1 / 52

Comparison of acceptances and radiative effects for W/Z for Pythia, Herwig and Mc@nlo

Comparison of acceptances and radiative effects for W/Z for Pythia, Herwig and Mc@nlo. Marc Goulette, Daniel Froidevaux ATLAS CAT Physics Meeting, CERN 18 April 2007. Outline. Introduction Selection cuts Electron acceptances Understanding the differences Conclusion and plans.

havard
Download Presentation

Comparison of acceptances and radiative effects for W/Z for Pythia, Herwig and Mc@nlo

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Comparison of acceptances and radiative effects for W/Z for Pythia, Herwig and Mc@nlo Marc Goulette, Daniel Froidevaux ATLAS CAT Physics Meeting, CERN 18 April 2007

  2. Outline • Introduction • Selection cuts • Electron acceptances • Understanding the differences • Conclusion and plans

  3. 1. Introduction • Study on We and Zee events at the generator level • Here the study is concentrated on We • Software • Release 11.0.4 • CBNT_Truth-00-00-41 • UserAnalysis-00-05-11 • standard DC3-jobOptions (including Photos) • Monte Carlo Generators • Pythia 6.323 Lowest order qq’ W Born • Herwig 6.507 Jimmy 4.0 Lowest order (approx.) qq’ W Born • MC@NLO 3.2 Next to Leading Order • At least ~100 K events produced for each generator

  4. Main Parameters • Same Mass and same Width for all • Herwig and Pythia cross sections in agreement to better than 1% !! • Higher McAtNlo cross section as expected (see spare slides for K-factors)

  5. Parton Density Function (PDFs) description • PYTHIA: • LHAPDF version 4.0 • CTEQ6L1-LO with LO alpha_s • HERWIG: • HWLHAPDF modpdf 10042 • LHAPDF version 4.0 • CTEQ6L1-LO with LO alpha_s • MCATNLO: • HWLHAPDF modpdf 10000 • LHAPDF version 4.0 • CTEQ6m (NLO version of CTEQ6L1-LO)

  6. 2. Selection cuts • LEPTON FILTER: • pT_cut = 0.0 • _cut = 5.0 • STANDARD CUTS: • pTe > 25 GeV • pT > 25 GeV • || < 2.5 • Electron and photon merged if R = (2+2) < 0.04

  7. 3. Electron acceptances for We Electron acceptances as a function of max with pTe > 25 GeV, pT > 25 GeV and || < 2.5: Acceptances in % Acceptance _cut Acceptance ratio • Important differences (413%) !! _cut

  8. 4. Source(s) of the difference ? • PDFs (exactly the same used, at least for Pythia/Herwig) • ISR/FSR/Underlying Event • pTw • Atlas interfaces to the generators • Photos interfaces to different generators • Other parameters (mw and w identical, s, QCD, …)

  9. W+ / W- rapidity (100 K events) W- W+ Number of events Number of events _ _ u.d u.d yw+ yw- • Shape difference due to p-p collision (more valence u than valence d) • Same fluctuations observed for both charged bosons • W+ chosen as an example in the analysis

  10. Rapidity and pseudorapidity Number of events Number of events yW W Important fluctuations for both ( next slide..) No cuts After cuts Ratio Ratio yW yW

  11. Pdfs for Pythia, Herwig and Mcatnlo xi = fraction of energy taken by the quark Pythia is quite different for all cases. Number of events Number of events Herwig and mcatnlo are close for some quarks (xc and xd). _ xu xd _ Why are pdfs different despite choice of same set in JobOptions ? Number of events Number of events _ xs xc

  12. _ Do we get mW from √(x1.x2.s) for e.g. udW ? HERWIG MC@NLO Number of events Number of events ~mW (GeV) ~mW (GeV) PYTHIA Important asymmetry for Pythia ( see next slides) Number of events ~mW (GeV)

  13. Remove Underlying Event pTe>25 GeV, pT>25 GeV and ||<2.5: Acceptances in % before Acceptance ratio  _cut Acceptance ratio after  _cut • Big effect !!

  14. No Underlying Event: effect on pdfs Pythia is still quite different for all (xc is best). Number of events Number of events _ xd xu Herwig and mcatnlo almost equal for all quarks (except xu). Number of events Number of events _ xs xc

  15. No UE and no Initial State Radiation pTe>25 GeV, pT>25 GeV and ||<2.5: Acceptances in % before Acceptance ratio _cut Acceptance ratio after • global increase of acceptance and big effect on ratios !! _cut

  16. No UE and no ISR: effect on pdfs Number of events Number of events _ xu xd Pdfs Better Number of events Number of events _ xs xc

  17. No UE and no ISR: Pythia 6.3vsPythia 6.4 Almost no changes (even on the acceptances) Number of events Number of events _ xu xd Number of events Number of events _ xs xc

  18. No UE, no ISR and no Matrix Element Correction pTe>25 GeV, pT>25 GeV and ||<2.5: Acceptances in % before Acceptance ratio _cut Acceptance ratio after • Some improvement but still not satisfactory. _cut

  19. No UE, no ISR, no ME: back on yW+ W+ W+ Number of events Number of events Standard 100 K events for all generators no UE, no ISR, no ME 100 K Mcatnlo 500 K Pythia & Herwig yw+ yw- • Error bars are correct ! • Pythia and Herwig normalized to Mcatnlo • Fluctuations still important and cannot be attributed to statistics

  20. No UE, no ISR and no ME: effect on pdfs Number of events Number of events _ Some improvement xu xd Number of events Number of events _ xs xc

  21. √(x1.x2.s) for all generators MC@NLO HERWIG Number of events Number of events ~mW (GeV) ~mW (GeV) PYTHIA The tail is still there and needs to be understood Number of events ~mW (GeV)

  22. Problem in Pythia to ATLAS interface? What happened here ? • - We should have : Py_s~+Py_c = Py_W and the same for Px and Pz … • Could it be due to a different frame between the original quarks and the W ?

  23. Problem in Pythia to ATLAS interface? Standard configuration, 5 K events produced Number of events Number of events (MeV) (MeV) But here this is clearly important .. Number of events (GeV)

  24. Problem in Pythia to ATLAS interface? No UE, no ISR, no ME, 5 K events produced Number of events Number of events (MeV) (MeV) And stays after UE, ISR and ME removal (but effect Small, could be precision of interface ?) Number of events (MeV) • Torbjorn S. checked on 10 K events (Thanks !): nice delta func. in standalone mode !!

  25. Problem in Herwig to ATLAS interface? No UE, no ISR, no ME, 3 K events produced Number of events Number of events (MeV) (MeV) Similar effect seen in Herwig Number of events (MeV)

  26. McAtNlo printout • - Events without a W in the • printout are all without Photons ! • These events are in fact • “W without jets” (76.3 % of all events). • Events with a W in the record • Are in fact “W+Jets” (23.7 %). • For these events, we can see • some photons coming out (as • expected..).

  27. Number of photons generated Thus, the total number of photons generated with MC@NLO compared to Pythia and Herwig is much lower: • Problem still present in 12.0.X. Thus for other (later) studies, the evgen files and • the samples will need to be redone (but Borut K. and Jon B. are aware of that…) • - Impact of Photos on acceptance is small (see spare slides).

  28. Applying a pTw weighting factor Nb of events pTw Mcatnlo Weighting factor Low statistics Nb of events pTw Herwig pTw (MeV/c) pTe Herwig after pTe Herwig before Weighting factor pTe (MeV/c)

  29. pTw: effect on electron acceptances pTe>25 GeV, pT>25 GeV and ||<2.5: Acceptances in % before Acceptance ratio _cut after Acceptance ratio • Almost no effect because pdfs obscure everything at the moment _cut

  30. Source(s) of the difference ? Status Done • Pdfs  partly understood with UE, ISR and ME • But after no ME, no ISR, no UE, all generators are not identical !! • pTw  YES, but obscured by pdfs • Photos interfaces to different generators  NO • Pythia shower  test with Pythia 6.4 NO To do • Pdfs: Atlas interfaces to generators  explain mW tail, pdfs and acc.? • ( try Mcatnlo with CTEQ 6L for complete consistency check) • Other parameters (s, QCD, …) ? • Running Herwig and Pythia in standalone mode ?

  31. Conclusion and plans Conclusions • W enu : Important differences found on the acceptance • QED radiation not correct for mcatnlo evgen samples • Differences now attributed to pdfs but not fully understood yet • There may be some problem in the generator to Atlas interface ? Plans • Continue study : interfaces ?, Other param., run in standalone mode ? • Revisit pTW impact on acceptance • Quantify effects for Z e e • Study in detail Photos effect • Derive estimate of systematic uncertainty on acceptance using TruthAtlfastFullSim

  32. Additional material

  33. Acceptances for Zee Electron acceptances as a function of _max with pTe1 > 25 GeV, pTe2 > 25 GeV and || < 2.5: Difference between Herwig and Pythia is smaller (less than 3.2 %), which better shows that the same pdfs have been used…

  34. Main Parameters Tevatron (2006): mW = 80.413 ± 0.48 GeV

  35. K factors • Here we have analytic calculations both at LO and at NLO (NLO code courtesy of James Stirling). • For any PDF we get the following total cross-section to be compared with those from the generators. • W- cross-section (pb) CTEQ6L1 with LO alphas CTEQ6L1 with NLO alphas CTEQ61m NLO • LO calculation 7379.5 6890.1 7570.0 • NLO calculation 8172.8 7552.0 8562.8 • W+Cross-section (pb) CTEQ6L1 with LO alphas CTEQ6L1 with NLO alphas CTEQ61m NLO • LO calculation 10159.3 9472.2 10107.1 • NLO calculation 11373.4 10530.1 11512.1 • SO 1) the fully correct Nlo calculations are in agreement between the analytic code and MC@NLO to 1% • 2) the fully LO calculation is within 1% for W+, a little worse for W- • 3) using LO PDf with LO alphas is closer to the truth than using LO PDF with NLO alphas but NLO is still needed • 4) the k-factors for both types of LO PDF are very similar • 5) if we want a k-factor which can take us from LO calculation with Lo PDF straight to NLO calculation with Nlo PDF then we now have it in the above table by dividing bottom right numbers by top left numbers. Then it explains the numbers fine. • Pythia CTEQ6L1-L0 with LO alphas Herwig CTEQ6L1-L0 with LO alphas Mcatnlo CTEQ6m W+ cross-section (pb) 100111008811456 • W- cross-section (pb) 719972248429 Thanks a lot to Amanda Cooper Sarkar !!

  36. Some numbers • Number of events generated: • MC@NLO W- with Photos: 105 000 • MC@NLO W- without Photos: 105 000 • MC@NLO W+ with Photos: 105 000 • MC@NLO W+ without Photos: 105 000 • HERWIG W with Photos: 105 000 • HERWIG W without Photos: 105 000 • PYTHIA W with Photos: 110 000 • PYTHIA W without Photos: 110 000 • Number of Photons generated: • MC@NLO W- with Photos: 19 236 • MC@NLO W- without Photos: 357 • MC@NLO W+ with Photos: 19 278 • MC@NLO W+ without Photos: 441 • HERWIG W with Photos: 80 703 • HERWIG W without Photos: 558 • PYTHIA W with Photos: 86 999 • PYTHIA W without Photos: 46 783 • Fraction W+ / W-: • Fraction of W- for Herwig: 43806 (41.72 %) • Fraction of W+ for Herwig: 61173 (58.26%) • Fraction of W- for Pythia: 46013 (41.83%) • Fraction of W+ for Pythia: 63987 (58.17%) • Nb of electrons / neutrinos after LeptonFilter cuts: • Mcatnlo W-: 105 000 / 104 916 • Mcatnlo W+: 105 000 / 105 000 • Herwig: 104 979 / 104 979 • Pythia: 110 000 / 110 000 • Number of events W+jets: • Mcatnlo W+: 24 864 (23.68%) • Mcatnlo W-: 24 990 (23.80%) • Number of events with weights -1: • Mcatnlo W+: 9 324 (8.88%) • Mcatnlo W-: 9 051 (8.62%)

  37. Info on JobOptions and samples • Pythia and Herwig samples : • http://alxr.usatlas.bnl.gov/lxr/source/atlas/Generators/DC3_joboptions/share/ • In particular DC3.005100.JimmyWenu.py and DC3.005104.PythiaWenu.py • Mcatnlo samples: • https://twiki.cern.ch/twiki/bin/view/Atlas/WZMcAtNloSample • - Use of DC3.005250.McAtNloWminenu.py and DC3.005254.McAtNloWplusenu.py

  38. Parton selection and event record Pythia Mcatnlo Herwig

  39. No Photos: effect on acceptances pTe>25 GeV, pT>25 GeV and ||<2.5: Acceptances in % before Acceptance ratio _cut after Acceptance ratio • Almost no effect for ||>1.0 !! _cut

  40. McAtNlo events with or without jets The events with/without Jets have a slightly different shape for the pTw: W without jet Number of events W with jet pTw (MeV/c) => But W+jet is a small fraction of all events (23.7%) and pTw doesn’t affect much the acceptances (see later…)

  41. McAtNlo weights Should not be forgotten: weight +1 (91.2%) or -1 (8.8%). Weight +1 Weight -1 All Number of events Number of events pTw (MeV/c) pTw (MeV/c)

  42. pTw and pTe Difference at low pT Ratio Number of events Same behaviour at high pT pTw (MeV/c) pTw (MeV/c) Ratio Number of events pTe (MeV/c) pTe (MeV/c)

  43. mW invariant mass Number of events mW (MeV)

  44. Cos(*) Number of events V-A compatible Cos(*)

  45. No UE and no ISR (1): HERWIG/MCATNLO HERWIG / MCATNLO  No real improvement yet …

  46. No UE and no ISR (2):PYTHIA/MCATNLO  Some improvement (~+6%)

  47. No UE and no ISR (3): HERWIG/PYTHIA HERWIG / PYTHIA • Important effect from UE: needs to be better understood… • Discussion with authors has started (M.Seymour, T.Sjostrand)

  48. No UE and no ISR: mW for all generators Number of events printout Number of events From leptons ~mW (GeV) ~mW (GeV) mWgen mWleptons Sqrt(x1.x2.s) Pythia Sqrt(x1.x2.s) is also calculated later for Herwig and Mcatnlo. Number of events ~mW (GeV)

  49. Pythia mWgen: directly from W, mWrec: from leptons, sqrt(..): from quarks Sqrt(x1.x2.S)/mWgen mWgen/mWrec Sqrt(x1.x2.S)/mWrec Sqrt(x1.x2.S) - mWgen mWrec-mWgen Number of events Number of events (MeV) (MeV)

  50. Herwig Sqrt(x1.x2.S)/mWgen mWgen/mWrec Sqrt(x1.x2.S)/mWrec Number of events mWrec-mWgen Sqrt(x1.x2.S) - mWgen Number of events (MeV) (MeV)

More Related