Data driven SUSY background estimation in one lepton mode

1
Data driven SUSY background estimation in one
lepton mode

• A. Koutsman, W. Verkerke
• 31 may 2007
• Simpelveld

2
One lepton mode SUSY
1 fb-1
Effective mass (GeV)

• Dominant backgrounds
• Top pair
• Wjets
• QCD
• Zjets

• GOAL estimate and understand backgrounds from
  data
• TARGET Develop methods to discover/exclude SUSY
  with 1 fb-1

3
Method-1 (S.Asai, K.Oe) CSC Note1/2

• Main idea separate data in two sets
• MTgt100 signal region
• MTlt100 control region
• Assumption-1 the shape of BG in control region
  is same as shape of BG in signal region ? Just
  need to scale with events
• Assumption-2 SUSY is negligible in control region

Top Wnjets
Top Wnjets SUSY
Estimated signal region X scaling
Works without SU3 in the game ?Assumption 1
is fairly good
Actual BG can be estimated correctly
Estimated BG over- estimated by factor 2
Problems with SU3 ?Assumption 2 is no
good
(Kenta Oe Shoji Asai)
(Kenta Oe Shoji Asai)
4
Improved Method-1

• Main Idea
• Look at 2D distributions
• Account for correlationsif necessary
• Account for SUSYcontamination in control region
• Extrapolate Wjets/Top background in 2D

Condition for success of improved Method-1
Variables must be uncorrelated or must have
(small) understandable correlation. If this is
not the case we cannot reliably extrapolate
estimate from control region to signal region.
5
Correlation

• Howto Find a factorizable (conditional
  dependence if necessary) p.d.f. for each sample

MT
MT
MT
MT
FACTORIZABLE?
Look at shape of missing ET in slices of MT and
observe if shape changes. Find conditional
dependence or prove that distributions are
uncorrelated
CONDITIONAL?
6
Factorizable and Conditional

• Semileptonic top, dileptonic top and Wjets are
  factorizable (low statistics for dileptonic
  sample)
• SUSY more complex, but no extrapolation needed

Wjets
Wjets
Fit an exponential to each slice Look at
distribution of slopes
Wjets

• Missing ET and MT are uncorrelated.
• No need to even consider a pdf of ET with a
  conditional dependence on MT.

7
Fitting PDFs
Semileptonic

• How do you model the background samples
• Semileptonic
• (11215 parameters)
• exponential in missing ET
• exponentialgauss in MT
• Dileptonic
• (123 parameters)
• exponential in missing ET
• gauss in MT
• Wjets
• (11215 parameters)
• exponential in missing ET
• exponentialgauss in MT

Dileptonic
Wjets
Problem likelihood fit with negative weights
8
Fitting to Data
Semileptonic

• For the combined background sample
• (3 in total 16 parameters)
• 3 fractions ( events each sample)
• First simple try
• get fixed shapes from MC-prefit to individual
  samples
• only float the fractions

Dileptonic
Fit
Truth Nsemi 423 35
432 Ndi 2111 167
1988 Nwjets 588 163 750
Wjets
Floating Parameter FinalValue Error GblCorr.
-------------------- ---------------------
Num_TDil_sample 4.2277e 3.45e
0.380082Num_TSemil_sample 2.1108e 1.67e
0.945936Num_W_sample 5.8834e 1.63e
0.946130 PARAMETER CORRELATION COEFFICIENTS
NO. GLOBAL 1 2 3 1
0.38008 1.000 -0.055 -0.080 2 0.94594
-0.055 1.000 -0.937 3 0.94613 -0.080
-0.937 1.000
Full sample

• Combine Wjets and Semileptonic top to one
  subsample

Fit
Truth Nsemiwjets 438 35
432 Ndi 2684 589 2738
9
Fitting to Data (2)
Semileptonic

• Float not only fractions, but also shapes of 3
  sub-samples in a combined fit

• Sub-samples not fitted correctly, high
  correlation coefficients, not stable
• More Work Needed!

Dileptonic

• Ideas to proceed
• 1) 3D fit to deal with high correlation in
  simple 3 component model
• 2) simplify shape floating fit for stability
  (W-mass constraint)
• 3) iterative procedure fit first fractions
  (shapes fixed from MC), then float shapes but fix
  fractions
• 4) fit with SUSY

Wjets
Full sample
10
SUSY Future

• SUSY distribution is flat in control region in
  both variables
• Add a flat pdf to the fit to deal with SUSY
  contamination
• Once background fitting is up and running, do toy
  MC study to check reliability stability of the
  fit

Alex J. Koutsman Wouter Verkerke

• Future (after CSC)
• Study SUSY theoretically (SuperJournal Club)
• Get more acquainted with different SUSY scenarios
  (18 models produced for CSC)
• Get ready for data streaming data, no MC

11
(No Transcript)
12
(No Transcript)