1 / 53

Search for X WZ 0  evjj Paper Seminar

Search for X WZ 0  evjj Paper Seminar. David Toback & Chris Battle Texas A&M Henry Frisch University of Chicago. Outline. Theory and Signature Overview of Analysis; Event Selection and What signal would look like; Acceptance Backgrounds Comparing Data, Signal and Backgrounds

aden
Download Presentation

Search for X WZ 0  evjj Paper Seminar

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Search for XWZ0  evjjPaper Seminar David Toback & Chris Battle Texas A&M Henry Frisch University of Chicago

  2. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  3. XWZ0  evjj • We want to search for new physics in a model independent manner • Many models predict new particles which decay via XWZ0 • The Wen and Z0jj have advantages: • Electrons are straightforward to trigger on and identify • Z0jj has a large branching fraction

  4. Feynman Diagram X 0

  5. Example Feynman Diagram • Heavy Charged Vector Boson: W’ WZ0 • Technicolor Rho: rT WZ0 W’ width, G(W’), can vary greatly Search for X as a function of mass and width 0

  6. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  7. W+2jets Event Selection & Summary • 1 electron • Missing ET • 2 Jets • 110 pb-1 of data from Run 1A and 1B

  8. Looking for Signal • Model X production using W’  WZ0 in Pythia • Look for X WZ0 mass bumps in both Mjj and MW+jj

  9. Overview of Analysis • Constrain PZn using W mass • Reconstruct dijet and W+dijet masses • Look for bumps in dijet vs. W+dijet mass plane using a fit Reconstruction procedure does a good job of reproducing W’

  10. Acceptance vs. Mass Good Acceptance as a function of mass G(X)<< MX

  11. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  12. Backgrounds • W+jets ( W  eν, W τν  eννν ) • Non W+jets • Fakes • `tt • `bt • WW • WZ0 • Z0 ( ee) + jets • Z0 (ττ) + jets

  13. Background Normalization • All but W+jets have an absolute normalization • W+jets has a large normalization error • Take normalization from the data: Ndata= NW+jets + Nother + Nsignal

  14. Summary of Backgrounds Estimated from data { Use PYTHIA and normalize to known cross sections Combination of VECBOS and PYTHIA. Norm to measured Z0ee data Use VECBOS for shape. Large k factor uncertainty. Take normalization from fit to data. Agrees with Duke Group results

  15. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  16. Dijet Mass Distributions • No evidence of Z0 produced in association with a W • W+jets normalized to data and non-W+jets (no signal assumption)

  17. W+dijet Mass Distributions • No evidence of W’ or other new particle production • W+jets normalized to the data and non-W+jets (no signal assumption)

  18. W+dijet in 3 Mass Regions • Use previous normalization and check Z0 region • Data outside Z0 mass region is well modeled telling us that the background estimate inside the Z0 mass region is well modeled (both norm & shape). • No evidence for WZ0 production. ( *Figure 1 in PRL)

  19. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  20. Turning the Crank • Searching the data for X • Look for excess in dijet vs. W+dijet mass plane • Fit the data to signal, W+jets and non-W+jets • Fix non-W+jets background • Allow W+jets and signal to float • Binned likelihood fit in the 2-d dijet vs. W+Dijet mass plane • Normalization mostly comes from outside signal region • Same technique as Dijet Mass bump search (R. Harris) • No evidence for signal (as seen in previous plots and in the fit results) • Get 95% C.L. cross section upper limit from the fit • Incorporate systematic errors

  21. Example Signal Fits I Data vs. background with no signal from “reference model” W’ with a mass of 300 GeV .

  22. Example Signal Fits II Data vs. expectations (back & signal) with best fit amount of signal from reference model W’ with a mass of 300 GeV .

  23. Example: Signal Fits III Data vs. expectations (back and signal): signal level which is excluded at the 95% C.L. (reference model, MW’= 300 GeV) .

  24. Example: Signal Fits IV Data vs. expectations (back & signal) with reference model; theoretical production cross section Excluded at the 95% C.L.

  25. Systematic Errors Use same (conservative) methods as dijet mass bump search and `bb mass bump search • Find the no-systematic 95% C.L. upper limit • Vary background or signal (depending on effect) by +1σ and –1σ and re-fit • Recalculate new limit • Take absolute value of % change in limit (even if the cross section limit goes down!) • Take the larger % of the two variations (+1σ and –1σ) as the % smearing • Take all variations and add them in quadrature • Use this as a Gaussian smearing to the likelihood

  26. Systematic errors Vary both signal and background separately to over-estimate the magnitude of the effect • Amount of non-W+jets (vary background) • Absolute jet energy scale (vary signal) • Energy resolution (vary signal) • Radiation (vary signal) • Q2 scale of W+Jets (vary background) • Structure functions (vary background) • Acceptance (add % error) • Luminosity (add % error)

  27. Systematic Errors • Absolute energy scale dominates the error • Shifts signal into region with lots more background • Checked with Pseudo-Expts

  28. Errors Cont.:Extended Gauge Model • Narrower width = less signal in high background region • Absolute energy scale again dominates the error

  29. Systematic Error Summary • Systematic errors for lots of effects • Conservative estimation methods • We are not pulled unreasonably by an unexpected fluctuation in the data • Data is well modeled • Set limits

  30. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  31. Setting Limits • We set generic 95% C.L. cross section limits on X production as a function of mass and width • Use W’ production as a good approximation • Use W’ production as a model (determines production cross sections) and set mass limits • Begin with an theoretical overview of W’

  32. Reference Model • Simplest W’ Model • W’ is the same as W only heavier (same couplings to quarks and leptons) • No new neutrinos • Call this “Reference Model” • Consequences • Large production cross sections • Γ(W’  WZ0)  M5W’ • Large branching fraction to WZ0 • Large total width, Γ(W’) • Model becomes unphysical at approx. MW  Γ(W’) which occurs at approx. MW’  425 GeV/c2

  33. Extended Gauge Model • Simplest W’ model unphysical (can be no W’WZ0 vertex in SM) • Simplest extension is W’-W mixing as in extended gauge models (e.g. L-R symmetry) • Effective W’WZ vertex; same as in reference model but vertex multiplied by  , which is estimated (non-trivially) by  = C(MW / MW’)2 where C is of order 1 • Γ(W’ WZ0)  MW’ • Narrow width  Small Br; most previous W’ searches assume this (e.g. W’  eν ) • Use  = C(MW / MW’)2 as general Г(W’) << MW’ • Large production cross section

  34. Theoretical Consequences • Comparison of the reference and extended gauge models • Drastic differences in width and branching ratios Wide Widths Small Br Narrow Widths Large Br

  35. Branching Ratio for W’WZ0 • Reference Model • W’ is the same as the SM W only heavier • Large width large branching ratio • Extended Gauge Model • Mixing factor between W and W’ • Small width • Small branching ratio

  36. Mass Dist. For Reference Model • 1000 PYTHIA generated events for the reference model • Width increases as a function of mass

  37. 95% C.L. Limits: Reference Model • We exclude the reference model of W’ from 200 to 480 GeV. • Taken in conjunction with low mass exclusions from the W’lν , we exclude the entire model

  38. 95% C.L. Limits: Ext. Gauge Model • 95% C.L. upper limits on cross section vs. W’ mass for the extended gauge model • No mass limits for very small factors (branching ratio is tiny) • Cross section limits applicable for any new particle production with narrow width XWZ0

  39. Cross Section vs. Mixing Factor 95% C.L. upper limits on cross section vs. W – W’ mixing factor

  40. Cross Section vs. W’ Width • 95% C.L. upper limits on the cross section vs. W’ width • These limits are good for any new particle production with XWZ0; narrow or wide width * PRL Figure 2

  41. Limits on Mixing Factor vs. W’ Mass 95% C.L. exclusion region for W-W’ mixing factor vs. W’ mass * PRL Figure 3

  42. Outline • Theory and Signature • Overview of Analysis; Event Selection and What signal would look like; Acceptance • Backgrounds • Comparing Data, Signal and Backgrounds • Fitting and Systematics • Limits • Conclusions

  43. Conclusions • No evidence forXWZ0in the enjjdecay channel • Narrow and width width approximations • Most comprehensive limits on direct W’ WZ0 • Reference model completely excluded • Large exclusions in an extended gauge model • Web page at -hepr8.physics.tamu.edu/hep/wprime/ • Documentation in CDF Note 5610 • PRL draft in CDF Note 5629

  44. Backups

  45. Acceptance vs. W’ Mass • Good Acceptance for W’ • Reference Model • Large width at large mass • Lots of low mass events • Lower acceptance

  46. Pseudo-Experiments: Check Re-run entire analysis on fake data generated from backgrounds only • Generate fake data set • Allow number of events to float • Re-estimate the effect of all systematic errors for the fake data set • Add errors in quadrature as for data • Re-estimate the limit from the fake data set • Repeat many times • Repeat for different masses and mixing factors

  47. Pseudo-Exp: Jet Energy Error • The effect on the limit (in %) of the jet energy scale uncertainty for a set of pseudo-experiments with W’ mass of 200, 300, 400, 500, & 600 GeV respectively. This is for the extended gauge model with C=1.

  48. Pseudo-Experiments: Total Error • The total effect on the limit (in %) due to all systematic uncertainties for a set of pseudo-experiments with W’ mass of 200, 300, 400, 500, & 600 GeV respectively. This is for the extended gauge model with C=1.

  49. Pseudo-Experiments: Limit • 95% cross section upper limit from a set of pseudo-experiments with W’ mass of 200, 300, 400, 500, & 600 GeV respectively. This is for the reference model.

  50. Pseudo-Experiments: Total Error • The total effect on the limit (in %) due to all systematic uncertainties for a set of pseudo-experiments with W’ mass of 200, 300, 400, 500, & 600 GeV respectively. This is for the reference model.

More Related