On the Verification and Interpretation of the Evaluated Data Covariances

Summary: • Input experimental data (results of measurements and their covariances) predetermine the evaluated data and their covariances calculated by the LSM for any linear model function; • As follows from the conservation laws relative decreasing (increasing) uncertainties of the evaluated data leads to pumping uncertainty information into the off-diagonal covariances • strict relationships between input experimental data and output evaluated data, restrictions imposed to the covariances of the experimental errors provide verification both the final and intermediate results of calculations • Integral uncertainties of the random vector (trace and determinant of covariance matrix) provide additional uncertainty information for the evaluated data • As seen today the adequate processing of USU (inconsistent experimental data) is a key for getting reliable uncertainty information for the evaluated data