Difference: SherpaJetParameters ( vs. 1)

• Sherpa 1.1.3 pp → jjj (ie. 3 partons, quark or gluons)
• parton p_T≥20 GeV/c, |η|<2.5
• Sherpa's default JetFinder sqr(QCUT/E_CMS) D with different values of QCUT and D. This is a k_T algorithm (Sherpa 1.2.0 allows other algorithms, including anti-k_T)
• Did not include ΔR cut for these tests, but this can be done using
  

  "Calc((sqr(DEta(p[0],p[1]))+sqr(DPhi(p[0],p[1]))>=0.36)+
        (sqr(DEta(p[0],p[2]))+sqr(DPhi(p[0],p[2]))>=0.36)+
        (sqr(DEta(p[1],p[2]))+sqr(DPhi(p[1],p[2]))>=0.36))" 93,93,93 2,3
  

  (spacing, included here for clarity, is not allowed in Sherpa's Run.dat) which requires at least two of the three ΔR between partons to be ≥0.6, ie. ΔR≥0.6 between one parton and both the other two.

• varying Qcut: Q_cut=5, 10, 20, 40 with D=1
• varying D: D=0.3, 0.5, 1, 1.5 with Q_cut=20

Comments:

• The PDG id is of the 3 partons per event. Do we really expect ~80% of the partons to be gluons, rather than quarks?
• Jet plots are for jets with p_T≥20 GeV/c and |η|<2.5 and at least three such jets.
• I plot the smallest, middle, and largest ΔR and Δp in the event. The smallest shows the cut-off due to generator parameters. The middle shows what we would cut on with a ΔR cut. Note the different jet ΔR/Δp samples don't have the same normalisation (especially for Q_cut=40,D=1), due to the 3-jet cut.
• The effect of varying Q_cut and D shows in the parton separations, but not much in the reconstructed jets. Is the effect really so washed out?
• After hours of LXR/dOxygen searching, I finally found how to run a jet finder on generator output. You need to set a flag to specify the input type: this is done automatically for other input types, but not for generator input. For the record, here's the magic incantation:

  from JetRec.JetRecFlags import jobproperties as jetFlags
  jetFlags.JetRecFlags.inputFileType = "GEN"
  from JetRec.JetGetters import make_StandardJetGetter
  collname = make_StandardJetGetter('AntiKt', 0.4, 'Truth').jetAlgorithmHandle().JetCollectionName

-- TimAdye - 2009-12-15