Evolution of Bias in Radio Galaxies

Evolution of Bias in Radio Galaxies
Sam Lindsay, Centre for Astrophysics Research, University of Hertfordshire
Introduction
The galaxy bias parameter quantifies how strongly a population traces the underlying dark matter structure. Observing at radio wavelengths, we target the
synchrotron emission of massive star-forming galaxies and AGN with a stronger bias than optically-selected surveys. The wide redshift range probed by
radio surveys further enables study into the evolution of clustering properties over cosmic time. The angular correlation function provides a 2D clustering
measure, while the redshift distribution allows further calculation of spatial properties over time. This poster shows the results of the analysis of a large area
(210 deg2 @ 1 mJy)
mJy) of the VLA FIRST survey cross-matched with redshift catalogues from GAMA (spec-z
(spec-z’’s) and SDSS/UKIDSS (photo-z
(photo-z’’s), and preliminary
results of a cross-correlation function analysis of deep VLA radio fields (2 x 1 deg2 @ 60 µJy)
Jy) with deep IR catalogues using the VISTA telescope.
Limber Inversion and Linear Bias
Angular Correlation Function
Comparing an angular distribution of galaxies on the sky with a randomly
scattered mock catalogue, we quantify the excess clustering as a function of
angular separation, defined as follows:
With the angular clustering parameters and a known redshift distribution we
may infer the spatial correlation length, r0 by Limber inversion (Limber 1953;
Peebles 1980):
where DD,
DD, DR and RR are normalized binned pair counts between data (D
(D)
and random (R
(R) catalogues. This correlation function is fitted by a power law
with amplitude, A and slope parameter (γ-1). The cross-correlation function
between two galaxy distributions (D1 and D2) is computed analogously with
D1D2, D1R and D2R pair counts.
The linear bias, b (defined as the square root of the ratio of the galaxy and
dark matter correlation functions) is a function of r0 and calculated by:
Radio + Optical Data
FIRST + GAMA/SDSS/UKIDSS
VLA + VISTA
• VLA FIRST radio data @ 1.4 GHz to ~1 mJy flux density limit and ~5”
~5”
resolution. Sources within 72”
72” of one another are collapsed to a single source
at the flux-weighted centre to find cores of multiple-component radio galaxies.
2 x 1 sq. deg. fields: XMM3/CFHTLS-D1 and VLA-COSMOS Large
• GAMA spec-z’
spec-z’s (r < 19.8) and SDSS/UKIDSS photo-z’
photo-z’s (r < 22) for radio
sources with optical counterparts within 3”
3”, covering 210 sq. deg.
• VISTA infrared data from VIDEO/UltraVISTA surveys, respectively, to
similar depths of Ks < ~23.5.
• VLA radio data @ 1.4 GHz to ~60 µJy
µJy flux density limit and 1.6”
1.6”/6”
/6”
resolution, respectively.
~14,000 radio sources with ~4,000 optical identifications
2 x ~100,000 IR sources with 2 x ~1,000 radio counterparts
w(θ ), and therefore bIR, is well
constrained for the Ks < 23 VIDEO
sources.
The two-point angular correlation
function, w(θ ), of FIRST sources
with GAMA/SDSS redshifts,
redshifts, with
bootstrap errors. Inset are 68%,
90% and 95% confidence
contours for power law amplitude
and slope.
The angular cross-correlation
function between radio and IR
sources gives the relative bias
between populations.
N(z) for the 1 mJy radio
population is assumed from the
SKADS simulations, allowing us to
infer the unmatched radio N(z) by
subtracting the matched radio N(z).
Conclusions
Correlation length
(top) and bias (bottom) against
redshift for matched radio sources
in red, unmatched in green and
all-radio samples in blue (the lower
value of which represents a wider
sample of the FIRST catalogue).
The solid lines are the bias
prescriptions (Mo & White 1996) for
different radio populations used in
SKADS (Wilman et al. 2008),
highlighting the increase of bias
with redshift which appears
exaggerated in our results.
• Identifying optical counterparts in GAMA/SDSS to FIRST radio sources
allows us to calculate a bias of ~2 to 4 from z ~ 0.3 to 0.6
• Adding an assumed radio redshift distribution (that used in SKADS
simulations) allows this redshift range to be extended to z > 1
• We find the bias to increase with redshift ahead of the rate predicted by
Mo & White (1996) models, suggesting an increasing galaxy mass/AGN
fraction being observed at high redshift
• Better constraints on the angular correlation function can be made using
narrower deep radio fields, and cross-correlating with an infrared
catalogue provides an indirect method of measuring the radio bias
• Expanding deep radio coverage over similar fields (e.g. with SKA
precursors/pathfinders) will allow finer redshift binning and smaller errors in
the radio galaxy bias over cosmic time.
References
Becker R. H. et al., 1995, ApJ,
(FIRST)
ApJ, 450, 559
Driver S.P. et al., 2011, MNRAS, 413, 917
(GAMA)
Jarvis M. J., et al. 2013, MNRAS, 428, 1281
(VIDEO)
Mo, H. J., White, S. D. M. 1996, MNRAS, 282, 347
Muzzin A. et al., 2013, ApJS,
(UltraVISTA)
ApJS, 206, 8
UltraVISTA)
Peebles P.J.E., 1980, The Large Scale Structure of the Universe,Princeton
Universe,Princeton University Press
Schinnerer E. et al., 2007, ApJS,
(VLA-COSMOS)
ApJS, 172, 46
Wilman R.J. et al., 2008, MNRAS, 388, 1335
(SKADS)
Contact Information
E-mail:
E-mail: Sam Lindsay
- [email protected]
Matt Jarvis (supervisor)
- [email protected]
Web:
Web:
www.samlindsay.co.uk
Image: ESO/B. Tafreshi