Presentation on the inclusive analysis

Adding the Single Top to the simultaneous cross-section

measurements using the AIDA technique at 8 TeV

- Ruturaj Apte

Aim

● Adding the Single Top channel (Wt) to the analysis of the other 3 channels namely the ttbar, z/γ -> ττ and the WW channels using the AIDA technique.

● The aim is to measure the cross-section of the Wt channel in an inclusive manner so as to reduce the dependencies on models and to search for any new physics.

● The plots shown on the subsequent slides have the requirement of exactly one electron and one muon with opposite signs, with no particular requirements on the number of b tagged jets.

● The plots only include MC data.

Motivation for measuring single top (Wt)

* CMS σ = 23.4 ± 5.4 pb June 2014 – first 5σ measurement (http://arxiv.org/abs/1401.2942) * ATLAS σ = 27.2 ± 5.8 pb Sep 2013

The top mass is very close to the energy of the electroweak symmetry breaking scale. It has been speculated that the single top plays an important role in the symmetry breaking. Any deviation from the predicted rate will be a signal of new physics.

CMS 8 TeV:

ATLAS 8 TeV:

σ

Missing Energy

Invariant Mass

Normal Jet binning- 2 bins

Normal Jet Binning- 6 bins

Jet Class

● Jet classes introduced to better resolve the Wt events from the t-tbar events

Class 0 1 2 3 4

bjets 0 0 1 ≥1 others

Total Jets 0 1 1 2

Jet Class Binning

Fitting to Pseudodata with the Jetclass binning – fitted yields for

single top channel

Fitting to Pseudodata with the Jetclass binning – Pull distributions

for st channel

Test Statistic

● The test statistic used to test for the significance of the signal is the

(-2) * ln( ratio of the likelihood functions)

In the numerator likelihood is maximised by fixing μ = 0. Wheras in the denominator μ is fitted to the value that gives the maximum likelihood.

● We checked the validity of the asymtotic limit formulae by comparing them to the test statistic distributions generated from pseudodata.

● They turned out to be in good agreement and hence we used the asymtotic limits for our significance studies.

Significance Study Results

● We did significance studies for the 2bin, 6bin and jet class binning using only statistical uncertainties.

● We found that the 2bin gives a much lower significance ~ 3sigma than the other two types.

● Both jet class and 6 jet binnning give a significance > 5 sigma.

● The significance of jet class binning was slightly greater than that of the 6 jet bin case.

Including Systematics

● Next thing was to include 80 different sources of systematic uncertainties related to electron, muon and jet variables in the analysis.

● We first looked only at the normalization uncertainties due to the systematics.

Normalization systematics

● These are the % differences in the event acceptance from the nominal distrubution of the four main rate systematics.

Jet Energy Resolution : 0.60 %

Electron Id SF : 2.35 %

Muon Id SF : 0.51 %

Jet Vertex Fraction : 1.18 %● In the next slides we shall look at what these

four variables mean.

What they mean

● Jet Energy Resolution : It is measured using two different techniques.

● The differences between the results of these two methods is a source of a systematic uncertainty.

● To evaluate the impact, the jet energies in MC simulations are randomly smeared by adding/subtracting a range of one standard deviation on jet energy resolution.

● This propagates to an acceptance change.

What they mean

● Electron and Muon Id SF : Various cuts are applied on the tracks and energy depositions in the detector before it is identified as an electron or a muon.

● There is a scaling factor associated with these cuts for the MC to match data.

● This scaling factor has an associated systematic uncertainty.

What they mean

● Jet Vertex Fraction : In a given jet, JVF is the fraction of tracks that come from the primary vertex. We have a cut on 0.5 for this.

● There is an uncertainty associated with the cuts (+- 0.1) that we apply to the Jet Vertex Fraction.

● This is another source of a systematic uncertainty.

Jet Energy Resolutionred -> nominal blue-> shifted template ratio-> nominal / shifted

Electron Id scaling factorred -> nominal blue-> shifted template ratio-> nominal / shifted

Muon Id Scaling Factorred -> nominal blue-> shifted template ratio-> nominal / shifted

Jet Vertex Fractionred -> nominal blue-> shifted template ratio-> nominal / shifted

Shape Systematics

● To determine the effect of a shape systematic we first generated pseudodata from the particular systematic.

● Then to this pseudodata we fitted the nominal histograms.

● We then calculated the percentage difference between the mean fit value of the Wt and the Wt yield in the nominal case.

● Repeated the procedure for all the 80 systematics

Important shape syst.

● Following percentages are calculated as % difference between the mean fit value to the fits done to pseudodata generated from systematics by the nominal channel and the nominal Wt yield.

Jet Energy Resolution : 2.75 %

Electron ID Scaling Factor : 2.82 %

B Tag Scaling Factor : 6.20 %

Jet Energy Scale : 5.94 %

Conclusions

● We have completed studying the systematic uncertainties for the signal Wt channel.

● Without systematics, we found the significance of jet class binning slightly better than 6 jet. But it can be seen that there is a large systematic uncertainty in the B Tag Scaling Factor.

● We need to extend the analysis of systematics for all the rest of the signal and background channels.

● Finalize the binning based on significance studies including all the systematics.

Backup Slides

Significance Testing using Asymtotic Limits – test statistic obtained by

fitting to the background only

Test Statistic

● Next we calculate it for a pseudodata that has been generated by adding the signal to the background templates.

● The median of this test statistic needs to be well resolved from the previous distribution for us to obtain a good significance.

● Here we have checked the validity of the asymptotic formulae to approximate the behaviour of these two distributions.

● In the future we will only be using the asymptotic formulae rather than generating a lot of pseudodata, since its validity has been tested.

Significance Testing using Asymtotic Limits - test statistic obtained by

fitting to the sig + backg

List of shape syst.

List of shape syst.

List of shape syst.

List of shape syst.