Un'applicazione del metodo del bootstrap ad un problema fisico di propagazione degli errori

Abstract

This paper deeals with an applications of the bootstrap method to the error analysis when the data come from both experiments and Monte Carlo simulations. The example concerns with X-ray microanalysis directed to the simultaneous determination of both composition and thickness of thin supported films. The bootstrap allows one to evaluate how experimental errors coming from simulations propagate themselves into the quantities of interest. The booststrao is also used in a predictive way, i.e., to asses the accuracy of the final results from the precision of the data, before any actual measurements or simulations.