NEURAL NETWORK METHOD FOR GALAXY CLASSIFICATION : THE LUMINOSITY FUNCTION OF E/S0 IN CLUSTERS
Emilio Molinari, R. Smareglia · 1998
Abstract. We present a method based on the non-linear be-haviourofneuralnetworkfortheidentificationoftheearly-typepopulation in the cores of galaxy clusters. A Kohonen Self Or-ganising Map applied on a three-colour photometric catalogueof objects enabled us to select in each passband the ellipticalgalaxies. We measured in this way the luminosity function ofthe E/S0 galaxies selected in this way. Such luminosity func-tions show peculiarities which disfavour the hypothesis of itsuniversality often claimed for rich clusters and that can be re-lated to the past dynamical history of the cluster as a whole.Keywords: galaxies:clusters:general–galaxies:ellipticalandlenticular – galaxies: luminosity function – galaxies: photome-try – methods: miscellaneous1. IntroductionBy now the determination of the luminosity function (LF) ofgalaxies in clusters has reached the key point of becoming atrue challenge to the supposed universality of its form (as inColless 1989). Different environment and cluster density con-ditions are clearly related to different morphological mixtures(Dressler 1980; Whitmore & Gilmore 1993) and therefore theevidence provided by the whole set of LF shapes in the Virgocluster (Binggeli et al. 1988) claims that one luminosity func-tion for all types of clusters can hardly be supported. The clearpresence of two distinct classes of normal and dwarf galaxiesand their different dynamical and luminosity evolution can leadto the search of substantial differences from one LF to the other.Unfortunatelysystematicsurveysspanasmallmagnituderange,going from (M