J/ApJ/782/41 231 AGN candidates from the 2FGL catalog (Doert+, 2014)
Search for gamma-ray-emitting active galactic nuclei in the Fermi-LAT
unassociated sample using machine learning.
Doert M., Errando M.
<Astrophys. J., 782, 41 (2014)>
=2014ApJ...782...41D 2014ApJ...782...41D (SIMBAD/NED BibCode)
ADC_Keywords: Active gal. nuclei ; Gamma rays
Keywords: catalogs - galaxies: active - gamma rays: galaxies -
methods: statistical
Abstract:
The second Fermi-LAT source catalog (2FGL) is the deepest all-sky
survey available in the gamma-ray band. It contains 1873 sources, of
which 576 remain unassociated. Machine-learning algorithms can be
trained on the gamma-ray properties of known active galactic nuclei
(AGNs) to find objects with AGN-like properties in the unassociated
sample. This analysis finds 231 high-confidence AGN candidates, with
increased robustness provided by intersecting two complementary
algorithms. A method to estimate the performance of the classification
algorithm is also presented, that takes into account the differences
between associated and unassociated gamma-ray sources. Follow-up
observations targeting AGN candidates, or studies of multiwavelength
archival data, will reduce the number of unassociated gamma-ray
sources and contribute to a more complete characterization of the
population of gamma-ray emitting AGNs.
Description:
The second Fermi-LAT source catalog (2FGL; Nolan et al. 2012, cat.
J/ApJS/199/31) is the deepest all-sky survey available in the
gamma-ray band. It contains 1873 sources, of which 576 remain
unassociated. The Large Area Telescope (LAT) on board the Fermi
Gamma-ray Space Telescope started operations in 2008.
In this work, machine-learning algorithms are used to identify
unassociated sources in the 2FGL catalog with properties similar to
gamma-ray-emitting Active Galactic Nuclei (AGN). This analysis finds
231 high-confidence AGN candidates (see Table3).
File Summary:
--------------------------------------------------------------------------------
FileName Lrecl Records Explanations
--------------------------------------------------------------------------------
ReadMe 80 . This file
table3.dat 45 231 List of high-confidence AGN candidates, ordered
by R.A.
--------------------------------------------------------------------------------
See also:
J/ApJS/209/9 : Unidentified gamma-ray sources. IV. X-ray (Paggi+, 2013)
J/ApJS/208/17 : 2nd Fermi LAT cat. of gamma-ray pulsars (2PC) (Abdo+, 2013)
J/ApJS/206/13 : Blazars with γ-ray counterparts. II. (Massaro+, 2013)
J/ApJ/779/133 : X-ray & radio fluxes of unassociated 2FGL sources
(Acero+, 2013)
J/MNRAS/424/L64 : AGN/pulsar distinction for 2FGL sources (Mirabal+, 2012)
J/ApJS/199/31 : Fermi LAT second source catalog (2FGL) (Nolan+, 2012)
J/ApJS/188/405 : Fermi-LAT first source catalog (1FGL) (Abdo+, 2010)
Byte-by-byte Description of file: table3.dat
--------------------------------------------------------------------------------
Bytes Format Units Label Explanations
--------------------------------------------------------------------------------
1- 13 A13 --- 2FGL Identifier in the second Fermi-LAT source
(2FGL) catalog (Nolan et al. 2012,
cat. J/ApJS/199/31) (JHHMM.s+DDMM or
JHHMM.s+DDMMc format)
15- 21 F7.3 deg RAdeg Right Ascension in decimal degrees (J2000)
23- 29 F7.3 deg DEdeg Declination in decimal degrees (J2000)
31- 34 F4.2 --- LRF [0.86/1] Random forest likelihood LRF (1)
36- 39 F4.2 --- LNN [0.92/1] Neural network likelihood LNN (1)
41 A1 --- Cl [AU] Class predicted by Mirabal et al. 2012
(cat. J/MNRAS/424/L64) (A=Active Galactic
Nucleus (AGN), U=Uncertain, NULL=Not present)
43 A1 --- C1 [a] Infrared counterpart in Massaro et al. 2013
(cat. J/ApJS/206/13)
44 A1 --- C2 [b] X-ray counterpart in Paggi et al. 2013
(cat. J/ApJS/209/9)
45 A1 --- C3 [c] AGN candidate in Acero et al. 2013
(cat. J/ApJ/779/133)
--------------------------------------------------------------------------------
Note (1): The sample of associated sources (1297 objects) was split into two
subsamples: training (70% of the sources) and test (30%). The training
sample was used to train the learning algorithms and optimize their
performance, while the test sample was set aside to evaluate the
performance of the classification methods once all the optimizations were
made. Two machine-learning classification methods, Random Forest (RF) and
Neutral Networks (NN) were selected. The overall classification scheme
(RF & NN) requires both RF and NN to label a source as AGN for it to be
considered an AGN candidate. Please refer to Section 3 in the paper for
further details.
--------------------------------------------------------------------------------
History:
From electronic version of the journal
(End) Prepared by [AAS]; Sylvain Guehenneux [CDS] 28-Jan-2016