[Rivet] Normalisation of histograms in MC_Z* analyses

[Christian Gumpert]

Dear RIVET experts,

when running some of the MC_Z* analyses, I observed differences in the 
way histograms are normalised. E.g. in the analysis MC_ZJETS all 
histograms inherited from the MC_JetAnalysis are normalised using 
Scale(hist, crossSection()/sumOfWeights()) which in my opinion seems to 
be the correct approach. On the contrary, the two histograms Z_jet1_dR 
and Z_jet1_deta are normalised using normalize(hist, 
crossSection()/picobarn) which can be quite misleading in my opinion. 
The cross section passed to RIVET usually corresponds to the full 
generated sample of events. However, the histograms with the Z 
properties are only filled in a phase space with a Z mass requirement 
(and thus will have in general a smaller cross section). In my opnion, 
it would preferable to have a standard way of normalising histograms 
using the Scale(hist, crossSection()/sumOfWeights()) approach. THis 
problem basically affects all MC_Z* analyses while the MC_W* analyses 
look fine.
Also, the description of MC_ZJETS (data 
<https://rivet.hepforge.org/trac/browser/data?order=name>/anainfo 
<https://rivet.hepforge.org/trac/browser/data/anainfo?order=name>/MC_ZJETS.info 
<https://rivet.hepforge.org/trac/browser/data/anainfo/MC_ZJETS.info?order=name>) 
is a bit outdated as e.g. the Z mass distribution is not available.

Cheers
Christian
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://www.hepforge.org/lists-archive/rivet/attachments/20150416/d6422207/attachment.html>