MUSE: computational aspects of a GUM supplement 1 implementation

摘要

The new guideline GUM Supplement 1--Propagation of Distributions Using a Monte Carlo Method (GS1) is currently published by JCGM/WG1. It describes an approximate method to calculate the measurement uncertainty in nearly all areas of metrology. In this way it overcomes the various limitations and drawbacks of the uncertainty propagation detailed in GUM. However, GS1 demands a software implementation in contrast to the uncertainty propagation. Therefore we have developed a software tool called MUSE (Measurement Uncertainty Simulation and Evaluation), which is a comprehensive implementation of GS1. In this paper we present the major computational aspects of the software which are the sampling from probability density functions (PDFs), an efficient way to propagate the PDFs with the help of a block design through the equation of the measurand and the calculation of the summarizing parameters based on these blocks. Also the different quality measures which are in place during the life cycle of the tool are elaborated.