1 / 15

Data-driven Wln and Znn backgrounds Estimating the backgrounds in the control sample

This presentation discusses a data-driven approach for estimating the backgrounds in the control sample (CS) for W+lnu and Z->nunu processes. The method utilizes the kinematic acceptance, ID efficiency, and W/Z ratio to apply weights to each CS event. The benefits of this data-driven approach include reduced JES/JER systematic and no luminosity systematic. The presentation also explores the use of different control samples and uncertainties in the estimation.

woodsa
Download Presentation

Data-driven Wln and Znn backgrounds Estimating the backgrounds in the control sample

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. Data-driven Wln and Znn backgroundsEstimating the backgrounds in the control sample Alex Pinder University of Oxford W/Z/Top informal meeting 29 March 2011

  2. Recap/motivation • Use Wen and Wmn control samples (CS) to get V+jets background (BG) normalisation and shape from data • To each CS event, apply a weight based on the CS kinematic acceptance, ID efficiency, and W/Z ratio • Benefits of DD approach: • Much reduced JES/JER systematic • No luminosity systematic • Measure shape as well as normalisation Weν BG Zνν BG Weν CS W/Z background (0-lepton)

  3. Backgrounds-within-backgrounds • The 1-lepton control sample is as follows: • 1 tight lepton (e/μ) as defined by SUSY obj defs • No other “medium” leptons • MET > 25 GeV • 40 GeV < MT < 80 GeV • Plus jets, Meff, Δφ etc as required by 0-lepton search • But not 100% Wlν: • Large amount of ttbar • Leptonic Wτν • Also QCD (in e channel) W/Z background (0-lepton)

  4. 2010 approach • All backgrounds in the control sample ignored, except ttbar • Subtract this using MC W/Z background (0-lepton)

  5. 2011 approach? • Have tried a simultaneous fit • Inspired by the 1-lepton SUSY analysis • Three control samples: • Wlν + Wτν (b-tag veto, tight lepton) WCS • Top (b-tag, tight lepton) TCS • QCD (medium-but-not tight electron) QCS • Control samples defined by object rather than kinematic cuts • Uncertainties from efficiencies and fake rates W/Z background (0-lepton)

  6. Method • For each signal region, count events in each control sample: • Can relate these to the true number of W/top/QCD by: • Matrix elements taken from simulation • Invert the matrix to find the result we want: W/Z background (0-lepton)

  7. Example: electron control sample W/Z background (0-lepton)

  8. Example: electron control sample W/Z background (0-lepton)

  9. Example: electron control sample W/Z background (0-lepton)

  10. Muon control sample • Muon channel has ~ 0 QCD contamination • Require only two control samples (and 2×2 matrix) W/Z background (0-lepton)

  11. Electron 2 control sample • Have another electron control sample where the electron-matched jet is not removed • Also electron isn’t “neutrinofied” • Used to estimate Weν b’ground where electron not identified • Again, QCD contamination is minimal W/Z background (0-lepton)

  12. Test on data (35 pb-1) WCSTCSQCS WCSTCS Electron (1) Muon • Reasonable results, consistent with the MC • Electron control sample (#1) – predicted W fraction in WCS: • Signal region 1: 0.94 • Signal region 2: 0.99 • Signal region 3: 0.92 • Signal region 4: 1.07  Because no events in TCS W/Z background (0-lepton)

  13. Wτν contamination • How to deal with this? • When estimating Wlν background, do not remove it • Get estimate of W  τν  lνν background for free • When estimating Zνν, remove it using MC • Fine if systematic error is huge W/Z background (0-lepton)

  14. Open questions • Are these the best control samples to use? • How best to estimate the effect of uncertainties? • B-tag efficiency / fake rate • Loose electron efficiency / fake rate • Tight electron efficiency / fake rate • Muon efficiency • Is there a simpler way? W/Z background (0-lepton)

  15. Final thoughts • Have presented a potential method to remove backgrounds from the W control samples • Uses a simultaneous fit to three sub-control samples • Seems to work rather well so far • But not clear yet how to handle systematics • People listening may have expertise here • When we estimate the W background, do we want to remove ttbar from the control sample? • If we leave it in, we estimate the ttbar  missed lepton background as well • Can subtract the W component to get an independent check on existing ttbar estimation techniques W/Z background (0-lepton)

More Related