Synopsis of presentation for RCM no. 3

A significant capability has been added to the SAMMY R-matrix code: Matrix algebra manipulation was used to determine what changes would be needed in the conventional definition of off-diagonal data covariance matrix (DCM) in order to produce agreement between results from two supposedly identical procedures. Within the SAMMY code, this new definition of DCM has been implemented for use with any parameter for which partial derivatives can be calculated. Implementation of the PUP option is a major step towards satisfying the long-standing desire to take all possible experimental uncertainties into account during an evaluation. 6Li(n,t), 6Li(n,n), and total cross section comparisons were made to compare various methods of treating the normalizations (as contributors to DCM or as fitting parameters). Preliminary results gave no surprises: with large uncertainties on the input normalizations, large differences in calculated cross sections (as high as 20%) could be found when comparing various treatments of the normalizations. In the near future, these comparison studies will be explored further using realistic input values corresponding more closely to those used in the EDA runs. The underlying causes for differences between the output of the R-matrix codes EDA and RAC will not be fully understood until relevant features of these codes are compared individually. Comparison tests confirmed that equivalent values of R-matrix parameters led to equivalent values for theoretical cross sections, provided EDA used non-relativistic kinematics. Iterations should be permitted, with only one parameter initially. Further tests would continue to add refinements one at a time: (a) off-diagonal data covariance matrix corresponding to normalization only, with one data set, one parameter, and no iteration; (b) more than one parameter; (c) with iteration; (d) more than one data set; (e) including other types of correlations. At every stage, the three codes must use exactly equivalent input. Carrying out detailed comparison tests of this nature will give us better insight into some of the reasons for differences between RAC and EDA results