ME/PS Matching Studies in DØ Daniel Wicke (Fermilab) Outline Introduction MLM CKKW Summary Revision : 1.11 TeV4LHC, 1-Dec-4 1
Introduction Parton shower models good in describing jet evolution, i.e. for soft radiation. restricted to 2 2 hard processes. Matrix element models good in describing hard interactions. no description of jet evolution (soft radiation). Analyses with many hard final state particles (including jets) need both. Simple connection leads to double counting, elaborate matching is required: CKKW: Restrict PS to cover only phase space not covered in ME. MLM: Identify and reject double counting events. This talk shows findings made with such matched MC samples in DØ. Daniel Wicke, ME/PS Matching Studies in DØ, Introduction TeV4LHC, 1-Dec-4 2
MLM Matching Prescription (Reminder) 1. Generate events of given multiplicity according to Matrix Element with cuts on p T and/or spacial distance (e.g. R cone ). 2. Perform parton shower (e.g. with Pythia or Herwig) 3. Process showered event before hadronisation with a jet algorithm 4. Match partons (from 1) with jets (from 3) Exclusive: Each parton is matched to exactly one jet and vice versa. Inclusive: Each parton is matched to a jet (unassociated jets may exist). 5. Combine samples with differing ME multiplicity (constant luminosity): use exclusive matched events except for the last multiplicity bin. e.g. W + N = W+j excl W+1j excl... W+4j excl W+5j incl Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 3
Implementation Match after the fact, i.e. MLM rejection done after full simulation (if any). Allows reuse of unmatched samples. Allows application of various matching parameters. Low matching efficiency requires high statistics w/ full detector simulation. Samples created Alpgen + Pythia (W +jets or Z+jets samples) Alpgen generation cuts include: p T > 8 GeV and R >.4 Varying cone jet parameters: p T = 1, 2, 3 GeV and R cone =.4,.7 Vary highest multiplicity taken from ME: N gen jets... 4 3 samples per process (W +jets, Z+jets, etc.) Results should be independent of exact matching parameters. Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 4
Where do the events come from? Fractional Contribution 1..9.8.7.6.5 W + N =...6 Jet Production Reconstructed: p T >15 GeV/c, R<.5 Generated: p T >1 GeV/c, R>.4 Matched: p T >1 GeV/c, R<.4 W+j exclusive W+1j exclusive W+2j exclusive W+3j exclusive W+4j exclusive W+5j exclusive W+6j inclusive Fractional Contribution 1..9.8.7.6.5 W + N =...6 Jet Production Reconstructed: p T >15 GeV/c, R<.5 Generated: p T >1 GeV/c, R>.4 Matched: p T >3 GeV/c, R<.7 W+j exclusive W+1j exclusive W+2j exclusive W+3j exclusive W+4j exclusive W+5j exclusive W+6j inclusive.4.4.3.3.2.2.1.1. 1 2 3 4 5 6 7 8 Jet Multiplicity Event mixture (by multiplicity bin) is strong function of matching criteria. Important for requesting sample statistics.. 1 2 3 4 5 6 7 8 Jet Multiplicity Gen. p T, R, N Mult. 1,.4,i4 3,.7,i4 9% 9% 1 2% 4.7% 2 5%.8% 3.7%.3% 4.1% Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 5
Stability wrt. matching cirteria Result should not depend on the generator cuts or matching criteria. σ (pb) 1 5 1 4 1 3 1 2 W + N =...6 Jet Production Reconstructed: p T >15 GeV/c, R<.5 Generated/Matched: p T >1 GeV/c, R<.4 Generated/Matched: p T >3 GeV/c, R<.7 Matched result lies between the generated cross sections and displays different multiplicity dependence. 1 1 1-1 1-2 ALPGEN: p T >1 GeV/c, R>.4 ALPGEN: p T >3 GeV/c, R>.7 1-3 1 2 3 4 5 6 7 8 Jet Multiplicity Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 6
W +jets (Muon channel) Analysis Selection t t cross-section analysis in l+jets channel (topological selection). Isolated high p T muon. E miss T > 2 GeV η < 2.5 Here: 2 jets p T > 2 GeV (more lacks MC statistics) Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 7
T T 6 6 2 3 g jet η 5 1 4 15 6 p T spectra W transverse for various mass M T inclusive bins 4 6 25 8 4 8 6 3 15 6 4 1 4 5 1 M T -3-2 5-1 1 4 1 15 6 2 3 W transverse leading 2nd leading jet mass M T p T jet η 25 6-1 1 2 3 2nd leading jet η 4 15 6 4 8 4 1 3 6 6 1 3 8 4 4 6 4 1 1 5 1 1 2 15 3 W φ(leading transverse jet, mass missing E T ) 5 1 15 W transverse mass 291 Evt. data % ttbar 94 % W+jets 6 % QCD 5 5 1 1 4 1 2 15 6 15 3 W φ(leading transverse 2nd leading jet, mass jet M missing jet T p T p T E T ) 5 1 1 15 leading jet p T 25 6 4 8 15 6 291 Evt. data % ttbar 94 % W+jets 6 % QCD 1 8 6 4 4 3 1 3 8 6 1 4 1 1 2 3 φ(leading jet, missing E T ) 1 2 3 φ(leading jet, missing E T ) 1 N =, 1 too soft. 1 291 Evt. data % ttbar 291 94 % Evt. W+jets data 6 % QCD % ttbar 94 % W+jets 6 % QCD N = 2, 3, 4 better. 5 5 1 1 1 2 15 15 3 W φ(leading transverse 2nd leading jet, mass missing jet p T E T ) 291 Evt. data 1 5 1 % ttbar 15 2nd leading 94 jet % W+jets p T 6 4 8 6 % QCD 1 1 4 4 4 5 5 3 1 1 8 6 6 2 3 g jet η -3-2 5-1 1 1 15 2 3 leading 2nd leading jet p T jet η -3-2 5-1 5 1 1 1 15 15 2 3 2nd leading 2nd leading jet jet p T p jet T η 5 5 1 1 15 15 2nd leading jet jet p T p T Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 8
W transverse mass. R =.4 R =.7 8 4 6 3 4 4 3 3 3 1 1 1 1 1 2 3 φ(leading jet, missing E T ) ev] 6 25 4 5 1 4 15 6 W transverse mass 291 Evt. data % ttbar 94 % W+jets 6 % QCD M T R =.4,.7 look 6 the same. 1 8 4 5 1 1 2 15 3 φ(leading W transverse jet, missing mass E T ) 291 Evt. data % ttbar 94 % W+jets Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 9 6 % QCD 1 8
4 6 M T 1 2 3 φ(leading p T 25 jet, cut missing E T ) 8 6 15 4 1 5 291 Evt. data % ttbar 94 % W+jets 6 % QCD p T > 1 GeV: better for p T -spectrum. -3-2 -1 4 1 6 2 3 5 1 15 2nd leading M T jet η 2nd leading jet p T 25 p T > 3 GeV: (and also p T > 2 GeV) 15 better in event shapes. 1 But 5our default unmatched samples look as good (maybe -3-2 even -1 better). 1 2 3 2nd leading jet η 6 4 4 3 1 5 1 15.2 1.4.6 2.8 1 3 W transverse mass φ(leading jet, missing Centrality E T ) 25 6 1 3 8 4 15 6 291 Evt. data % ttbar 94 % W+jets 6 % QCD 5 1 15.2 5 1.4 1.6 2 15.8 1 3 W transverse leading jet mass p T φ(leading 2nd leading jet, Sphericity missing Centrality jet p T E T ) 6 4 5 1 15 leading jet p T 1 14 1 8 1 6 8 6 1 4 4 1 4 5 1 5 1 4 8 3 6 1 291 Evt. data % ttbar 94 % W+jets 6 % QCD 4 1.2 5.4 1.6 15.8 1 2nd leading Sphericity jet p T 1 8 6 4 Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 1
W +jets (Electron channel) t t cross-section analysis in l+jets channel (with b-tagging). Leading jet η distribution in e + 1jet Unmatched Matched 7 6 eta_j 7 6 eta_j 5 5 4 4 3 3 1 1-2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5-2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5 Discrepancy for the matched sample,... Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 11
Leading jet η distribution e + 2jets e + 3jets 9 25 eta_j 8 7 eta_j 6 15 5 4 1 3 5 2 1-2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5-2.5-2 -1.5-1 -.5.5 1 1.5 2 2.5... but higher jet multiplicities seem OK. Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 12
Z+jets (Muon channel) Analysis Selection t t cross-section analysis in dimuon channel: Vertex quality 2 muons: Medium tagged, p T > 15 GeV, η < 3. 75 < M µµ < 15 GeV Investigating Z inclusive and Z + 2jet events. Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 13
Leading jet p T Z inclusive Z + 2 jets 8 Data MLM 1.4 3 7 MLM 2.4 3 MLM 3.4 3 14 Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 Matched better in Z incl. 6 5 4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia N gen max = 12 1 8 6 4 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia N gen max = But not in Z + 2j, here Pythia is wider data even more 1 2 1 2 3 4 5 6 7 8 9 1 Leading Jet Pt [GeV/c] 2 4 6 8 1 12 14 Leading Jet Pt [GeV/c] 7 6 5 4 3 1 1 2 3 4 5 6 7 8 9 1 Leading Jet Pt [GeV/c] Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia N gen max = 3 14 Data 12 1 8 6 4 2 MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia N gen max = 3 2 4 6 8 1 12 14 Leading Jet Pt [GeV/c] Matched Algen-Pythia Pythia Data Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 14
Leading muon p T 2 18 16 14 1 1 8 6 4 Z inclusive 1 2 3 4 5 6 7 8 9 1 Leading Muon Pt [GeV/c] Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia 16 14 12 1 8 6 4 2 Z + 2 jets 2 4 6 8 1 12 14 Leading Muon Pt [GeV/c] Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia Matched better in Z incl. But not in Z + 2j again matched too narrow N gen max = 2 18 16 14 1 1 8 6 4 1 2 3 4 5 6 7 8 9 1 Leading Muon Pt [GeV/c] Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia 14 12 1 8 6 4 2 2 4 6 8 1 12 14 Leading Muon Pt [GeV/c] Data MLM 1.4 3 MLM 2.4 3 MLM 3.4 3 MLM 1.7 3 MLM 2.7 3 MLM 3.7 3 Pythia N gen max = 3 Matched Algen-Pythia Pythia Data Daniel Wicke, ME/PS Matching Studies in DØ, MLM Matching TeV4LHC, 1-Dec-4 15
CKKW Matching Introduction The CKKW prescription restricts the PS such that double counting is avoided. Implementations F. Krauss and co. in Sherpa. Steve Mrenna pre-produced samples using Madgraph II and Pythia. Samples Sherpa: Z+jets (ME up to Z + 3j) with Z µµ and full det. simulation S. Mrenna samples are currently processed... Daniel Wicke, ME/PS Matching Studies in DØ, CKKW Matching TeV4LHC, 1-Dec-4 16
Z+jets Sherpa compared to Pythia with and without underlying event. p T (Z) UE has not significant effect; good agreement between models; but Pythia doesn t fully describe data. normalized entries.1.5 Sherpa Pythia w/ UE Pythia w/o UE normalized entries -1 1-2 1 1-3 Sherpa Pythia w/ UE Pythia w/o UE 5 1 Z p t -4 1 5 1 Z p t Daniel Wicke, ME/PS Matching Studies in DØ, CKKW Matching TeV4LHC, 1-Dec-4 17
Jet multiplicity Cone algorithm R =.5, p T > 2 GeV. UE increases jet multiplicity. Sherpa has more jets (p T > 2 GeV). normalized entries 1 Sherpa Pythia w/ UE Pythia w/o UE -1 1 1-2 -3 1 2 4 6 incl. jet multiplicity (E t > 2 GeV) Daniel Wicke, ME/PS Matching Studies in DØ, CKKW Matching TeV4LHC, 1-Dec-4 18
Leading jet p T Small modifications due to UE. Sherpa shows harder spectrum. normalized entries.2.1 Sherpa Pythia w/ UE Pythia w/o UE 5 1 st 1 jet E t normalized entries.4.2 Sherpa Pythia w/ UE Pythia w/o UE normalized entries.6.4 Sherpa Pythia w/ UE Pythia w/o UE.2 5 1 nd 2 jet E t 5 1 rd 3 jet E t Daniel Wicke, ME/PS Matching Studies in DØ, CKKW Matching TeV4LHC, 1-Dec-4 19
Summary and Outlook DØ has investigated MLM and CKKW matching schemes. MLM with Alpgen+Pythia and various prescription param. in Z+jets, W +jets. (Tuned) pythia seems to be better in many cases. Statistics isn t sufficient to be decisive (yet?). Studies with heavy flavours are in progress. CKKW with Sherpa in Z+jets. Underlying event model needed for fair comparison of Sherpa with data. Expected in next version. CKKW with Madgraph+Herwig in progress. Acknowledgments Thanks for the support in preparing this talk to Michael Begel, Tobias Golling, Sasha Khanov, Jessica Leveque, Thomas Nunnemann, Jovan Mitrevski... Daniel Wicke, ME/PS Matching Studies in DØ, Summary and Outlook TeV4LHC, 1-Dec-4 2