.107M (SIMBAD/NED BibCode)
ADC_Keywords: Active gal. nuclei; QSOs ; X-ray sources ; Photometry ; Accretion
Keywords: accretion, accretion disks - methods: data analysis - catalogs -
galaxies: active - galaxies: fundamental parameters -
galaxies: photometry
Abstract:
We investigate the physical nature of active galactic nuclei (AGNs)
using machine learning (ML) tools.
We show that the redshift, z, bolometric luminosity, LBol, central
mass of the supermassive black hole (SMBH), MBH, Eddington ratio,
λEdd, and AGN class (obscured or unobscured) can be
reconstructed through multi-wavelength photometric observations only.
We trained a random forest regressor (RFR) ML-model on 7616
spectroscopically observed AGNs from the SPIDERS- AGN survey, which
had previously been cross-matched with soft X-ray observations (from
ROSAT or XMM), WISE mid-infrared photometry, and optical photometry
from SDSS ugriz filters. We built a catalog of 21 050 AGNs that were
subsequently reconstructed with the trained RFR; for 9687 sources, we
found archival redshift measurements. All AGNs were classified as
either type 1 or type 2 using a random forest classifier (RFC)
algorithm on a subset of known sources. All known photometric
measurement uncertainties were incorporated via a simulation-based
approach.
We present the reconstructed catalog of 21050 AGNs with redshifts
ranging from 0<z<2.5. We determined z estimations for 11363 new
sources, with both accuracy and outlier rates within 2%. The
distinction between type 1 or type 2 AGN could be identified with
respective efficiencies of 94% and 89%. The estimated obscuration
level, a proxy for AGN classification, of all sources is given in the
dataset. The LBol , MBH , and λEdd values are given for
21050 new sources with their estimated error.
The release of this catalog will advance AGN studies by presenting key
parameters of the accretion history of 6dex in luminosity over a wide
range of z. Similar applications of ML techniques using photometric
data only will be essential in the future, with large datasets from
eROSITA, JSWT, and the VRO poised to be released in the next decade.
Description:
This catalog includes the 21050 reconstructed sources, with results
from the obscuration classifier and estimation of z, LX, LBol, MBH,
LEdd, and λEdd with associated reconstruction uncertainties. In
addition, the 7613 SPIDERS sources used in the training sample are
also included. A description of the catalog's columns is given below.
Features providing X-ray, IR, and optical information are also given.
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
catalog.dat 842 28663 Positions, photometric magnitude and errors,
redshift, reconstructed AGN physical parameters
--------------------------------------------------------------------------------
See also:
VII/258 : Quasars and Active Galactic Nuclei (13th Ed.) (Veron+ 2010)
II/328 : AllWISE Data Release (Cutri+ 2013)
IX/53 : XMM-Newton slew survey Source Catalogue, version 2.0
(XMM-SSC, 2017)
J/A+A/588/A103 : Second ROSAT all-sky survey (2RXS) source catalog
(Boller+, 2016)
Byte-by-byte Description of file: catalog.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 6 A6 --- XName 2RXS or XMMSL2 X-ray detection (1)
8- 23 A16 --- Name X-ray identifier (JHHMMSS.s+DDMMSS)
25- 46 F22.18 deg RAdeg X-ray right ascension (J2000)
48- 70 E23.15 deg DEdeg X-ray declination (J2000)
72- 90 F19.15 [mW/m2] logFlux X-ray 0.5-2 keV band flux
92-115 E24.16 [mW/m2] e_logFlux ?=- X-ray 0.5-2 keV band flux error
119-137 A19 --- ALLWISE WISE All-Sky Release catalog name,
JHHMMSS.ss+DDMMSS.s (2)
141-151 F11.7 deg RAWdeg AllWISE right ascension (J2000)
153-163 E11.3 deg DEWdeg AllWISE declination (J2000)
165-182 F18.15 mag W1mag AllWISE W1 magnitude (Vega)
184-201 F18.15 mag W2mag AllWISE W2 magnitude (Vega)
203-221 F19.16 mag W3mag AllWISE W3 magnitude (Vega)
223-241 F19.16 mag W4mag AllWISE W4 magnitude (Vega)
243-263 F21.19 mag e_W1mag AllWISE W1 magnitude error (Vega)
265-284 F20.18 mag e_W2mag AllWISE W2 magnitude error (Vega)
286-305 F20.18 mag e_W3mag AllWISE W3 magnitude error (Vega)
307-326 F20.18 mag e_W4mag AllWISE W4 magnitude error (Vega)
328-349 F22.15 e/s FG Gaia G mean flux
351-371 F21.16 e/s e_FG Gaia G mean flux error
373-379 A7 --- Class Broad spectral classification computed by
the SDSS-DR16 spectroscopic pipeline
383-403 A21 --- SubClass Detailed spectral classification computed
by the SDSS-DR16 spectroscopic pipeline
406-413 F8.5 mag umagPSF SDSS PSF magnitude in the u-band (AB)
415-422 F8.5 mag gmagPSF SDSS PSF magnitude in the g-band (AB)
424-431 F8.5 mag rmagPSF SDSS PSF magnitude in the r-band (AB)
433-440 F8.5 mag imagPSF SDSS PSF magnitude in the i-band (AB)
442-449 F8.5 mag zmagPSF SDSS PSF magnitude in the z-band (AB)
451-463 F13.9 mag e_umagPSF SDSS PSF magnitude error in the u-band (AB)
465-478 E14.6 mag e_gmagPSF SDSS PSF magnitude error in the g-band (AB)
480-493 E14.6 mag e_rmagPSF SDSS PSF magnitude error in the r-band (AB)
495-508 E14.6 mag e_imagPSF SDSS PSF magnitude error in the i-band (AB)
510-521 F12.9 mag e_zmagPSF SDSS PSF magnitude error in the z-band (AB)
523-546 E24.16 --- z Redshift of the source
548-568 F21.19 --- e_z Uncertainty in the redshift
570-587 F18.15 [10-7W] logLX X-ray luminosity in the 0.5-2keV band
589-611 E23.15 [10-7W] e_logLX ?=- Uncertainty on the X-ray luminosity
in the 0.5-2keV band
613-630 F18.15 [10-7W] logLbol Bolometric luminosity
632-655 E24.16 [10-7W] e_logLbol Uncertainty on the bolometric luminosity
657-675 F19.16 [Msun] logMBH Black hole mass
677-700 E24.16 [Msun] e_logMBH Uncertainty on the black hole mass
702-719 F18.15 [10-7W] logLEdd Eddington luminosity
721-744 E24.16 [10-7W] e_logLEdd ?=- Uncertainty on the Eddington luminosity
746-767 E22.14 [-] logREdd Eddington ratio
769-791 E23.15 [-] e_logREdd ?=- Uncertainty on the uncertainty
793 I1 --- Recons [0/1]? Flag indicating whether the source and
values from columns (z to e_logEddRatio)
are from (3)
795 I1 --- knownz [0/1]? Flag indicating whether the redshift
values (z and e_z) come from (4)
797-820 E24.16 --- obscur [0/1] Value between 0 and 1 indicating
whether the source is obscured
(obscuration ∼ 1) or not (obscuration ∼ 0),
from the ML classifier
822-842 F21.19 --- e_obscur Uncertainty on the obscuration value
--------------------------------------------------------------------------------
Note (1): For 2RXS, Boller et al., 2016A&A...588A.103B 2016A&A...588A.103B, Cat. J/A+A/588/A103,
for XMMSL2, XMM-SSC, 2017, Cat. IX/53.
Note (2): From AllWISE, 2013, Cat. II/328.
Note (3): Flag as follows:
0 = from SPIDERS AGN spectroscopic catalog, Coffey et al.,
2019A&A...625A.123C 2019A&A...625A.123C
1 = from this work's ML reconstruction
Note (4): Flag as follows:
0 = from this work's ML reconstruction
1 = from Dwelly et al., 2017MNRAS.469.1065D 2017MNRAS.469.1065D or
Veron-cetty & Veron, 2010A&A...518A..10V 2010A&A...518A..10V, Cat. VII/258
--------------------------------------------------------------------------------
Acknowledgements:
Sarah Mechbal, sarah.mechbal(at)desy.de
(End) Patricia Vannier [CDS] 16-Feb-2024