Prepare next production

cp3-llbb / TTAnalysis

TTbar analysis for CP3 llbb framework

0 stars 4 forks source link

Prepare next production #30

Closed swertz closed 8 years ago

swertz commented 8 years ago

Corrected muon isolations
New CSVv2 working points
Moved to a single config file (as HHAnalysis)
Moved to 76X
Use latest JEC
Add JEC/JER systematics

Anything else (no merging until we're sure)?

blinkseb commented 8 years ago

Great thanks!

I'd prefer to keep two different configuration file. It's much easier to handle when launching on crab than to remember that you have to edit the file before sending. If you want, you can move the command part to a common python file, and call it from the MC and DATA cfg.

swertz commented 8 years ago

There is no way to adapt the config on-the-fly for data or MC from GridIn?

blinkseb commented 8 years ago

Not currently :(

swertz commented 8 years ago

I've updated the PR. In the end I kept two files as before...

Should we think about adding the Top Pt reweighting and the trigger scale factors/efficiencies at this stage or try to include them at the plotting stage? I think it'd be more flexible to have this included later...

blinkseb commented 8 years ago

I'm still a bit uncomfortable wrt top pt reweighting. I'm not sure would should apply it ( https://twiki.cern.ch/twiki/bin/view/CMS/TopSystematics#pt_top_Reweighting )

Concerning trigger efficiencies, we have SFs for Double Muons already in the framework, but we need to store them in our tree if we want to use it later. There's nothing for electrons and it has been clearly mentioned that we have to derive our own efficiencies / SFs :(

swertz commented 8 years ago

All right, let's keep Pt reweighting on hold, this is something that we can add later if needed...

For the trigger I was thinking of somehow incorporating these results: https://indico.cern.ch/event/487265/contribution/3/attachments/1220074/1783203/Talk_Till_mumu.pdf (where they have MuMu, EE and MuE) The ID/iso working points they use are different than what we've been doing, but they show a significant discrepancy between data and MC trigger efficiencies. So either we change our selection and ask them for their numbers, or we try to measure trigger efficiencies ourselves... Either way these is stuff we can add in the plotting or skimming step, so I think we can move on for now...

blinkseb commented 8 years ago

Ah nice, good to keep that in mind so :) I'm merging