Open GilesStrong opened 1 year ago
I agree with making it optional. Numerically, it should be an important effect only for large objects, but users interested in small objects may be happy to switch it off. Are there implications also on speed of execution? I presume not a lot, because Eloss can be computed analytically from formulas readily available e.g. in the PDG, but I might be forgetting some subtlety.
@giamman if there are analytic formulas, then the time increase should be negligible, I think.
Well one could do that in a discrete stochastic approach (à la GEANT) or with an analytical continuous approach, and I strongly believe that we should go analytic whenever we can, given TomOpt's scope.
@MaximeLagrange has shown that energy loss during propagation can be significant, and can lead to large differences in the scattering distributions. Currently, muons keep the same momentum throughout their flight. We should add in a simulation of this E-loss in the muon scattering. However, since this E-loss will also affect inference, and may complicate the optimisation, we should make the E-loss be optional.