David's suggestions about shifts: to do 1 as a general Trigger Shifter, 1 technical shift, and maybe next year (2023) do 1 week as a shift leader. (asked for Trigger Shift, not available until 2023, wrote for a technical shift, waiting for an answer).
the current MC production is Run 2, using the CMS 2018 UL workflow so the detector is Run2 detector not Run3! MC for Run3 is not available yet. The official production will be starting soon.
We received an email yesterday evening: "According to last week's MccM meeting(Oct 5th), PdmV announced that the campaign for Run3 sample production (Run3Summer22) is coming soon and hopefully by this week"
we are now comparing my dataset and Joe's because they are produced basically with the same card (the same card, but different versions of madgraph and pythia and small differences in the cards too (e.g. qcut etc)), mine are Nanoaodv9 and his are NanoAodv7.
so I have the ParticleNet branches that he hasn't (because v9 and not v7), but I can also have the previous btagger he used, so when I use the tag "--btagHbb" I have his btag and the comparison should overlap in this case, when I use the tag "--ParticleNet" I can see the improvement with Particlenet, different from Joe's result
We will have to re-produce the MC production from scratch for Run3!
But the procedure will be the same, just changing the gridpacks and the step1-7 config files to the CMS Run 3 production.
We won't have much Run 3 data to look at for a while, so we don't have much choice but to use Run 2 at the moment. Also, the new analysis will use the combination of Run2 + Run3 data (not Run 3 on its own). So we need both Run2 and Run3 MC.
Where is the detector information and energy information changing between Run2 and Run 3?
I think GEANT4 (the detector simulation) is run in Step2 (we have 7 steps to follow when producing NanoAOD) so for Run3, we'll need a new step2 config (with an updated detector simulation).
So what would be feasible probably by the end of my PhD is:
Re-doing the analysis for Run 2 using ParticleNet, Run3 preparation studies using MC only. We can compare the Run2 and Run3 MC, see what has improved, etc
Jacob asked me to produce these datasets:
When we're happy with the validation plots, I'd suggest to start the signal production with 150k events for each of the following 15 mass points (the "cross" shape I mentioned at the meeting):
plus one single squark sample for comparison: mH_60, mSquark = 2800
I launched them during the weekend the all failed, I need to understand why and fix this
crab exit code is : 50115 - cmsRun did not produce a valid job report at runtime (often means cmsRun segfaulted) (Check also the file Report.pkl which might have some more information, e.g., malformed job report xml or something).
So I need what is wrong with cmsRun
To do next:
-solving this issue
- a plot GenPart collection with pdgId==35