Lidia DellAsta Universit degli Studi INFN Milano Status

Lidia Dell’Asta Università degli Studi & INFN Milano

� Status of the analysis in ATLAS and Italian effort � W→τν & Z→ττ analysis: � Analysis aims for an integrated luminosity of 100 pb-1 event selection Signal&Background expected � First data/MC comparison loosened selections first results � Usage of: D 3 PDs Benchmark analysis Good Run List Cleaning cuts � Problems � Conclusions&Outlook Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 2

� � � The analysis are developed within the Tau Performance working group and two separate subgroups in the SM W/Z Signatures group. People involved from Milano: � W→τν: L. D. A. , A. Andreazza � Z→ττ: D. Cavalli, S. Consonni, C. Pizio Recent work in which we have been deeply involved: � Development of the Z→ττ benchmark analysis. All the future analysis should be validated against it. � Note on the analysis of Z→ττ events on MC data at 10 Te. V (D. Cavalli, S. Consonni, C. Pizio, ATL-COM-PHYS-2010 -105) � Note on the analysis of W→τν events on MC data at 10 Te. V (L. D. A. , A. Andreazza, ATLCOM-PHYS-2010 -067). The results are based on simulated data samples corresponding to 100 pb-1 of integrated luminosity. We assumed 10 Te. V center of mass energy collisions and an instantaneous luminosity of 1031 cm-2 s-1. Optimization of the cut flow for the analysis on 7 Te. V data for both channels. � Early data/MC comparison. � We are working to contribute to the first Tau CONF Note on 7 Te. V data with a section devoted to W→τν analysis. � Responsibility for the production of Tau Performance group D 3 PDs from MC (L. D. A. ) � Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 3

� � W→τν analysis: � Studies possible only at low luminosities (up to 1032 cm-2 s-1) � Most abundant source of τ in early running � Analysis aiming to the observation of the first τ candidate � First possibility to access the tau identification capability by: observing the tau signal measuring the cross section ratio (W→τν)/(W →lν) Signature: � Tau decaying to hadrons (hadronic jet, well collimated and with few charged particles) � Missing ET Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 4

� Event selection: � Trigger: combined trigger which asks for a minimum amount of missing ET and for a τ with a minimum p. T of 12 Ge. V → EF_tau 12_loose_xe(10, 15, 20). � Missing ET: the offline reconstructed missing ET must be higher than 35 Ge. V. � Tau identification: there must be only one τ which passes strict selection cuts, both on identification and on momentum. � Veto events with leptons with p. T > 5 Ge. V and loose identification criteria. � Veto events with jets with p. T > 20 Ge. V in the crack region. � Topological cut: Δφ(τ, MET) > 2. 5. � Lidia Dell'Asta With such a selection, expected at the end of the cut flow, in 100 pb-1: 5000 events, with a S/B ratio of 3. 7. For further details about event selection, see back-up slides Sestri Levante, 14 Giugno 2010 5

� For first data analsis and first W→τν observation: � Stream selection: L 1 Calo Stream + DESD_MET skimming � Good. Run. List selection: tau CP GRL � Collision candidate & jet cleaning � Slightly modified cut flow: Trigger: L 1_J 5 Lowered missing ET cut (MET_Topo > 20 Ge. V) � In this situation we are overwhelmed by QCD background. Trying to exploit some new cut to help in the rejection of QCD background → MET significance With this new cut (still under investigation) we are able to relax some cuts (such as MET cut) and to have both higher S/B ration and signal selection efficiency. � To be proved that this quantity is well known even at the beginning of data taking and with pile up becoming more important. � � � With the new loosened event selection + new cut, in 10 nb-1: � ∼ 1 signal evts expected � ∼ 0. 12 background evts expected Missing - plot Delta. Phi First W →τν candidate is just around the corner! Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 6

� � � We are participating to the preparation of a Tau Performance CONF Note. The goal is to prove the good agreement data/MC for tau identification variables, considering only tau candidates. The W→τν group wants to contribute with the same comparison but “biasing” the selected sample toward the sample we will select in our analysis, i. e. cutting on missing E T. Standard cut flow: § GRL § Trigger L 1_J 5 § Collision candidate & Jet cleaning § MET cut: MET_Topo > 15 Ge. V § Tau Reconstruction: the leading of the merged taus must have ET>15 Ge. V and not in crack region DATA: run 153565 MC: QCD di-jets samples Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 7

� Z→ττ: motivations � Control samples for channels containing τ � Tau-jet scale determination � Tau-jet ID efficiency determination � Missing ET scale determination � Measurement of the cross section pp →Z x BR(Z→ττ→lτhad ) � MC analysis (aimed at a ∼ 100 pb-1 scenario) - Visible mass analysis: � Statistic preferred to purity � No cuts on ET Miss (the analysis must be done before the ET Miss scale is set) � Reconstruct the invariant mass only of the visible decay products (visible mass) → depends on the tau-jet scale and not on the ET Miss scale � The main motivation for this analysis is the tau-jet scale determination � Signature: � Lepton � Tau decaying to hadrons � Missing ET due to three neutrinos produced in the tau decays � Lepton and tau jet with opposite sign � Missing ET pointing in between the phi directions of the decay products Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 8

The event selection depends on the channel (electron or muon) considered: � Trigger: single lepton trigger (mu_10 or e_10_medium) � Lepton preselection: the leptons (both muon and electron) and the tau are preselected requiring a minimum momentum and some loose identifications criteria. The overlap removal is then performed between the selected objects. Finally, the leptons and taus are selected and further requirements on lepton isolation are applied. � Z/W backgrounds rejection: further cuts, both topological and kinematical, are performed to reject background events coming from Z and W � QCD background rejection � Cut on visible mass in the window [35, 75] Ge. V � � � With such a selection, expected at the end of the cut flow, in 100 pb-1: 968 signal events, with a ratio S/B = 2. 95. If the analysis is restricted to the visible mass window [35, 75] Ge. V, in 100 pb-1: 914 signal events, with a ratio S/B = 4. 85. For further details about event selection, see back-up slides Lidia Dell'Asta � Open issue: QCD di-jets background: Statistics available is too low (used JF 17) Need to use some factorization methods (on isolation and tau ID) Sestri Levante, 14 Giugno 2010 9

Studies on 10 Te. V MC data � Strategy: reconstruct the invariant mass of the visible decay products, proportionality to the tau-jet scale � � Reference from Monte. Carlo Truth: 58. 86 Ge. V, 1% systematics (different MCs) Keep control over the residual backgrounds � Evaluate bg, especially QCD � Subtraction procedure: OS-SS events (delicate for W+jets) Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 10

For first data analysis and early data/MC comparison: � Stream selection: Muon channel: Muonsw. Beam Stream + DESD_SGLMU skimming Electron channel: L 1 Calo/L 1 Calo. EM Streams + DESD_SGLEL skimming � Good. Run. List selection: combined GRL from muon/egamma + tau + MET � Collision candidate & jet cleaning � Modified cut flow: Trigger: L 1_MU 0/L 1_EM 2 No isolation required Loosened identification criteria (on all objects) Only background expected with this selection � � � Electron channel: 33 events (Lumi = 5. 1 nb-1) � Muon channel: 7 events (Lumi = 6. 4 nb-1) � Signal events expected: � ∼ 0. 05 evts expected from the benchmark analysis � ∼ 0. 19 evts expected for the loosened analysis Lidia Dell'Asta For further details about event selection, see back-up slides Sestri Levante, 14 Giugno 2010 11

� � � Our analysis (both W and Z) is based on compiled C++ code running on D 3 PDs. � Have to run locally → need to copy to Milano all the datasets (both data and MC) → need to subscribe everything to Milano → suffering from disk space problems @CNAF. � This scheme will not work soon (too much data) → have to lern to use prun (not managed so far… The datasets we use (Tau Performance D 3 PDs) are produced centrally � Sometimes we have to wait for the central production: data – shifter’s task MC – managed by volounteers (L. D. A. and another Ph. D from Friburg) � No Tau disk space in Italy: we should have asked before… � As they are common to the whole Tau group (both performance and analysis), D 3 PDs are becoming bigger and bigger. We should think about having different D 3 PDs for different analysis. MC D 3 PDs from April reprocessing > 1 TB ! Need some more MC samples produced with different generators (i. e. POWHEG), as we are using only Pythia sample produced centrally. Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 12

� Milano is working on the analysis of taus both from W and from Z. � We are deeply involved in the commissioning activities of the ATLAS Tau Performance group � We have updated our analysis for the 7 Te. V data analysis � We have started the comparison between data and MC and we keep on updating our results as new data is available � We are participating to the first tau CONF Note with 7 Te. V data with a chapter dedicated to tau reconstruction performance for the W→τν analysis � We are involved in the production of the common analysis ntuples from MC Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 13

Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 15

Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 16

Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 17

Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 18

Lidia Dell'Asta Sestri Levante, 14 Giugno 2010 19