Paper detail

Order statistics of the early-type galaxy luminosity function

We apply order statistics (OS) to the bright end ($M_r < -22$) of the luminosity distribution of early-type galaxies spectroscopically identified in the SDSS DR7 catalog. We calculate the typical OS quantities of this distribution numerically, measuring the expectation value and variance of the $k^{th}$ most luminous galaxy in a sample with cardinality $N$ over a large ensemble of such samples. From these statistical quantities we explain why and in what limit the $k^{th}$ most luminous galaxies can be used as standard candles for cosmological studies. Since our sample contains all bright galaxies including the brightest cluster galaxies (BCG), based on OS we argue that BCGs can be considered as statistical extremes of a well-established Schechter luminosity distribution when galaxies are binned by redshift and not cluster-by-cluster. We presume that the reason behind this might be that luminous red ellipticals in galaxy clusters are \em not random \em samples of an overall luminosity distribution but biased by the fact that they are in a cluster containing the BCG. We show that a simple statistical toy model can reproduce the well-known magnitude gap between the BCG and the second brightest galaxy of the clusters.