Covariance propagation in R-matrix model fitting

This work is done for improving the current international standard cross section of nuclear reaction. The features of covariance propagation in R-matrix model fitting for 7Li, 11B and 16O systems are researched systematically with Code RAC, and the results about propagation of non-diagonal elements of covariance matrix are presented. It is found that in R-matrix model fitting, short-energy-range parameters result in relatively smaller covariance propagation coefficient (CPC), medium and long-energy-range parameters produce relatively larger CPC. Especially the medium-energy-range component of systematic error plays very important role in propagation of covariance. In the evaluation procedure of nuclear data both long-energy-range component (LERC) and medium-energy-range component (MERC) of systematic error should be considered in experimental data-base file. Furthermore, these conclusions are suitable for the similar model fitting in other science fields