Multiple Coulomb Scattering

[freddieknets]

As currently implemented in Everest, MCS will be performed stepwise in order to take edge effects into account (see section 4.1 of https://cds.cern.ch/record/447077?ln=en). However, slicing up MCS is not a good idea as it changes the statistics slightly. Here is an example for 10 slices:

We can clearly see that the tails get emptied in favour of the core.

However, we cannot just ignore the edge effects either..

[freddieknets]

Maybe the issue is that MCS is essentially a statistical approach, while we try to address the edge effect in steps, i.e. point-like. Hence, we should try to address the edge effect statistically as well. Around the average path of the particle, we can draw bands (representing the sqrt(s)(1 + 0.038ln s)), potentially cut by 3 sigma (but probably there is no need to cut and we can address that statistically as well). By calculating the population of particles in the region within these bands that is outside of the collimator, we have a certain percentage/probability of trajectories that is outside. We could sample this with an extra dice.