gpre2¶
One of the Data Preproccessing Layers that is mostly not useful for anything not Particle Physics related, since it assumes the Input to be a list of lists of Momentum 4 vectors. Is Outdatet, buggy and only here for consistency
Arguments
- gs: The Number of Particles used, will become the Number of Nodes (Graph size)
- numericC=10000: A Numerical Constant that will be sometimes used to keep things finite
Produced Features
- eta-mean(eta)
- phi-mean(phi)
- ln(pt)
- ln(E)
- -ln(pt/sum(pt))
- -ln(E/sum(E))
- sqrt((eta-mean(eta))**2+(phi-mean(phi))**2)
- iszero (a flag to filter out missing(zero) particles)