The updated photon conversion code has changed the behavior found in 10.4. In BetheHeitlerModel (and now G4PairProductionRelModel in 10.6) the correlation between the e+ and e- cos(theta) from the photon direction has changed fairly significantly. This change is most clearly seen by looking at how cos(theta) is chosen: 10.4: G4BetheHeitlerModel.cc:270-274 shows one random value is picked from the angular distribution which is applied for each daughter. 10.5 and 10.6: G4ModifiedTsai.cc:129-135 shows a random value is picked separately for each daughter, therefore decorrelating the two. This difference is most plainly seen by plotting KE(e-)*cos(e- angle from photon) vs KE(e+)*cos(e+ angle from photon) which goes from being a diagonal line (10.4) to a 2D distribution (10.5 and 10.6). Is this an expected change in behavior solving a previous bug? I was unable to find a reference to any change here other than code cleaning in the release notes. Best, Michael
Hello, since Geant4 10.5 we have introduced so called 5D model of gamma conversion which handles correctly nuclear recoil and final state e+e-. This model in Geant4 10.6 is included in Option4 (EMZ) physics. We also tried update Tsai angular generator for e+e- using comparisons with this 5D model. It cannot be as accurate as the 5D model. If this question is important for you, please, try to make a study of 3 existing generators, where 5D should be the reference.
Hello, Tsai generator is based on parameterisation of original Tsai angular distribution of bremsstrahlung. Extension of Tsai angular distribution to sampling of e+e- pair production is done using comparison of gamma conversion to e+e- pair with 5D model, which is fully based on QED differential cross section. In the current version of Tsai generator used for e+e- production there is less inaccuracy of the final state.
It is not a bug but request to explain what was changed in Geant4.
