# Stat

Document Sample

```					Few activities in Statistics:
Higgs + Exotics

Samir FERRAG
University of Glasgow
Introduction
• Very useful Stat Workshop at CERN last week
• Interacted and triggered some projects:
– Alex Read (Shape analysis for ttH, Saverio)
– Kyle (Stat Framework for ATLAS based on Roofit)
–
• Feasibility of ttH
– Extract the constraints on the background uncertainty
– If feasible: define the priorities on real data:
• PDF, Background, ttH…

– Counting experiment constraints to be extended to
shape analysis
Theoretical uncertainties on ttbb
Relative uncertainty (%)

High-x gluons are not the main
Source of uncertainties ~3%

PDF number

• ttbb production via gluon-gluon and quark-antiquark fusions
• ttbb cross section from AlgGen: sttbb= 2.55 pb
• Theoretical uncertainties:
+/- 6%     from PDFs (classical recipe)
+/- 55% from factorisation and renormalisation scales 1/2m0 <m<2m0
• PDF uncetainties: can be irreducible and constitutes a serious source to
constraint the analysis
constraint on ttH
• For a relative systematic uncertainty  on the background, the significance is:

S
N
(1   2 B) B

• By extracting the relation between S/B and , we obtain the constraint relation:

1                      2
1       2   4
S         S            N
B       
S              2 2
Constraint on ttH for 10fb-1
• Higgs mass=120GeV
S/B

5s

4s
Requested S/B
For 3s significane
For 6% PDF uncertainty

3s

S/B=0.14 From                    2s
Markus analysis
(ATLFAST)
systematic uncertainty on B (%)

6% of PDF uncertainties (on ttbb+ttjj)
Need 1 to 2 % of accuracy
On background to reach
2s significance
Constraint on ttH for 100fb-1
• Higgs mass=120GeV
S/B

5s

4s

3s
Requested S/B
For 3s significane
For 6% PDF uncertainty                                               2s
S/B=0.14 From
Markus analysis
(ATLFAST)

systematic uncertainty on B (%)

Need 4% of accuracy 6%   of PDF uncertainties
On background to reach
3s significance
Variation with the Higgs mass
• Higgs mass=130GeV   • Higgs mass=120GeV

• 10 fb-1                                                   5s
4s
3s
2s
1s

ATLFAST
analysis

• 100 fb-1

ATLFAST
analysis
Conclusion: Counting
experiment…!
• Is ttH feasible ?
– Reduce the PDF uncertainties to << 6%
• if not:
PDF uncertainties are enough to cancel the discovery power
– Background (ttbb+ttjj) measurement accuracy at < 5%
•   Including PDF uncertainties above (?)
•   Systematic uncertainties from Luminosity (?)
•   Reconstruction and performance tools… (?)
•   If Likelihood, Probability Density Function parameterisation (?)
•   …
– For 100fb-1 (in 2011) and Background accuracy of 5% :
• S/B=30% for a 5s discovery (mHiggs=120 GeV)
• S/B=16% for a 3s evidence (mHiggs=120 GeV)
Extend to Shape Analysis
From the Statistic workshop at CERN:

Counting Experiment
Shape (PDF)

Saverio ?
Chris
First Steps in Roofit

• Roofit part of the package ROOT 5-02
• Compiled and installed on my Laptop
• Goal: use it and develop the framework for:
• Compute limits (Exotics)
• Research of bumps (Higgs)
•Very nice Topic for Glasgow/UK collaboration

```
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
 views: 12 posted: 2/26/2012 language: English pages: 10