March 2005Sarah Allwood WW Scattering at ATLAS Sarah Allwood University of Manchester IOP HEPP conference 2005, Dublin

March 2005Sarah Allwood Introduction W L W L → W L W L is described at low energy by an effective Lagrangian: the EWChL. a 4 and a 5 parameterise the “new physics”. EWChL made valid up to higher energies by unitarity constraints: this can predict resonances ~1 TeV in WW scattering. Map of a 4 -a 5 space obtained using the Padé unitarisation protocol. Taken from hep-ph/ J.M. Butterworth, B.E. Cox, J.R. Forshaw. Without some new physics W L W L → W L W L violates perturbative unitarity at E~1.2 TeV.

March 2005Sarah Allwood Signal Scenarios Five representative signal scenarios were chosen: A: scalar resonance of 1.0 TeV, B: vector resonance of 1.4 TeV, C: vector resonance of 1.9 TeV, D: double resonance of a scalar at 800 GeV and a vector at 1.4 TeV, E: scenario with no resonances (the continuum). This follows on from the hadron level analysis in hep-ph/  How sensitive is ATLAS to these resonances in WW→WW→l qq ?  Optimum jet parameters to use?

March 2005Sarah Allwood Signal and Backgrounds Backgrounds: W+jets (W  l ), σ~60,000 fb, and, σ~16,000 fb (cf signal σ<100 fb). high p T lepton high E T miss Jet(s) with high p T and m ~ m W. Little hadronic activity in the central region (|η|<2.5) apart from the hadronic W. Tag jets at large η (|η|>2). high p T W

March 2005Sarah Allwood Overview of Analysis Leptonic: Reconstruct W from high p T lepton and E T miss. Reject events with p T W < 320 GeV. Events generated in Pythia (modified to include EWChL), simulated/reconstructed in ATLFAST. Cells smeared before clustering. Pile-up for high/low luminosity added to cells. Cell threshold of 1 GeV for low luminosity, 2 GeV for high luminosity. Underlying event included. Hadronic: Reject events with p T W < 320 GeV. Reject events outside the range m W ± 2σ Environment cuts

March 2005Sarah Allwood Jet Finding K T algorithm (inclusive mode): For each object, calculate  d kl (~p T 2 of k with respect to l)  d kB (~p T 2 of k with respect to the beam direction) Scale d kB by the R-parameter: d k =d kB R 2 If d k < d kl, k is a jet. If d kl < d k, merge k and l (add their 4-momenta) and define this as a new object. Repeat until all objects are in jets. Decide which R-parameter to use based on resolution of hadronic W… beam k l

March 2005Sarah Allwood Hadronic W Mass of the highest p T jet in the event for different R-parameters and cone radii: cone From a Gaussian fit to the peak: Best resolution for cone: ΔR=0.7 Best resolution for k T : R=0.5 (hep-ph/ used R=1.0) kTkT W resolved into 2 jets High mass tail from underlying event and pile-up

March 2005Sarah Allwood Hadronic W Reconstruct hadronic W as 1 k T jet, but follow this with a “subjet” cut. Run k T algorithm in subjet mode on the cells in the highest p T jet. Clustering is stopped at a scale y cut p T 2 → clusters remaining are subjets. Scale at which jet is resolved into two subjets is ~m W 2 for a true W. Make a cut at 1.55 < log(p T √y) <2.0

March 2005Sarah Allwood Hadronic W Use cones of ΔR = 0.2 to find 2 jet centres Sum 4-momenta of all cells within ΔR = 0.4 of jet centres. kTkT cone k T gives higher s/b and better hadronic W resolution than cone 1 jet approach. Similar resolutions achieved for k T and cone (7.2 GeV). But a fairer comparison is given by reconstructing hadronic W as two overlapping jets… (similar to 1 TeV Higgs study in ATLAS TDR)

March 2005Sarah Allwood Environment Cuts Top mass cut: Reject events where m (W+jet) ~ m top 130 GeV < m (W+jet) < 240 GeV Tag jet veto: Require forward and backward jets with E > 300 GeV and |η| > 2. Reduces top background by a factor of 10 Reduces W+jets by a factor of 200, top by factor of 100 s/b = (k T ), (cone) efficiency = 3.1% (k T ), 3.3% (cone) s/b = 1.0 (k T ), 0.9 (cone) efficiency = 1.1% (k T ), 1.1% (cone)

March 2005Sarah Allwood Environment Cuts p T cut: Expect p T (WW+tag jets) ~ 0. Reject events with p T (WW+tag jets) > 50 GeV. Minijet veto: Reject events that have more than one jet (p T > 15 GeV) in the central region. s/b = 1.5 (k T ), 1.6 (cone) efficiency = 1.1% (k T ), 1.0% (cone) s/b = 1.3 (k T ), 1.4 (cone) efficiency = 1.1% (k T ), 1.1% (cone)

March 2005Sarah Allwood Low Luminosity Results For 30 fb -1 : K T s/bCone s/b A:3.3 B:2.2 C:1.9 D:4.2 E:1.5 A:3.7 B:2.5 C:2.1 D:4.5 E:1.6 kTkT cone K T efficiency % Cone efficiency % A:1.40 B:1.33 C:1.25 D:1.13 E:1.06 A:1.40 B:1.36 C:1.24 D:1.10 E:1.01

March 2005Sarah Allwood Low Luminosity Results KTKT Cone

March 2005Sarah Allwood High Luminosity Results K T s/bCone s/b A:3.3 B:2.1 C:1.8 D:3.9 E:1.4 A:3.3 B:2.2 C:1.8 D:4.0 E:1.4 For 100 fb -1 K T efficiency % Cone efficiency % A:1.21 B:1.15 C:1.06 D:0.93 E:0.89 A:1.20 B:1.15 C:1.04 D:0.93 E:0.90

March 2005Sarah Allwood Summary With 100 fb -1 a wide range of resonances can be observed, and their spins measured. k T and cone results give similar efficiencies and signal/background. Final signal/background > 1 in all cases. Using R-parameter of 0.5 instead of 1 has improved signal/background and made some cuts less sensitive to underlying event. Reconstructing the hadronically decaying W as 1 jet followed by a subjet cut is similar to the cone 2 jet reconstruction.

March 2005Sarah Allwood Extra Slides

March 2005Sarah Allwood Underlying Event ATLAS default: mstp(82) = 4 parp(82) = 1.9 parp(90) = 0.16 Switch off multi-parton interactions in Pythia and compare to ATLAS default…