Status Report Gerrit Van Onsem New Physics meeting

Status Report Gerrit Van Onsem New Physics meeting 16/02/10

Outline ( p distribution and p 95 for TTJets + LM 4 (last meeting: LM 0 to LM 3) [for 2 and 3 variables] ) Added 'approximate' uncertainty of 15% on estimation to see effect on TTJets + LM 1 distributions [3 variables] Without this 15% uncertainty: added one more variable (Angle between hadronic top and muon) to see evolution of the distributions for LM 1 w. r. t. 3 variables (MET, Muon. Pt and Jets. Ht) from last meeting Observables not yet chosen from a ranked table Documented (and committed) classes in Stat. Procedure Tools to run over grid: being worked on Auto-binner not yet implemented 2

Pseudoexperiments on TTJets+LM 4 with 2 variables Adding LM 4 (XS = 6. 70 pb) to TTJets (XS = 414 pb): 500 ps. exps. [MET, Muon. Pt] [Landau function to fit V distribution, not made use of this] p 95 = 0. 83 Adding LM 4 shifts p values to lower values – as expected 3

Pseudoexperiments on TTJets+LM 4 with 3 variables Adding LM 4 (XS = 6. 70 pb) to TTJets (XS = 414 pb): 500 ps. exps. [MET, Muon. Pt, Jets. Ht] p 95 = 0. 781 Adding sensitive variable: p 95 smaller 4

Adding approximate uncertainty to estimation with 3 variables: TTJets Adding approximate uncertainty to 'estimation' of TTJets histograms, faking bkg estimation methods: binerror = sqrt( bincontent + (0. 15*bincontent)² ) 2000 TTJets ps. exps. [MET, Muon. Pt, Jets. Ht] -> reference V distribution with addition of 15% uncertainty on estimation without addition of 15% uncertainty (from last meeting) V values lower (variances in denominator of weight of event) 5

Adding approximate uncertainty to estimation with 3 variables: TTJets+LM 1 Adding LM 1 to TTJets: 500 ps. exps with addition of 15% uncertainty on estimation p 95 = 0. 454 V values also lower, but p values more shifted towards 0 (w. r. t. no addition of 15% uncertainty). . . ? Plots at the right from last meeting (no addition of 15% uncertainty): without addition of 15% uncertainty p 95 = 0. 689 6

Pseudoexperiments on TTJets with 4 variables Adding variable: Angle between hadronic top and muon Chose some binning (binner not yet implemented) 'Estimation' redone and 2000 TTJets ps. exps. redone with 4 variables (no addition of 15% uncertainty). 'Estimation' for the new variable V distribution for 2000 TTJets pseudoexperiments [ Had. Top = (selected. Jets[0] + selected. Jets[1] + selected. Jets [3]) Lepton = selected. Muon[0]; Variable = ROOT: : Math: : Vector. Util: : Angle(Had. Top, Lepton) ] 7

Pseudoexperiments on TTJets+LM 1 with 4 variables Adding LM 1 to TTJets: 500 ps. exps. Example distribution for a TTJets + LM 1 pseudoexperiment for the new variable Compare slide 6 (bottom right) p 95 value for these 4 variables > for 3 variables. . . p 95 = 0. 783 8

To be done Update macros to run over variables with Observables class Work on tools to run the code on the grid (needed to proceed in efficient way, but couldn't finish it by now. . . ) Implement auto-binner Apply all this to highest ranked observables from NPVTable + see evolution of p 95 with #observables, effect of estimation uncertainty, binning. . . for different LMX points 9