Run3 Trigger Scale Factors

AE Introduction

ok, first you need to modify the config file from here: https://gitlab.cern.ch/cms-cmu/coffea4bees/-/blob/master/analysis/metadata/trigger_weights.yml?ref_type=heads#L9-11

  • use_vectorized is the new trigger weights (vectorized because the old code was still doing loops)
  • tagger is the tagger we use in Run3. (Marina has 3 types of taggers, so I just copied them)
  • make_classifier_input is where you store the new friendtrees.

then you will need to run something like this:

./run_container python runner.py -p coffea4bees/analysis/processors/processor_trigger_weights.py -d TTToSemiLeptonic -y 2022_EE -m coffea4bees/metadata/datasets_HH4b_Run3/ -c coffea4bees/analysis/metadata/trigger_weights.yml -op output/ -o test.coffea

This code has not been validated, so it could be that there are some bugs. I just visually look at the weights and see that they are between 0-1.

Logs

12 February 2026 Thursday

  • 3b looks OK python -i coffea4bees/plots/iPlot.py --combine coffea4bees/Run3TriggerSFs/trigWeight_TTbar_wHLT.coffea coffea4bees/Run3TriggerSFs/trigWeight_data.coffea -m coffea4bees/plots/metadata/plotsAll_TTCR_3b.yml
plot("selJets_ll.n",region=sum,yscale="linear",cut="passElMu",norm=0,rebin=1,doratio=1,year="Run3")

2026-02-12_14-10-27_screenshot.png

  • 4b looks off python -i coffea4bees/plots/iPlot.py --combine coffea4bees/Run3TriggerSFs/trigWeight_TTbar_wHLT.coffea coffea4bees/Run3TriggerSFs/trigWeight_data.coffea -m coffea4bees/plots/metadata/plotsAll_TTCR_4b.yml

2026-02-12_14-09-42_screenshot.png

  • OK will now try to fit JCM with floating 4b TT to see if this helps at all…

11 February 2026 Wednesday

  • Adding threeTag and fourTag plots to diplepton hists
  • Changed selection to be >2 btags
  • Test job not running… taking forever to preprocess files

-/store/user/smurthy/XX4b/2025_Run3_skims/TTToHadronic_2023_BPix/

  • When I eos ls I see:

error: MGM root://cmseos.fnal.gov not online/reachable

  • Trying again later… more files can be accessed but still halting… now running!
  • Adding the three and four tag selection to dilepton plots
  • Running on all TTMC… finished
  • running on all Data… finished!

31 January 2026 Saturday

  • Running all the TTbar samples … Had to go to 8GB of condor memory
  • 116,895 events/s total (156244645/1336.629465341568)
  • Now have ttbar nJets… on to fitting JCM

30 January 2026 Friday

  • Looks like the ttbar files are missing the xs
  • [X] Eta cut in ht defitinos 2.4 -> 2.5
  • Adding xs from Marina's AN https://cms.cern.ch/iCMS/jsp/db_notes/noteInfo.jsp?cmsnoteid=CMS%20AN-2023/151
  • Re-Running 2L2Nu with xs… norm is much better !
  • Updating friend trees … fixed btagging and jet eta cut
  • Running 2L2Nu on condor outputing to eos
  • [X] Add lepton plots
  • [X] Add seljet plots
  • [X] Add Mll
  • [X] Add emu selection
  • [X] Add MeT

  • Remade hists data and TTbar

    mumu: 2026-01-30_22-22-24_screenshot.png emu:

    2026-01-30_22-22-50_screenshot.png

  • Probably good enough to move on!
  • Adding quarkonina and Z vetos to mumu
  • Pushing … Fixing the CI… updating the cutflow…
  • Make friend trees for SemiLeptonic and Hadronic

29 January 2026 Thursday

  • [X] Check cutflows of TTbar skims
    • two tag data is in the 2L2Nu skims.. about half the total dataset …
  • Running on condor with out trig weights
  • python coffea4bees/analysis/printCutFlow.py -i coffea4bees/Run3TriggerSFs/test_trigWeight_TTTo2L2Nu.coffea -p TTTo2L2Nu -e 2023_preBPixMC
  • [X] Turning on the dilepton selection
  • Fix to get the ttbar plotting region working plot("tagJetsdilepttbar.eta",region=sum,yscale="log",cut="passDilepTtbar",norm=0,rebin=1,doratio=0,year="Run3",process="TTbar")
  • [X] Add plots relevant for trigger turn ons
  • Pushed
  • Running w/Trigger weight TTTo2L2Nu condor 14,424 events/s total (10153441/703.9495573043823)
  • Running w/o Trigger weight … 17,281 events/s total (10153441/587.5605149269104)
  • Ran data: 55,313 events/s total (69615572/1258.5722591876984)
  • Looks like the ttbar weights are way off ! ..

28 January 2026 Wednesday

  • Making the trigger weight with PNet
  • Running all on condor… outputing to eos… Condor fails.. .running locally … works
    • I think the condor job fails because of memory (only ran with 2GBs…)
  • Testing the trigger weights ./run_container source coffea4bees/scripts/weights-trigger-analysis-Run3.sh
  • Running all the TTMC with trigger weights … condor taking forever… too small chunk size.. now 100k
  • Condor still fails with 100k.. THink its b.c memeory 2GB -> 4GB rerunning … still problems now 6GB

Plan:

  • [X] TTTo2L2Nu without trig SF
  • [X] TTTo2L2Nu w/trig SF
  • [X] Data 3tag ?

27 January 2026 Tuesday

  • Starting
  • First attempt … TTTo2L2Nu 2022 test job .. failed
  • Pinged AE… Now fixed
  • Running all years, no test
  • Finshed… Need to put on eos

Backlinks