Systematics for legacy 2017

[simonepigazzini]

We should prepare the systematic workflow to handle the legacy samples, starting from 2017. Probably it makes sense to start with with diphoton systematics where probably it still makes sense to have a separate file for legacy 2017 (that might eventually be merged with the one we are using now once the usage of ReReco17 fades out).

The main points on the list is of course to get the EGM energy correction right and the GTs.

@rchatter

[simonepigazzini]

https://github.com/cms-analysis/flashgg/blob/45a6a51f166d4a4746c6cea863e897011f342e5f/MetaData/data/MetaConditions/Era2017_legacy_v1.json#L4

Mc is fine, the recommended one for data is v28 (https://twiki.cern.ch/twiki/bin/view/CMS/PdmVRun2LegacyAnalysisSummaryTable). I think v17 is still fine for microAOD production, we will update to v28 with any coming PR related to this.

@alesauva

[simonepigazzini]

@rchatter can you setup the scale and smearing for lagacy 2017? I would go again with a separate diphotonSystematics config file that is then specified in the meta conditions.

@threiten @edjtscott I think we need at least scale and smearings to go ahead with the differential pho ID corrections

[rchatter]

Hi Simone,

Yes I have everything to push this diphoSyst file(privately provided to a few people already) as you asked. Just want to explicitly test it on the DoubleEG Legacy17 dataset before committing[unfortunately there seems to have been an issue in the MuAOD production]. The reason for this extra caution is that nominally for all of CMS we changed the scale correction file format and the corresponding EGamma scale correction class to deal with it (This was a long standing request from multiple people, to migrate the scale correction to DB and this change in format is the first step). All future Legacy scale correction files will be in the new format. However for continuity in HGG I also produced the file in the old format[like we currently have for the 3 years]. Hence also checking that it works fine with the dipho data.

Rajdeep

[simonepigazzini]

Thomas is going to produce tnp ntuple for the pho ID correction training, I guess we can use those to have the validation (he will share them with you in case you want to take a look), so I think it is best if you can make a PR with the changes needed to run with the old format and we will migrate everything together once the new format is available for all years. What do you think?

[edjtscott]

I think we can almost close this issue @simonepigazzini since @rchatter made the PR and it has been merged. However I notice that this line in the legacy meta conditions file was not changed - it should be updated right? If so that should be fixed first.

[rchatter]

Nice catch! Completely forgot to update the metaconditions. Shall be cross-check if the updates needed for UL17 JEC/JER and also update the B-tagging SFs on top of this and then close the PR?

[edjtscott]

Yes good idea, we can keep this open until those are confirmed / sorted

[rchatter]

I believe we can close this now right?

[youyingli]

Yes, all systematics for UL2017 have been confirmed!