JetToolbox_94X including fix in ECF and remove jettoolbox.root

cms-jet / JetToolbox

Python framework for configuration of jet tools via the jet toolbox.

https://twiki.cern.ch/twiki/bin/viewauth/CMS/JetToolbox

7 stars 36 forks source link

JetToolbox_94X including fix in ECF and remove jettoolbox.root #63

Closed alefisico closed 6 years ago

alefisico commented 6 years ago

Fixing the ECF to include parameters of the algorithm and reduce time by not calculating all the variables N2 and N3
changing associateTask parameter to saveJetCollection. By default it run the jetToolbox and not save the jetCollection in the edm rootfile (to avoid the creation of jettoolbox.root). If needed (saveJetCollection=True) the pat:Jet collection is saved.
adding meaningful test (using one of the CMSDAS examples)

kpedro88 commented 5 years ago

@alefisico how is it intended to work if you want to calculate more than one kind of ECF? (The previous settings replicated the central miniAOD settings, which have both beta=1 and beta=2.)

Also, ECF N3 is still stored even if it wasn't computed: https://github.com/cms-jet/JetToolbox/pull/63/files#diff-f1fc8e7a8a193e13cf51904bafa14b26R993 (and further down)

alefisico commented 5 years ago

Hi @kpedro88 this was a request from a user, after also discussing with @rappoccio. The problem is that it takes quite a lot of time to calculate ECF N3 and calculating it by default, even if the user will not use it (which Sal mentioned that most of the analysis are not using it), was pointed out to not be very useful and computationally expensive. However, I tried to keep it as useful as possible. Therefore, if you turn on the option ecfN3=True, it will calculate ecfN3. Also, I know that the variable is stored even if you dont turn on the option. It was just part of this quick fix. If you want to modify it, please be my guest.

kpedro88 commented 5 years ago

@alefisico I understand the motivation behind disabling N3 by default. But it seems like it would at least be useful to be able to calculate both nb1 and nb2. I will think about how this could be supported flexibly.