Rjets WjetsZjets Status Report Evelin Meoni On behalf
Rjets (W+jets/Z+jets) : Status Report Evelin Meoni On behalf of the Rjets group: G. Arabidze, C. Chavez Barajas, H. Beauchemin, M. Bona, G. Brandt, B. Brau, N. Calace, S. Chouridou, M. Dunford, G. Fletcher, T. Hayashi, J. Huston, R. King, A. Meade, E. Meoni, M. Mondragon, J. Nielsen, J. B. Sauvan, C. Sawyer, A. Tricoli 1 st Ed Board Meeting 23 th January 2013 1
Outline This outline “roughly” represents the current skeleton of the Supporting Note with some differences to better summarize the status of the work (what has been done/what we still have to do): • Motivation • Measurements and Analysis Strategy • Data and MC sample • Event selection • Detector level plots: - background estimation - overall systematics at detector level • Unfolding: - method - systematic uncertainties For each point outlined in red, work to be done and on-going activities ( to do list summary at the end) • Theoretical predictions • Results • Conclusions At the end few ideas/proposals about how synchronize the review paths of Rjets and Wjets papers 2
Motivations Boson + Jets measurements in general: - important bkg for new physics and test of the p. QCD Ratio measurement (W+jets)/(Z+jets) in particular: - high precision test of p. QCD reduced uncertainties both at experimental level (strongly reduced JES, LUMI) and theoretical level (significantly reduced dependence on PDFs) , but information about dynamics of V+jets production conserved in the ratio - model-independent sensitivity to new physics : new physics preferentially coupling to W or Z could cause a deviation of the ratio in a particular kinematic regime - useful. PDFs tool uncertainties for background predictions in searches mapping W/Z +jets in different regions test similarity Met Z ee+jets e e W en+jets 3
Measurement Past measurement with 2010 dataset (Phys. Lett. B 708 (2012), 221 -240): “A measurement of the ratio of the W and Z cross sections with exactly one associated jet in pp collisions at sqrt(s) = 7 Te. V with ATLAS”: - measurement for the first time as a function of jet p. T thr - measurement done for electron and muon channel and combination -comparison with LO +PS MCs (Pythia and Alpgen) and NLO calculations (MCFM): good agreement btw data and predictions Goal for measurement with 2011 dataset: Njets : inclusive and the ratio (njet/(njet-1) ) Njets for pt_lead > 150 Ge. V: njet/(njet-1) ( Test scaling differences between W and Z) Jets: p. T and y of 1 st , 2 nd , 3 rd and 4 th jet for Njets>=1, 2, 3, 4 Dijets variables: m 12, DR 12, Df 12, Dy 12 Complex : HT and ST for Njets>=1, 2, 3, 4 and for Njets==1, 2, 3, 4 Distributions in VBF phase space (mjj>350 Ge. V, |Dh|>3): p. T and y of the 3 rd jet Comparison with LO MCs (Alpgen and Sherpa) and with NLO calculations (Black. Hat+Sherpa) 4
Strategy (1/2) Measurement at detector level : - Data (2011 stat) and MC (mc 11 c, pileup reweighting applied, corrected for lepton momentum/energy scale and lep eff SF) - Signal measured as Data – BKGs: Hadron level: -Correct for detector effect (trigger and reconstruction eff. and resolution) with Bayesian Unfolding (Roo. Unfold) in the same phase space of the detector level measurement with Alpgen (dressed leptons) Comparisons of unfolded data with LO MCs+PS (Alpgen and Sherpa) and with NLO calculation (Black. Hat+Sherpa) corrected Parton level: NLO calculation corrected by NP effects (fragmentation and UE) and QED effects (born level dressed) 5
Strategy (2/2) • Measurement done first for electron and muon in as similar as experimentally possible phase space: Electrons (dressed level) Z: p. T(e)>20 Ge. V |h|<2. 47 (crack excl), 66 < m. Z < 116 Ge. V W: p. T(e)>25 Ge. V |h|<2. 47 (crack excl), p. T(v)>25 Ge. V, MT > 40 Ge. V Jet : p. T> 30 Ge. V |y|<4. 4 • Combination in We’ll publish just the combination Muons (dressed level) Z: p. T(m)>20 Ge. V |h|<2. 4, 66 < m. Z < 116 Ge. V W: p. T(m)>25 Ge. V |h|<2. 4, p. T(v)>25 Ge. V, MT > 40 Ge. V Jet : p. T> 30 Ge. V |y|<4. 4 the common generic phase space : Combination (dressed level) Z: p. T(l)>20 Ge. V |h|<2. 5, 66 < m. Z < 116 Ge. V W: p. T(l)>25 Ge. V |h|<2. 5, p. T(n)>25 Ge. V, MT > 40 Ge. V Correction of unfolded electron and muon results for space-phase done with Alpgen Combination Code and References • About Combination: Chi 2 method : measurement of Ratio(e) and Ratio (m) first Combination of the 2 Ratios 6
Data and MC samples • 2011 Data: - period D-M - GRL: data 11_7 Te. V. period. All. Year_Det. Status-v 36 -pro 10_Cool. Run. Query-00 -04 -08_WZjets_allchannels. xml - 4. 64 fb-1 for each stream (EGamma and Muon) - SMWZ D 3 PDs tag p 833 • MC: -mc 11 c samples - Process Generator W mn Alpgen+Jimmy(0 p. . 5 p) and Sherpa v 1. 4 Z mm Alpgen+Jimmy(0 p. . 5 p) and Sherpa v 1. 4 W en Alpgen+Jimmy(0 p. . 5 p) and Sherpa v 1. 4 Z ee Alpgen+Jimmy(0 p. . 5 p) and Sherpa v 1. 4 Z tt Alpgen+Jimmy(0 p. . 5 p) WW, ZZ, WZ Herwig ttbar Alpgen+Jimmy (0 p. . 5 p) Single top Acer. MC Normalized to (N)NLO XS: https: //svnweb. cern. ch/trac/atlasgrp/ browser/Physics/Standard. Model/ Common/Winter 2012/mc 11 c_p 833_info. txt -QCD estimated with data driven-techniques TO BE DONE: some checks (see next slides) -top : currently estimated via MC TO BE DONE: use Data-Driven estimation (see next slides) 7
Event selection Rjets analysis tool on SVN : Physics. Analysis/Standard. Model. Phys/WZJets/Wzratio Common W/Z baseline selection: https: //twiki. cern. ch/twiki/bin/viewauth/Atlas. Protected/WZElectroweak. Common. Topics 2011 Preselection: -trigger Electron Trigger W: e 20_medium(D-I), e 22_medium (K), e 22 vh_medium 1 (L-M) Z : 2 e 12_medium(D-I), 2 e 12 T_medium (K), 2 e 12 Tvh_medium (L-M) Muon. Trigger: mu 18_MG (D-I), mu 18_MG_medium (J, M) - primary vertex (at least one good vertex with >=3 track ) -MET cleaning and LAr Hole Veto Lepton selection: Electrons: Author 1 or 3, Quality flag (OQ&1446 !=0) p. T>25 Ge. V (W), 20 Ge. V (Z) |n|<2. 47 (excluding 1. 37 -1. 52) Muons: STACO Combined with cleaning cuts (MCP recommendations), p. T>25 Ge. V (W), 20 Ge. V(Z) |n|<2. 4 Z selection: 2 OS leptons, 66 Ge. V< mll<116 Ge. V Electrons: 2 OS medium electrons Muons: 2 OS muons, at least the 1 st with d 0 sign (<3), z 0(<10 mm), isolation requ. (ptcone 20 / pt < 0. 1) W selection: MET > 25 Ge. V, m. T(W) > 40 Ge. V MET: MET_Ref. Final, JES and lepton scale prop with Missing. ETUtility Electrons: 1 tight++ electron Muons: 1 muon with d 0 sign, z 0, isolation requ. Jets selection: Antikt 4 Topo. EM at EM*JES scale p. T > 30 Ge. V, |y| < 4. 4 |JVF| > 0. 75 ( for jets with |η| < 2. 4) ΔR(jet-lepton) > 0. 5 8
QCD bkg estimation: electrons Njets=1 QCD mainly originated by light jets faking an electron W channel: QCD template from data : loose trigger and events with loose electrons, passing at least one medium and failing tight is. EM bit and anti-isolated - QCD normalization by fitting MET on data with QCD templ+ sign and other BKGs (from MC): - fit done in the MET region 10 -100 Ge. V - fit done in exclusive jet bins up to 5 jets - Systematics : change of the fit range, change of the template (changing the electr. definition) -Note: estimation done separately for per. D-K and per. L-M (different pileup and trigger pre-scale) and sum up (Somef checks on –going see conclusions and W+jets Ed. Board of Friday) =QCD/Data W Njets=0 Njets=1 Njets=2 Njets=3 Njets=4 Njets=5 Nominal (%) 6. 35 18. 09 17. 97 16. 72 11. 50 6. 11 Rel Uncert (%) 16. 30 12. 90 10. 72 8. 70 6. 76 37. 64 QCD Z channel - QCD template from data : select events with>=2 loose electrons fails medium - QCD normalization by fitting mee with QCD templ + Signal and other BKGs (from MC): -fit done in the region 54 -140 Ge. V- fine done in excluding bins up to 2 jets - Systematic uncertainties : fit range, electron Njets=2 energy scale variation Z f =QCD/Data Njets=0 Njets=1 QCD Nominal (%) 0. 31 0. 36 0. 62 Rel Uncert (%) 20. 16 34. 86 24. 86 9
QCD bkg estimation: muons Njets=1 QCD mainly originated by muon coming from heavy flavour decays W channel - QCD template from data : inverting d 0 sign cut (d 0 sign>3. 0) - QCD normalization fixed by fitting MET on data with QCD templ. + signal and other BKG (from MC): fit in 10 -100 Ge. V, fit done in excl jet bins up to 4 jets and inclusive for njets>=5 Z channel: - QCD template from data: inverting isolation -QDC norm fixed by fitting di-muon mass on data with QCD temp. + signal and other BKG (from MC): fit in 45 -150 Ge. V; fit done in excl jet bins up to 2 jets and inclusive for njets>=3 Systematics : Fit Range, Fit variable, alternate QCD template W Njets=0 Njets=1 Njets=2 Njets=3 Njets=4 Njets>=5 Nominal (%) 4. 22 14. 42 13. 45 13. 06 10. 93 8. 28 Rel Uncert (%) 23. 9 10. 35 13. 43 19. 57 37. 19 44. 80 Z Njets=0 Njets=1 Njets=2 Njets>=3 Nominal (%) 0. 33 0. 53 0. 92 0. 81 Rel Uncert (%) 9. 48 10. 73 63. 78 91. 33 10 f. QCD=QCD/Data
ttbar data-driven estimation Still to be included in the analysis framework Currently detector level plots and Unfolded plots done with ttbar from MC, in the next switch to data-driven. Here just few words (more details in Ed. Board of Friday for W+jets paper; W+Jets and Rjets group joined in one group sharing codes/results) ttbar template from Data: Njets=3 Selected events with : >=2 jets , MET and MT cut Required at least 1 b-jets Independent variable to fit the normalization: Fit on transverse sphericity the ttbar template subtracted by W+jets (h. f and light jets) contributions estimated with from MC template Systematics currently : different W MC templates (Alpgen vs Sherpa large contribution in the Njet=3 bin region currently), subtraction of W template (25% variation considered), statistical uncertainty on the fit Ongoing activities syst. from: fit variable (Mtop and MET), fit range Work to be finalized 11 Njets=5
Detector level : Njets (exclusive) Here shown the style of our detector level plots (what kind of information are in) Systematic band: JET: Jet Energy Scale and Jet Energy resolution ; Leptons: Muons : Reconstruction, Trigger Efficiency, Momentum scale and resolution Electrons: Reconstruction and Identification, Trigger Efficiency, Energy scale and resolution MET Bkgs: unc on XS for Normalization for bkg taken from MC; for QDC from data-driven estimation (Tag of the package in bkup slides) 12
Systematics on detector level plots: Njets (exclusive) W- m ch W- e ch Here shown the style of our detector level systematics plots Z- m ch Z- e ch TO BE DONE: On going checks on few small differences between electron and muon channel In the Supporting Note: for each variable shown detector level plot and systematic plots for W and Z for muon and electron. A few of the results in the main text (for sure Njet, leading p. T and y) , the rest in the appendices 13
Unfolding Unfolded W and Z in each channel (e, m) separately Performed R(m) and R(e) Comb Unfolding technique: Bayesian iterative unfolding with Roo. Unfold, implemented a - “match method” (2010 and 2011 Z+jets group method): applied a match reco-truth jets (best DR match regardless p. T order). - build : response matrix and fake correction (to account for un-matched reco jets) respecting the truth p. T ord. - fake correction applied as a bin-by-bin external correction before unfolding - “unmatch method” (similar not identical to 2010 W+jets group method): -no match reco-truth jets - build fake correction (to account for reco jets in events not passing the truth selection) -fake correction applied as a bin-by-bin external correction before unfolding Currently all the results shown in the supporting note are based on “match approach” with a fixed number of iteration (3 it) regardless the variable. TO BE DONE/On going activities : -Check full set of results with “un-match method” if working use it as nominal method - Fix the convergence criterium to define the proper number of iteration for each variable (on going study , close to the end, see backup slides) 14
Unfolding matrix W+ jets MC details: -Used Alpgen signal sample to unfold data -Closure test done with Sherpa emulating the data Muons Electrons Here shown the style of our matrices representation in the supporting note: Z + jets Each bin normalized to the Hadron level entries of the corresponding Column Shown migration & efficiency in the same plot Muons Electrons In the supporting note, plan to put all the matrices in an appendix 15
Muon Channel Closure test Sherpa (used to emulate Data) Unfolded with Alpgen W+ jets Z + jets Bayesian method with different number of iterations provide optimal closure in the region with high statistics, some differences observed in the region with low statistics uncertainty term relating to unfolding (see next slides) should cover this effects W+ jets Z + jets 16
Systematics on unfolded results (1/3) W mn + jets Z mm + jets For each source varied signal (matrix and fake correction) and bkg estimated with MC (for unfolding unc source moved Signal only, for bkg unc source moved bkg only) Experimental Systematics: JET, Muon, Electron, BKGS as for detector level plots Unfolding Systematics: Method: Bayesian vs. Bin-by-bin ; Modeling: Unfolding with Alpgen vs. Sherpa Statistical Uncertainty (to be implemented): Toy MCs to fluctuate the MC inputs (matrix and distributions) 17 To Do List: ) -Produce uncertainty plots on unfolded results also for electron channel (debugging on going) -In muon channel to be added the statistical -uncertainty on unfolding
Systematics on unfolded results (2/3) For each systematics source, the corresponding unfolded result for W and Z taken to form the varied Ratio W mn + jets Z mm + jets Ratio is a very precise measurement at low multiplicity, at high multiplicity (Njets>=4) precision limited by BKGs uncertainty : different level of top bkg in W and Z it doesn’t cancel in the ratio (needed ttbar datadriven estimation with an uncertainty <10% for Njets>=4 ) 18
Systematics on unfolded results (3/3) Systematics drastically reduced in the ratio W mn + jets Systematics suffer from statistical fluctuations of MC TO BE DONE: Apply a partial rebin of some variables in the region with low statistics (use first final Z+jets binning) In particular for Rapitidy Plots: show absolute value Results 19 Ratio Z mm + jets
Predictions • NLO p. QCD preditions with Black. Hat+Sherpa : CT 10 PDFs , ren and fact scale at HT/2 (scalar sum of all particles/partons in final state) • Corrections : • Nominal correction derived from Alpgen+Herwig (Auet 2 -cteq 6 l 1 tune )syst estimated with Alpgen+Pythia (Perugia 2011 C)(UE and hadronization) and Sherpa (QED corrections) W 20 W W 20
Predictions For Z+jets used same prediction of Z+jets paper, for W+Jets used results produced in the same way : -pdfs uncertainties computed using the complete eigen vector set, impact of as evaluated, scale uncertainty evaluate varying it by a factor of two W+Jets full sets results arrived this Monday TO BE DONE: Work of the uncertainties of the prediction (scale , pdfs and QED and acceptance Corrections varying correlated uncertainties in parallel calculating the ratio) Currently on supporting note just shown central value for the predictions 21
Ratio vs Njets exclusive Discrepancy of Data/MC in the Ratio at high multiplicity seems an issue related with unfolding in W (data/MC at detector level agree well) in a region with low statistics and large migrations Effect covered by the large uncertainties at high Multiplicity Z. Ch results fully consistent With Z+jets paper results Muon channel all syst in W Z Muons RATIO Muons W Muons Z Ele RATIO Electron channel: just unf sys in (TO BE UPDATED) 22
Leading jet p. T W Z Muons RATIO Muons Significant deviations Between Data and Predictions (mainly in muo ch) Shapes seems rather consistent Investigation through the double ratio (vs inclusive) Ele Ele 23
W Muons Significant deviations between Data and Predictions Z Muons RATIO Muons Shapes seems rather consistent Investigation through the double ratio (vs inclusive) Just overall normalization Improved Data/MC agreement Double Ratio Currently in Supporting Note shown Ratio of Absolute XS. For some distributions we want to show 24 defined Double ratio (just overall normalization). List to be
Leading jet y W Muons Significant deviations Between Data and Predictions W Ele Z Muons RATIO Muons Z Ele Shapes seems rather consistent 25
W Z Muons RATIO Muons Just overall normalization Improved Data/MC agreement W Z Double Ratio 26
Conclusions Rjets is a very high precision measurement and with 2011 statistics we are able to significantly extend the 2010 measurement to a larger number of observables. Some good agreement of data with NLO and LO MCs+PS and some deviations larger than uncertainties in the Ratio provide very useful information for QCD Wait for the finalization of the different parts of the analysis for final Statements. 27
To Do List (1/2) QCD data-driven: QDC estimation in W increases compared to 2010 in both ch (3 xele, 4 xmuon)(differences in evt selection, different level of pileup) needed some investigation (check heavy-flavor component in e-ch, try estimation periods also in m ch, Cross-check with incl. group results) ttbar data-driven: Finalize the results (work on systematics) Include in the analysis results Detector level plots: complete (just some on-going checks on the systematics) Unfolding: -Check the impact of the method without matching truth-reco -Finalization of convergence criterium -Electron channel: Propagation of the systematics on unfolded results -Muon channel: Add last missing systematics: statistical uncertainty on unfolding - Missing in both channel a few variables ( scaling: (njets+1)/njets) Predictions: Complete set of results just arrived , work on uncertainty estimation Combination: Code ready, we need to complete systematics propagation in ele chan to show results 28
To do list (2/2) Documentation: Supporting note already in place. The svn location is: https: //svnweb. cern. ch/trac/atlasphys/browser/Physics/Standard. Model/WZPhysics/Analy ses/Ratio. Jets/Rjets. Internal 5 fb Expected to be a very extended supporting note (basically 3 times the Z+jets one) Currently 250 pages and a lot of material still missing What is there? -Basically the full set of results (plots) we currently have : detector level plots with syst ones, unfolded result and syst ones (matrices, closure), Prediction corrections for almost the full set of variables -About the text : first part is the more documented: MC and data samples, selection, QDC bkg, minimal on the rest (expecially on the unfolding) -We should prefer to improve it before invite you to read it (maximum one week) In order to clean it , make it more organic and readable (a lot of material is it, but it need to be organized) This should facilitate and speed the things at the end 29
Rjets and Wjets papers In December at the time we required an Ed. Board for Rjets , the W+jets analysis had a longer time scale In extended discussions (with SM convernors , W/Z convenors and among the 2 groups) we felt it was important to publish the W+jets results on the same time scale (Z+Jets paper will give the open presentation next week). So: -R+jets and W+jets agreed to work together to accelerate the W +jets publication (with reduced in scope to realistically meet the new timeline) - W+jets and R+jets will be 2 publications with the same Ed. Board to facilitate this goal Having same Ed. Board and being the same group of persons for both the analyses, if possible try to optimize the work/time: - Do we want maintain the Ed. Board meetings separated or do we want joined Ed Board meetings? As analysis group we should prefer joined meetings -Do we want 2 separated supporting notes (one for each paper) or one supporting note for both papers? Different opinions within the analysis group, we think the choice has to be driven by preference of the Ed. Board in line with the convenors (this option has not been discussed before with them) 30
BACKUP 31
Technical remarks MC specific corrections: -Pileup: Pileup. Reweighting-00 -02 -01 -Leptons: Electrons: -Correct for Energy scale and reconstruction, ID and trigger efficiencies: : egamma. Analysis. Utils-00 -02 -76 Muons: -Correct for: momentum scale, resolution, reconstruction and trigger efficiencies: Muon. Efficiency. Corrections-02 -01 -00 Muon. Momentum. Corrections-00 -07 -00 Trig. Muon. Efficiency-00 -01 -10 MC/data : JETS: Apply. Jet. Calibration-00 -01 -06 MET: Missing. ETUtility-01 -00 -09 Systematics: -Leptons: previous packages -JETs: Apply. Jet. Resolution. Smearing-00 -00 -03, Jet. Resolution-00 -07 -14, Jet. Uncertainties-00 -07 -04 - MET: previous package 32
Detector level: leading jet p. T 33
Converge criteria for unfolding Convergence criterium, 3 method under evaluation: 1) 2011 Z+jets paper approach: Used Sherpa to emulate the data and Alpgen to unfold 2) (close to) 2010 W+jets paper approach: Stop iterations when difference on 2 consecutive is smaller of the statistical unc 3) Best iteration is the one for which the unfolding uncertainty on XS results is minimal I meth Z II meth Z 34 III meth Z
Predictions • NLO p. QCD preditions with Black. Hat+Sherpa : CT 10 PDFs , ren and fact scale at H T/2 (scalar sum of all particles/partons in final state) • Corrections : • Nominal correction derived from Alpgen+Herwig syst estimated with Alpgen+Pythia (UE and hadronization) and Sherpa (QED corrections) Z Z Z 35
- Slides: 35