root-project
TGenPhaseSpace weight normalization
Check duplicate issues.
- [X] Checked for duplicates
Description
The event weight obtained from TGenPhaseSpace::Generate() that is used to get a flat distribution is not normalized to 1. The number rather seems to correspond to the square of the portion of the phase space occupied by the kinematically allowed final states, e.g., for three massless particles, the shape is a triangle that occupies 1/2 of the m_{12}^2 and m_{23}
^2 ranges, and the normalization is indeed 1/4 (instead of 1). Maybe this is a feature and not a bug, but it should be documented in the TGenPhaseSpace::Generate() method.
It would also be useful to have a method that returns the correct event weight. At the moment, in order to get a correctly normalized weight, one can run a Monte Carlo to obtain an overall normalization of the distribution for chosen masses and then apply this normalization as a correction on individual weights of generated events.
Reproducer
A simplest example can be the tutorial macro: PhaseSpace.C
ROOT version
Present in v6.30 but the behavior was the same in v5 and possibly before.
Installation method
built from source
Operating system
Probably all systems (tested on Linux Centos 7, Centos 8, Rocky 9, macOS Sonoma 14)
Additional context
No response
