Bootstrap methods applied to indirect measurement

Abstract

A biased bootstrap technique is presented to obtain robust parameter and measurement estimates. Moreover, the estimation of a measurement probability density function (pdf) using classical bootstrap techniques is presented as our final goal. Most of the time, large scale repetition of an experiment is not economically feasible, the Monte Carlo method cannot be used for uncertainty characterization and bootstrap methods are proved to be a potentially useful alternative. The measurement characterization is driven by the pdf estimation in a non-linear non-Gaussian case and with limited observed data.