Shape of beam source

What if you sample random points in the cylindrical volume and track each photon for the mean path length? Would that suffice? If yes, one option is to use GPS and define the volume shape and dimension in macro. Like this?