User Tools

Site Tools



A quick study on the impact of irreducible-background subtraction

If you are interested in contributing please subscribe to the mailing list. You will be asked to log in with your CERN account. If you don't have a CERN account, please fill this form to create a lightweight account.

Jon Butterworth, Vitaliano Ciulli, Paolo Francavilla, Frank Krauss, Carlo Pandini, Luca Parrozzi, …

github repository: (see below for details)

Processes to consider:

WW versus ttbar

14 TeV (maybe also 13 TeV), B-tag up to y=2.5 (allowing a veto on b-jets in this region if it helps). Also look at the impact of pseudo-top vetos.

Double leptonic channel (eμ only):

1. Consider WWjj as the signal. Look at WW scattering-like WWjj topologies. Study the contribution to the signal coming from double-resonant (DR), single-resonant (SR) and continuum (C) ttbar processes, and the sum.

2. Consider WWj or WW as the signal. Look at jet-binning for for all WW (not just the VBS topology). Same study (DR/SR/C/Sum) The ATLAS paper is here: The cuts which define the fiducial phase space are:

  • pT μ > 20 GeV (muon or electron), leading lepton pT > 25 GeV
  • muon |η| < 2.4, electron |η| < 1.37 or 1.52 < |η| < 2.47,
  • no jets with p > 25 GeV, rapidity |y| < 4.5 and separated from an electron by ∆R > 0.3
  • pT (ll') > 30 GeV
  • m(ll′) > 10 GeV, pT,Rel > 25 GeV

I (Jon) suggest we simplify this to something like:

  • pT leptons all > 25 GeV
  • |η| leptons all < 2.4
  • m(ll′) > 10 GeV
  • count the jets, and b-jets, with p > 25 GeV, rapidity |y| < 4.5

… but I haven't check the CMS paper yet.

3. Consider WWbb as the signal, motivated by WWH (H→bb). Same study (DR/SR/C/Sum). For HH→WWbb, a possible proposal could be:

Objects Definition:

  • electrons and muons: pT > 25 GeV; |eta| < 2.4
  • jets: pT > 25 GeV; | eta | <4.5
  • bjets: jets with | eta |< 2.5 and b-labelled

NOTE 1: in the ATLAS WW analysis, the jets are build from all the particles, excluding muons and neutrinos. If we adopt this definition, we will need to run an Overlap Removal between electrons and jets, to avoid counting jets which are in fact electrons. NOTE 2: The ATLAS WW analysis is vetoing events if (among the other selections) :

  • there are jet (and b-jet), or
  • there are extra leptons.

If we want to extend to events with b-jets, we should consider how to handle the semi-leptonic b-decays in jets ()to avoid to veto the event because of the leptons in jets.

Event Selection: fill()CutFlow

  • 2 opposite charged leptons
  • 1 muon, 1 electron

fill(CutFlow) fill(MET_rel)

  • MET_rel>25 GeV,

fill(CutFlow) fill(mll)

  • mll>10 GeV,

fill(CutFlow) fill(b-jets, jets) #b-jets VS #jets fill(jets), pt, eta

  • 2+ bjets

fill(CutFlow) fill(bjet 1) pt, eta fill(bjet 2) pt, eta fill(electron) pt, eta fill(muon) pt, eta fill(MET) fill(mT(MET,e,mu))

  • 100<mT(MET,e,mu)/GeV<150

fill(CutFlow) fill(electron) pt, eta fill(muon) pt, eta fill(b-jet1) pt, eta fill(b-jet2) pt, eta fill(m(bb))

  • 100<m(bb)/GeV<150

fill(CutFlow) fill(b-jets, jets) #b-jets VS #jets fill(jets), pt, eta

NOTE: it would be nice to use some top veto. the pseudo-top definition could be a nice idea, but it is defined in a straightforward way for the Semi-leptonic channel. One can test the b-lepton mass (associating the b to the closer lepton).

Semi-leptonic channel:

4. Consider ttbar as signal. Look at distortion in b-lepton mass from SR/C/Sum/WWbb-non-top contributions.

Wb versus t

Link to the ATLAS paper:

This would be a nice demo of some of the ideas above with a real measurement. Treat Wb excluding t as the signal, use the ATLAS Wb (unsubstracted) measurement and re-evaluate the single-top subtraction and systematics.

The paper reports an measurement (2011 data) of Wb, in two versions. In the one case, single top is considered as part of the signal; in the second it is subtracted.

Here's the plot of the unsubtracted measurement, with the predicted single-top contribution shown:

Integrated over pT, the measurements are:


  • σfid (1 jet) = 5.9 ± 0.2 (stat) ± 1.3 (syst) pb, (fractional syst = 22%)
  • σfid (2 jet) = 3.7 ± 0.1 (stat) ± 0.8 (syst) pb, (fractional syst = 22%)
  • σfid (1+2 jet) = 9.6 ± 0.2 (stat) ± 1.7 (syst) pb.(fractional syst = 18%)


  • σfid (1 jet) = 5.0 ± 0.5 (stat) ± 1.2 (syst) pb, (fractional syst = 24%)
  • σfid (2 jet) = 2.2 ± 0.2 (stat) ± 0.5 (syst) pb, (fractional syst = 23%)
  • σfid (1+2 jet) = 7.1 ± 0.5 (stat) ± 1.4 (syst) pb.(fractional syst = 20%)

So there is a small but noticeable effect. The main contributions to the systematic errors quoted in the paper are:

  • Jet energy scale 10-50%
  • ISR/FSR, including on single top and ttbar 2-30%
  • b-tagging 1-8%
  • MC modelling (but only of the Wb “signal”) 2-8%

So I guess the fact that JES dominates is why the effect is fairly small. The “ISR/FSR” thing, which should be reduced for the unsubtracted measurement, varies a lot with jet pT. Indeed, if you compare Table 4 with Table 9 in the paper, you can see this. In the highest pT bin the systematic uncertainty goes from 16% before subtraction to to 54% after it.

github instructions

To setup your local repository

  1. go to and get an account
  2. In the top-right corner of the page, click Fork
  3. open a shell where git is available
  4. clone the repository with the command: git clone
  5. enjoy

To start contributing

  1. modify/add a file
  2. add file(s) to local repository: git add filename.bla
  3. commit file(s) to local repository: git commit -m “commit messate” filename.bla
  4. “push” modifications to your remote (i.e. github) repository : git push
  5. inform Luca to include your modifications to the main repositiry

To synch your repository with the main repository

2015/groups/tools/backgrounds.txt · Last modified: 2015/07/14 08:31 by philippe.gras