[Rivet] A comment on InvMassFinalState

[Frank Siegert]

Hi Adam,

Adam Davison, Wednesday 10 June 2009:
> I've been trying to use the InvMassFinalState class but the interface
> seems a bit difficult. It returns all particles that could be used to
> make, for example a Z, excluding duplicates. This means that if there's
> more than one candidate in an event, it can do some wierd things, like
> return a list of 3 or 5 particles. This makes it quite hard to apply
> any further cuts to the candidates, like pt, since in order to identify
> which particles are actually matching pairs, you have to loop over
> everything again anyway.

Yes, I agree that it is awkward, but when I used it (for the ZFinder) I 
wasn't brave enough to replace it. Unfortunately there are a few places 
which still seem to use it, and I didn't want to validate them after such 
a change:

$ grep InvMassFinalState $(svn ls --recursive) | cut -d ":" -f 1 | uniq
ChangeLog
include/Rivet/Makefile.am
include/Rivet/Projections/InvMassFinalState.hh
src/Analyses/CDF_2006_S6653332.cc
src/Analyses/CDF_2008_S7541902.cc
src/Analyses/CDF_2008_S8095620.cc
src/Analyses/D0_2008_S6879055.cc
src/Projections/InvMassFinalState.cc
src/Projections/Makefile.am
src/Projections/ZFinder.cc

> The ZFinder, which is the only place in Rivet I saw this being used
> returns if the size of the list isn't 2. Which unless I've
> misunderstood, potentially throws away events with good Z candidates
> in.

Yes, this is a deliberate choice. If we wanted to pick one of the Z 
candidates for each event, we would introduce a bias (for example towards 
a known Z mass), so better reject that event. That said, I don't think it 
will make much of a difference, because the number of events where enough 
leptons pass cuts and form multiple Z's is quite certainly negligible.

Frank