Skip to content

Latest commit

 

History

History
111 lines (86 loc) · 3.72 KB

WORK.md

File metadata and controls

111 lines (86 loc) · 3.72 KB
Raw

Work program for comparing various subtraction methods in identical setups

Ideally, each method should be studied by at least two people/groups

Code and results should be uploaded to a subdirectory of the directory "Comparisons/" on github

Signal samples

  • dijets (pt > 20, pt > 100, pt > 500), with UE off and massless particles
  • both full and CHS

The samples are in the usual location (/afs/cern.ch/user/p/puws2014/public/events or http://cern.ch/puws2014/events/).

lhc14-pythia8-4C-dijetsel20-noUE-nevsel1e5.pu14.gz
lhc14-pythia8-4C-dijetsel100-noUE-nevsel1e5.pu14.gz
lhc14-pythia8-4C-dijetsel500-noUE-nevsel1e5.pu14.gz

Each event in those files has at least one jet with |y|<2.5 and above the pt [GeV] indicated after the "dijetsel" tag. They are intended to be used respectively for the analyses with 20, 50 and 100 GeV pt cuts.

The files have been produced with a generation cut at 80% of jet pt selection cut. For validation purposes, there are files (on afs) labelled .res that indicate the average number of jets with pt>20 GeV, |y|<2.5 in each sample.

List of subtractors

  • Safearea
  • SafeNpC
  • ConstituentSubtractor
  • SoftKiller
  • Puppi
  • Cleansing
  • corrJVF
  • ......

Each one including trimming (Rsub=0.3, fcut=0.05) where possible

Observables

  • pt
  • mass
  • angularity/width/girth with alpha = 1

Pileup levels: 30, 60, 100, 140

particles and jet selection

particles: |y| < 4 jet definition: antikt R=0.4

jets: take the two hardest jets, then apply a selection of |y|<2.5, and pt > 20 (or 100, or 500), and study the impact of pileup on any jets that pass that selection. Pileup jets are matched to the hard jets with a deltaR = 0.3 criterion.

A separate study counts the jets above 20 GeV with |y| < 2.5. That count is only in events that have at one jet from the selection described in the preceding paragraph.

Note that in example03.cc prior to revision 107 (git rev-list --count HEAD) or hash d1b6590c2f2758d765c3... the number of jets was counted for all events.

How to compare (quality measures)

  • offset v. dispersion (use trimmed jet as ref when using trimming). There' a template offset-v-dispersion.gp gnuplot macro in the example/ directory for quick plotting
  • number of jets above 20 GeV as a function of npu

File format for results

Use the following file format for each subtractor/observable/sample (write out the "+-" to output):

 # comments  
 # npu    jet_ptmin  <DeltaO>     sigma_DeltaO    corr.coeff.     #_of_jets>20_GeV   matching_efficiency    <O_hard>           obs & method names 
   xxx    xxx        xxx +- yyy   xxx +- yyy      xxx             xxx +- yyy         xxx +- yyy             xxx +- yyy       #  e.g. pt_areasub

Template code

Look at example/example03.cc to see code that has the selection, matching and output as illustrated above. Run it for example with

./example03 -hard ../sample-events/lhc14-pythia8-4C-dijet50-nev20.pu14.gz \
            -pileup ../sample-events/lhc14-pythia8-4C-minbias-nev100.pu14.gz \
            -massless -npu 5 -nev 20 -jet-ptmin 20 > output.dat

Adapt the -jet-ptmin option depending on what you plan to study. There's also a -chs option for running with CHS type events.

To be considered at a later stage

boosted W, pt > 500 R=1 (trimmed mass) angularity alpha=1 tau_32, beta=2, 1-pass-kt-axis tau_21, beta=1, 1-pass-kt-axis, m > 40