220 likes | 326 Vues
This study analyzes three models (InertHiggs, SUSY, LHT) at 500 GeV with detailed event generation, simulation, and analysis. The models involve WIMPs and visible particles, with mass and angle calculations for production. Signal and background events are generated using specific tools, with selection and efficiency cuts applied. Mass determination, production angle calculation, discriminant distributions, and separation power evaluation are key components. The analysis also investigates uncertainties related to statistics and mass measurements, showcasing good separation power between models. Additionally, the study explores threshold scan strategies to understand cross-section responses near energy thresholds for different models.
E N D
Model ID @ 500 GeV Taikan Suehara ICEPP, The Univ. of Tokyo
Three models • InertHiggs • WIMP: hI (scalar) • Visible: h±(scalar) • SUSY • WIMP: c0 (fermion) • Visible: c± (fermion) • LHT • WIMP: AH (vector) • Visible: WH± (vector) All give 2W + 2WIMPfs.
Analysis overview • √s = 500 GeV, 500 fb-1 • full simulation • NO Initial polarization • Use same massesfor all the models m(visible) = 231.57 GeV, m(WIMP)=44.03 GeV • Cross sectionis normalized (200fb & 40fb). • Use W hadronic decay(cut leptonicevents) • Mass-> production angle calculation • Full SM backgroundincluded Weighted to 500fb-1
ILDdetector &framework Event generation(signal, SM background) Stdhep event files Full-MC simulation(GEANT4) Hit, MCinfo(LCIO file) • Tracking • Particle Flow • Flavor tagging etc. Event reconstruction ILDdetector(ILD00) • Vertex: Si pixel (3x2 layers) • Silicon Tracker: 4 layers • TPC(main tracker) • ECAL (0.5x0.5 cm tile, Si/W) • HCAL (3x3 cm tile, Sci/Fe) • Solenoid(3.5Tesla) • Muon detector Particle information (ROOT tree) Analysis • Event weighting • Separation cuts • Analysis (mass, angle…)
Event generation • Signal: JSF/physsimStructure • HELAS for helicity amp. calculation • BASES for numerical integration • SPRING for event generation • Pythia for quark fragmentation • ISR/Beamstrahlung included • ~0.1 M events for each process generated • Background: SLAC (ILCstandard sample) • whizard/pythia for all SM processes • ~14 M events in total (80-200 fb-1 for 2/4/6 fermion events, 0.1-1 fb-1 for eg/gg events)
Selection cuts BG suppression cuts “qqqq (non-b) + missing” • 4-jet clustering (Durham) • # Track>=20 • 160 < Evis < 400 GeV • each Ejet > 5GeV • |cosq|jet < 0.99 • unlike 3jet(yth,3 > 0.001) • each jet has >= 2tracks • no > 25 GeVleptons • |cosq|miss < 0.9 • sum |cosq| < 2.6 • sum b-quark prob. < 1 • kinematic fit converged • 65 < mW < 95 GeV
Selection efficiency • IH eff: 59.8%, pur (200 fb): 78.6%, pur (40 fb): 42.3% • SUSY eff: 58.3%, pur (200 fb): 77.7%, pur (40 fb): 41.0% • LHT eff: 58.9%, pur (200 fb): 78.2%, pur (40 fb): 41.8% Acceptable selection performance
W energy distribution (for mass) MC truth (signal only) Reco (kin-fitted) MC & reco have a good correlation.Difference of shapes (esp.WH) is notused for the discrimination in this analysis.
Mass determination • Fitting: 3rd polynomial x Voigt function (10 params) • All 10 params free • Fix 8 params (except edge positions):obtain errors of the edge positions
Mass resolution • Edge positions from the fit: (MC: 96.32/174.78) • Masses (MC: 231.57/44.03)
Production angle • Production angle of the visiblenew particle can be calculatedfrom W directions with massesof new particles andback-to-back assumption. • The angle has spin info.but two-fold ambiguity (solution of 2nd equation) exists.
Discriminant distribution • The distribution looks reasonable. • ~30% of the events cannot be used.(discriminant > 0 events selected)
1D distribution 200fb • MC & reco have reasonable correlation. • Three models seem to be discriminated.
2D distribution (2 solutions) IH SUSY SMBG LHT Chi-square can be derived using these distributions.
Separation power • Powerful for larger templates • Confirmed by comparing “same” model but statistically independent samples • Uncertainty exists with finite statistics • s reflects the statistical fluc. of templates between Di and Ti
Separation power results • Good separation with 200 fb cross section • Nearly 3s separation with 40 fb • IH/SUSY has better separation • Consistent with intuitive view Uncertainty from statistics : ~2 Uncertainty from statistics : ~.3
Effect of mass measurement • Uncertainty of masses can cause uncertainty on the angular distributions • We are examining such effects by simulation with shifted masses (full or generator level) IH 40fb IH 200fb
Results… • With full simulation, no significant deviation can be seen between shifted and original samples(within MC statistics…) • Higher statistics with all shifted masses is not practical • Generator analysis has not been finished… • Need to ‘adapt’ the MC distributions to reco ones… (with some coefficients) • Uncertainty of that adaptation should be also investigated…
Threshold scan • Cross section response to beam energy near the threshold reflects spin structure • SUSY has (ECM-ETH)0.5 response whileIH/LH has (ECM-ETH)1.5 response.
Strategy for threshold scan • Assume 40 fb cross section • 470, 500, 530 GeV with 50 fb-1 stat. each • Threshold energy is treated unknown • SM bg and signal efficiency is assumed to be the same as 500 GeV • Toy-MC with 3-point fittings • Examine c2 of the fitting • SUSY should give low c2 for n=1/2, high c2 for n=3/2 • Opposite for IH/LH
Chi-square results • Power of 0.5 result shows good separation between SUSY/other two models Power of 0.5 Power of 1.5
Summary/status • Analysis almost finished • Cut optimization / mass fitting • Model ID with 2d angular distribution • Some work left in the mass shift analysis • Threshold scan • Paper writing is ongoing (at final round?) • LCWS proceedings for both theory and 500 GeV experiment (this work) are already available • Need consistency between 500 GeV and 1 TeV work