An asymptotically model free method in statistical analysis of spatial point patterns

Abstract

In making inference on Spatial Point Processes often there it is the problem of knowing (or estimating) the distribution of relevant statistics. In the present paper it is proposed a method for estimating the distribution function of any strongly convergent statistic in a large class of Spatial Point Processes. The method is based on conditional simulation and doesn't depend on process' distribution. Asymptotically the statistics generated by this method of "autosimulation" have the same distribution as those generated by Montecarlo methods when the probability structure of the Process is completely known.