FxFx merging in tt+(up to) 2jets with aMC@NLO in PYTHIA8 J. Fernandez (with many thanks for the comments and prompt answer from Josh Bendavid) Setup and settings (I) • pp>t t~ +0,1,2j @13TeV generated in separated multiplicity samples (100k events each) using aMC@NLO v2.1.1 (No PS): A. B. With FxFx merging option (ickkw=3) With NO merging option (ickkw=0) • Same options for the rest: – Running options: • • • • • • – Jets options: • • • • 8th April 2014 cteq6_m = pdlabel ! PDF set average = event_norm ! Normalize events to sum or average to the X sect. F = fixed_ren_scale ! if .true. use fixed ren scale F = fixed_fac_scale ! if .true. use fixed fac scale 4 = maxjetflavor PYTHIA8 = parton_shower ! PS set to OFF in any case 1 = jetalgo ! FastJet jet algorithm (1=kT, 0=C/A, -1=anti-kT) 0.7 = jetradius ! The radius parameter for the jet algorithm 10 = ptj ! Min jet transverse momentum -1 = etaj ! Max jet abs(pseudo-rap) (a value .lt.0 means no cut) J. Fernández 2 Setup and settings (II) • Hadronized with PYTHIA8 (interfaced through CMSSW_7_1_0_pre7) – Script provided by Josh Bendavid – Key parameters: • • • • • • • 8th April 2014 'JetMatching:coneRadius = 1.', 'JetMatching:qCut = 30.', 'JetMatching:qCutME = 10.', 'JetMatching:nQmatch = 4', 'JetMatching:nPartonsNow = 0', # 1 for 1j, 2 for 2j 'TimeShower:nPartonsInBorn = 2', # 3 for 1j, 4 for 2j 'JetMatching:nJetMax = 1‘ #2 to merge 0j+1j+2j J. Fernández 3 Some numbers Process NNLO Xsec (pb) Frac Neg weights Matching Effective Matching Effective Eff 1jmax Xsec(pb) Eff 2jmax Xsec(pb) Gen CPU time* tt0j FxFx 693.0 0.25 0.26 244.7 0.26 244.7 1h tt1j FxFx 746.4 0.39 0.50 485.2 0.27 262.9 6.5h tt2j FxFx 600.7 tt0j NoFxFx 652.3 0.25 - - - - <1h tt1j NoFxFx 715.4 0.40 - - - - 5.6h tt2j NoFxFx 656.4 0.41 - - - - >24h >24h *Generation time on a dedicated 32-core multiCPU Xeon. In addition grid process time must be added… - Xsec values do not match those from Table 5 of arXiv:1405.0301v1 (same ptj value?) - Many problems running t t + 2j NNLO: 359 subprocesses, some (residual) subprocesses getting stuck over and over, and crashing after 16h… (P0_gg_ttxgg/GF62, P0_gg_ttxgg/GF61, P0_gg_ttxgg/GF22). Need them all successfully completed in order to merge properly the full result… J. Fernández 8th April 2014 4 Caveats for plots • Gen Jet: any Jet AK4 • Radiated Jet: any Gen Jet not matched (dR<0.1) to any top decay product • qCut = 30 GeV • Two validation checks: – FxFx merging vs Born level contributions – FxFx merging vs NoFxFx (simple adding of histograms according to weights) • Normalization required to compare same scales 8th April 2014 J. Fernández 5 1 jet Max comparison 8th April 2014 J. Fernández 6 Total number of jets (1jmax) FxFx FxFx NoFxFx 8th April 2014 NoFxFx J. Fernández 7 Radiated pT>10 jets & pT>30 FxFx FxFx 1j 2j NoFxFx 8th April 2014 NoFxFx J. Fernández 8 1st and 2nd leading jets FxFx FxFx NoFxFx 8th April 2014 J. Fernández NoFxFx 9 1st and 2nd radiated jets FxFx FxFx 30GeV 30GeV NoFxFx 8th April 2014 NoFxFx J. Fernández 10 3rd and 4th leading jets FxFx FxFx NoFxFx 8th April 2014 J. Fernández NoFxFx 11 3rd and 4th radiated jets FxFx FxFx NoFxFx 8th April 2014 J. Fernández NoFxFx 12 Normalized FxFx vs NoFxFx FxFx FxFx NoFxFx 8th April 2014 J. Fernández NoFxFx 13 Normalized FxFx vs NoFxFx FxFx FxFx NoFxFx 8th April 2014 J. Fernández NoFxFx 14 Questions • 0.7 = jetradius ≠ 'JetMatching:coneRadius = 1.‘ – aMC@NLO seems to force it to 1. when FxFx is enable in a warning message but LHE file keeps the 0.7 value • Grid pack working in 7XY? Using code from aMC@NLO independent of CMSSW for the moment • LHAPDF interface, full or lite? • CPU time in tt +2j? • If SIM&RECO CMSSW step fails for some events, are the weights still valid? – Should use sum = event_norm instead? • Differences in xsec obtained wrt those in Table 5 of arXiv:1405.0301v1 • Cross/section dependes on parton_shower/ickkw parameters? 8th April 2014 J. Fernández 15