Created attachment 220 [details] Comparison of Bertini cascade elastic nucleon-nucleon cross sections with parametrization from Cugnon et al. It appears that the NP and (PP or NN) elastic scattering cross-sections are swapped. I'm referring to the elastic cross sections in G4CascadeNPChannel, G4CascadePPChannel and G4CascadeNNChannel. Attached are two plots, comparing the values in G4CascadeNPChannel and G4CascadePPChannel to the parametrization of Cugnon et al (NIM B 111 (1996), 215). The NP and PP curves seem to have been interchanged in geant4. This will affect predictions at low energies where elastic scattering is dominant.
Thank you for noting this; your observation and comparison with data is probably correct. These cross-sections have been present in the "Bertini"-inspired code for quite a long time. It is our belief (as the present maintainers) that they were put in this way, coupled to a variety of other numerical parameters, to provide a "reasonable" (define it as you wish) match with data. We have made a few attempts in the past year to replace the nucleon-nucleon cross-sections with more physically motivated values, the most recent being a detailed calculation of pp, np, and nn. When we have introduced those "corrected" values, we have found that macroscopic observables, such as neutron production in shielding or angular differential spectra, degrade substantially in comparison with reference data. A long term effort of the current developers is to make as much of the parametrization (including these final state cross-section tables) as "physically motivated" as possible. The seemingly tight coupling among all of the different components of the model has made that a rather difficult process, but your input is a welcome contribution.
