Systematic Sharing and Reconstruction of Monte Carlo Output Distributions for Metrological Uncertainty Evaluation

Abstract

The Guide to the expression of uncertainty in measurement (GUM) sets the “transferability” as a necessary feature for the method of expressing the uncertainty. If the analytical GUM method is used, few easy-transferable pieces of information are enough to share the knowledge about the measurand. However, in case of the stochastic Monte Carlo (MC) technique is used according to the supplement 1 to the GUM (GUM-S1), the provided transferability guidelines are unclear and require sharing a large amount of data. In this paper, a complete system is proposed for sharing the output of MC and resampling from its distribution. The system suggests a small-file -size electronic sharing format; the shared data are processed, and random values are redrawn by the synergy of three software modules which work in a systematic manner. Mathematically, the proposed technique reconstructs the cumulative distribution function (cdf) of MC’s output distribution depending on a continuous piecewise linear approximation and uses it for resampling. The eligibility of the proposed technique was confirmed via successfully reconstructing the cdfs of two familiar distributions and resampling from them.