Programas del algoritmo de clasificacion de poblaciones normales trivariantes a partir de la entropia de mezcla

Abstract

The complete FORTRAN codified programs of the segregation algorithm described in Alcobé (2001) are described. The classification algorithm is applied to study the trivariate velocity distribution of stars from the star catalogs, which can be locally approximated by a superposition of two or more normal components. An auxiliar sampling parameter P (such as the velocity module referred to a specific point, the absolute value of one peculiar velocity component alone, the distance to the galactic plane, etc.) is introduced in order to define the sample boundaries. The sampling paramenter must induce a hyerarchical incorporation of stars to the population components, in the sense that the greater the P value, the greater the number of stars in each component. For a fixed P, a sample S(P) is drawn from the global catalog. Depending on the sampling parameter the population entropy H(P) of a two-component mixture is computed from the mixing proportions. The purpose is to find the optimal P value in order to maximize H(P). Then the algorithm is used recursively in order to segregate a global sample in more than two populations. Moreover, for each subsample S(P), the goodness of the superposition approximation is evaluated by reconstructing the sample central moments up to fourth-order from the population parameters. A chi-square test, taking into account the sampling distribution moments, is evaluated to measure the fitting error. For subsamples S(P) a total accordance between the minimum chi-square and the maximum population entropy H(P) is produced.