How do we store Monte Carlo truth for a generated event? It is a tree of particles and vertices? The following is a UML diagram constructed by visual inspection of the header files for McParticle and McVertex
The diagram is not compete, because it does not show any methods. However, they are all simple Get and Set methods, since these classes have no behavior: they are just a database. Also, it should show that the associated classes belong to other packages.
It is rather complex, and hard to set up properly, and also hard to traverse, due to the two classes involved. I think that the Gismo solution to this problem is better: there is only one class (GParticle) and it has methods to construct daughters that guarantee that it has been set up properly.
Proposal:
Merge McVertex into McParticle, eliminating the cross-connections. Examine methods or constructor types that set up the proper associations.
Toby Burnett 28-01-2002 09:13 -0800