search

key4hep / EDM4hep

Generic event data model for HEP collider experiments
https://cern.ch/edm4hep
Apache License 2.0
25 stars 35 forks source link

`RecoParticleVertexAssociation` duplicates relation information #321

Closed tmadlener closed 3 months ago

tmadlener commented 3 months ago

Somewhat related to the discussion in #320 but then again separate enough to warrant a dedicated issue. The currently existing RecoParticleVertexAssociation is in principle superfluous, as it connects Vertex and ReconstructedParticle; a connection that already exists with internal relations of the two datatypes

https://github.com/key4hep/EDM4hep/blob/bd9f45039149781a9c54799a6d03df56b1b5ed49/edm4hep.yaml#L680-L687 ReconstructedParticle: https://github.com/key4hep/EDM4hep/blob/bd9f45039149781a9c54799a6d03df56b1b5ed49/edm4hep.yaml#L591-L592 Vertex: https://github.com/key4hep/EDM4hep/blob/bd9f45039149781a9c54799a6d03df56b1b5ed49/edm4hep.yaml#L563-L564

Or in pictures: image

Historically the existence of this associations seems to be caused by the fact that something in iLCSoft in either the ILD or CLIC reconstruction(?) creates an LCRelation between a Vertex and ReconstructedParticle, which then required this association during the conversion between EDM4hep and LCIO. The main culprit is most likely LCFIPlus and another one is here.

I am currently not entirely sure whether it's easily possible to remove this.

gaede commented 3 months ago

We (BH, JMC, FG) suggest to keep it, following our proposal in #320 to not have the link to vertex in the RecoParticle.

tmadlener commented 3 months ago

Keep this regardless of the decision in #320, because we need this to keep information about the start vertex.