A Search for the Standard Model Higgs Boson in the

All-Hadronic channel using a Matrix Element Method using 2.0 fb-1


Aart Heijboer (contact), Joseph Kroll
University of Pennsylvania

Daniel Whiteson
University of California, Irvine

 Rong-Shyang Lu, Ankush Mitra, Song-Ming Wang
 Academia Sinica


higgs group conveners: Ben Kilminster and Eric James


CDF public note (#9366)

abstract

This web page summarizes a  Matrix Element (ME) based search for the Standard Model (SM) Higgs Boson, where the Higgs is produced in association with a W or Z boson (denoted collectively as V, for 'Vector boson'), and where the H decays into a \bbar pair and the V decays hadronically. The data are in agreement with the background model and we set a limit on the VH production cross section as a function of MH.

Summary

We select events with at least four jets, of which two are identified (tagged) as b-jets. The hadronic decay of the vector boson has a large branching ratio (about 2/3),  which makes for a large signal, relative to other analyses where the Vector boson decays to leptons. However, there is a very large background from QCD multijet production with a $b \bar{b}$-pair. A main ingredient of the analysis was the development of a data-driven model for this background.
In order to optimally distinguish signal events from the background, a Matrix Element (ME) technique was used. Using parton-level differential cross-sections, we compute the probability of the observed events for both the signal and background hypotheses. The discriminating variable is than formed by the ratio of these two likelihoods.
A model for the QCD background was constructed from  z large sample of events with only a single b-tag. These events are weighted by the ratio of  double-tagged to single tagged events, which is measured separately on events known to contain no signal. In order to correctly model the shape of the discriminating variable,  the tag ratio was measured as a function of several  kinematic variables.


Results

We have observed no excess over the background model. We set limits on the production cross section as a multiple of the SM cross section.


Table with expected and observed values for the limit in terms of the SM cross-section.






Plots




Feynman diagram for the VH ->jjbb process
Table showing event yields for the WH and ZH signal  

Cross-checks of the background modeling method performed in ttbar mc 1/3

Cross-checks of the background modeling method performed in ttbar mc 2/3

Cross-checks of the background modeling method performed in ttbar mc 3/3
showing the modeling of the ME discriminant works well in ttbar MC; residuals
are used to derived a 'mismodeling systematic'.

Explanation of the three systematic uncertainties considered in the background model.

The three systematic uncertainties (in terms of #events) applied to the background shape.
The Jet energy transfer functions that are used in the Matrix Element calculations. (derived from MC)

 The systematic uncertainties on the signal acceptance and the signal shape
(todo: add legend).
cross-check: modeling the data in the 'ctrl' region (events with M(untagged jets) uncompatible
with W or Z).
Checking several kinematic variables in data 1/3
Checking several kinematic variables in data 2/3
Checking several kinematic variables in data 3/3
Cross check in events with more than exactly 4 jets ( a sub-sample of all data, depleted in signal).
Data! and background model. For a test-mass of Mh=100 GeV
Data! and background model. For a test-mass of Mh=110 GeV
Data! and background model. For a test-mass of Mh=120 GeV
Data! and background model. For a test-mass of Mh=130 GeV
Data! and background model. For a test-mass of Mh=140 GeV
A nice version of the data and background model for Mh=120 GeV
Another version of the data and background model for Mh=120 GeV.. similar plot in eps .. similar plot in gif
Same as above but with sig. MC superimposed and not stacked... similar plot in eps .. similar plot in gif
Expected and observed limit, as a multiple of the SM cross-section [eps ] [png ] [gif ]
Expected and observed limit, as a multiple of the SM cross-section - log. y axis [eps ] [png ] [gif ]
Expected and observed limit, as absolute cross-section [eps ] [png ] [gif ]
Expected limit as a multiple of the SM cross-section; without any systematic uncertainties. [eps ] [png ] [gif ]