On the possibility of a multidimensional kinematic information analysis by means of nearest-neighbour estimations of dimensionality

Abstract

Abstract A new method of analysis of multiple final states in high-energy physics is described. It can be recommended for the preliminary analysis of events with high multiplicity and is able to indicate the existence of various bound states. The method allows a visualization of multidimensional data, this being particularly important in the dialog mode of data handling. New algorithms based on the estimation of the multivariate probability density are used for the determination of correlation dimensionality. These algorithms are more suitable and precise as compared to the earlier suggested ones, due to the natural scale introduced and the account of the distribution function of the correlation integral. The serviceability of the algorithms is checked in a series of Monte Carlo simulations.