cmsnotes/ZMet2011/introduction.tex


\section{Introduction}

In this note we describe a search for new physics in the 2011 
opposite sign isolated dilepton sample ($ee$, $e\mu$, and $\mu\mu$).  
The main sources of high \pt isolated dileptons at CMS are Drell Yan and \ttbar.
Here we concentrate on dileptons with invariant mass consistent
with $Z \to ee$ and $Z \to \mu\mu$.  A separate search for new physics in the non-\Z
sample is described in~\cite{ref:GenericOS}.

We search for new physics in the final state of \Z plus two or more jets plus missing 
transverse energy (MET). We reconstruct the \Z boson
in its decay to $e^+e^-$ or $\mu^+\mu^-$. Our search regions are defined as 
MET $\ge$ \signalmetl~GeV (loose signal region) and MET $\ge$ \signalmett~GeV 
(tight signal region), and two or more jets. We use data driven techniques to predict the
standard model background in these search regions. 
Contributions from Drell-Yan production combined with detector mis-measurements that 
produce fake MET are modeled via MET templates based on photon plus jets or QCD events. 
Top pair production backgrounds, as well as other backgrounds for which the lepton
flavors are uncorrelated such as WW and DY$\rightarrow\tau\tau$, are 
modeled via $e^\pm\mu^\mp$ subtraction.

As leptonically decaying \Z bosons are a signature that has very little background, 
they provide a clean final state in which to search for new physics. 
Because new physics is expected to be connected to the Standard Model Electroweak sector, 
it is likely that new particles will couple to W and Z bosons. 
For example, in mSUGRA, low $M_{1/2}$ can lead to a significant branching fraction 
for $\chi_2^0 \rightarrow Z \chi_1^0$. 
In addition, we are motivated by the existence of dark matter to search for new physics with MET.
Enhanced MET is a feature of many new physics scenarios, and R-parity conserving SUSY 
again provides a popular example. The main challenge of this search is therefore to 
understand the tail of the fake MET distribution in \Z plus jets events.

The basic idea of the MET template method~\cite{ref:templates1}\cite{ref:templates2} is 
to measure the MET distribution in data in a control sample which has no true MET 
and a similar topology to the signal events. 
%Start the qcd vs photon discussion
Templates can be derived from either a QCD sample (as was done in the original implementation) 
or a photon plus jets sample.
%In our case, we choose a photon sample with two or more jets as the control sample.
%Both the control sample and signal sample consist of a well measured object (either a 
%photon or a leptonically decaying $Z$), which recoils against a system of hadronic jets. 
In both cases, the instrumental MET is dominated by mismeasurements of the hadronic system,
and can be classified by the number of jets in the event and the scalar sum of their transverse
momenta.
The prediction is made such that the jet system in the control sample is similar to that of the
signal sample.
By using two independent control samples--QCD and photon plus jets--we are able to illustrate
the robustness of the MET templates method and to cross check the data driven background 
prediction.

This note is organized as follows. 
In sections \ref{sec:datasets} and \ref{sec:trigSel} we descibe
the datasets and triggers used, followed by the detailed object definitions (electrons, muons, photons,
jets, MET) and event selections in sections \ref{sec:evtsel} through \ref{sec:jetsel}.
We define a preselection and compare data vs. MC yields passing this preselection in 
Section~\ref{sec:yields}.
We then define the signal regions and show the number of observed events and MC expected 
yields in Section~\ref{sec:sigregion}.
Section~\ref{sec:templates} introduces the MET template method and discusses its derivation 
in some detail and is followed by a demonstration in Section~\ref{sec:mc} 
that the method works in Monte Carlo.
Section~\ref{sec:topbkg} introduces the top background estimate based on opposite flavor subtraction, 
and contributions from other backgrounds are discussed in Section~\ref{sec:othBG}.
Section~\ref{sec:results} shows the results for applying these methods in data.
We analyze the systematic uncertainties in the background predictions and in signal acceptance
in Section~\ref{sec:systematics}. 
We then proceed to calculate an upper limit on the non-SM contributions to our signal regions 
in Section~\ref{sec:upperlimit}. 
Efficiencies which can be used to test specific models of new physics are given 
in Section \ref{sec:outreach}.
Finally, in Section~\ref{sec:models} we calculate upper limits on the quantity \sta\
assuming efficiencies and uncertainties from sample benchmark SUSY processes. 

Revision:	1.6
Committed:	Wed Jul 13 12:28:35 2011 UTC (13 years, 10 months ago) by warren
Content type:	application/x-tex
Branch:	MAIN
CVS Tags:	HEAD
Changes since 1.5:	+4 -3 lines
Log Message:	Add model dependent limits
#	User	Rev	Content
1	benhoob	1.1
2			\section{Introduction}
3
4	warren	1.2	In this note we describe a search for new physics in the 2011
5	benhoob	1.1	opposite sign isolated dilepton sample ($ee$, $e\mu$, and $\mu\mu$).
6	warren	1.2	The main sources of high \pt isolated dileptons at CMS are Drell Yan and \ttbar.
7	benhoob	1.1	Here we concentrate on dileptons with invariant mass consistent
8			with $Z \to ee$ and $Z \to \mu\mu$. A separate search for new physics in the non-\Z
9			sample is described in~\cite{ref:GenericOS}.
10
11	warren	1.3	We search for new physics in the final state of \Z plus two or more jets plus missing
12			transverse energy (MET). We reconstruct the \Z boson
13			in its decay to $e^+e^-$ or $\mu^+\mu^-$. Our search regions are defined as
14			MET $\ge$ \signalmetl~GeV (loose signal region) and MET $\ge$ \signalmett~GeV
15			(tight signal region), and two or more jets. We use data driven techniques to predict the
16			standard model background in these search regions.
17			Contributions from Drell-Yan production combined with detector mis-measurements that
18	warren	1.4	produce fake MET are modeled via MET templates based on photon plus jets or QCD events.
19	benhoob	1.1	Top pair production backgrounds, as well as other backgrounds for which the lepton
20	warren	1.3	flavors are uncorrelated such as WW and DY$\rightarrow\tau\tau$, are
21			modeled via $e^\pm\mu^\mp$ subtraction.
22	benhoob	1.1
23	warren	1.4	As leptonically decaying \Z bosons are a signature that has very little background,
24	warren	1.3	they provide a clean final state in which to search for new physics.
25			Because new physics is expected to be connected to the Standard Model Electroweak sector,
26			it is likely that new particles will couple to W and Z bosons.
27			For example, in mSUGRA, low $M_{1/2}$ can lead to a significant branching fraction
28			for $\chi_2^0 \rightarrow Z \chi_1^0$.
29	benhoob	1.1	In addition, we are motivated by the existence of dark matter to search for new physics with MET.
30	warren	1.3	Enhanced MET is a feature of many new physics scenarios, and R-parity conserving SUSY
31			again provides a popular example. The main challenge of this search is therefore to
32	benhoob	1.1	understand the tail of the fake MET distribution in \Z plus jets events.
33
34	warren	1.3	The basic idea of the MET template method~\cite{ref:templates1}\cite{ref:templates2} is
35	warren	1.4	to measure the MET distribution in data in a control sample which has no true MET
36			and a similar topology to the signal events.
37			%Start the qcd vs photon discussion
38			Templates can be derived from either a QCD sample (as was done in the original implementation)
39			or a photon plus jets sample.
40			%In our case, we choose a photon sample with two or more jets as the control sample.
41			%Both the control sample and signal sample consist of a well measured object (either a
42			%photon or a leptonically decaying $Z$), which recoils against a system of hadronic jets.
43			In both cases, the instrumental MET is dominated by mismeasurements of the hadronic system,
44			and can be classified by the number of jets in the event and the scalar sum of their transverse
45			momenta.
46			The prediction is made such that the jet system in the control sample is similar to that of the
47			signal sample.
48			By using two independent control samples--QCD and photon plus jets--we are able to illustrate
49			the robustness of the MET templates method and to cross check the data driven background
50			prediction.
51	benhoob	1.1
52			This note is organized as follows.
53	warren	1.4	In sections \ref{sec:datasets} and \ref{sec:trigSel} we descibe
54			the datasets and triggers used, followed by the detailed object definitions (electrons, muons, photons,
55			jets, MET) and event selections in sections \ref{sec:evtsel} through \ref{sec:jetsel}.
56			We define a preselection and compare data vs. MC yields passing this preselection in
57			Section~\ref{sec:yields}.
58			We then define the signal regions and show the number of observed events and MC expected
59			yields in Section~\ref{sec:sigregion}.
60			Section~\ref{sec:templates} introduces the MET template method and discusses its derivation
61			in some detail and is followed by a demonstration in Section~\ref{sec:mc}
62			that the method works in Monte Carlo.
63			Section~\ref{sec:topbkg} introduces the top background estimate based on opposite flavor subtraction,
64			and contributions from other backgrounds are discussed in Section~\ref{sec:othBG}.
65	benhoob	1.1	Section~\ref{sec:results} shows the results for applying these methods in data.
66	warren	1.5	We analyze the systematic uncertainties in the background predictions and in signal acceptance
67			in Section~\ref{sec:systematics}.
68			We then proceed to calculate an upper limit on the non-SM contributions to our signal regions
69	warren	1.4	in Section~\ref{sec:upperlimit}.
70	warren	1.6	Efficiencies which can be used to test specific models of new physics are given
71			in Section \ref{sec:outreach}.
72			Finally, in Section~\ref{sec:models} we calculate upper limits on the quantity \sta\
73			assuming efficiencies and uncertainties from sample benchmark SUSY processes.
74	warren	1.4