We present the 3D real space clustering power spectrum of a sample of \sim 600 , 000 luminous red galaxies ( LRGs ) measured by the Sloan Digital Sky Survey ( SDSS ) , using photometric redshifts .
These galaxies are old , elliptical systems with strong 4000 Å breaks , and have accurate photometric redshifts with an average error of \Delta z = 0.03 .
This sample of galaxies ranges from redshift z = 0.2 to 0.6 over 3 , 528 { deg } ^ { 2 } of the sky , probing a volume of 1.5 h ^ { -3 } { Gpc } ^ { 3 } , making it the largest volume ever used for galaxy clustering measurements .
We measure the angular clustering power spectrum in eight redshift slices and use well-calibrated redshift distributions to combine these into a high precision 3D real space power spectrum from k = 0.005 h { Mpc } ^ { -1 } to k = 1 h { Mpc } ^ { -1 } .
We detect power on gigaparsec scales , beyond the turnover in the matter power spectrum , at a \sim 2 \sigma significance for k < 0.01 h { Mpc } ^ { -1 } , increasing to 5.5 \sigma for k < 0.02 h { Mpc } ^ { -1 } .
This detection of power is on scales significantly larger than those accessible to current spectroscopic redshift surveys .
We also find evidence for baryonic oscillations , both in the power spectrum , as well as in fits to the baryon density , at a 2.5 \sigma confidence level .
The large volume and resulting small statistical errors on the power spectrum allow us to constrain both the amplitude and scale dependence of the galaxy bias in cosmological fits .
The statistical power of these data to constrain cosmology is \sim 1.7 times better than previous clustering analyses .
Varying the matter density and baryon fraction , we find \Omega _ { M } = 0.30 \pm 0.03 , and \Omega _ { b } / \Omega _ { M } = 0.18 \pm 0.04 , for a fixed Hubble constant of 70 { km / s / Mpc } and a scale-invariant spectrum of initial perturbations .
The detection of baryonic oscillations also allows us to measure the comoving distance to z = 0.5 ; we find a best fit distance of 1.73 \pm 0.12 { Gpc } , corresponding to a 6.5 \% error on the distance .
These results demonstrate the ability to make precise clustering measurements with photometric surveys .