| 1 |
|
\section{Background Estimates from Data} |
| 2 |
|
\label{sec:datadriven} |
| 3 |
+ |
|
| 4 |
+ |
To look for possible BSM contributions, we define 2 signal regions that preserve about |
| 5 |
+ |
0.1\% of the dilepton $t\bar{t}$ events, by adding requirements of large \MET\ and \Ht: |
| 6 |
+ |
|
| 7 |
+ |
\begin{itemize} |
| 8 |
+ |
\item high \MET\ signal region: \MET $>$ 275~GeV, \Ht $>$ 300~GeV, |
| 9 |
+ |
\item high \Ht\ signal region: \MET $>$ 200~GeV, \Ht $>$ 600~GeV. |
| 10 |
+ |
\end{itemize} |
| 11 |
+ |
|
| 12 |
+ |
For the high \MET\ (high \Ht) signal region, the MC predicts 2.6 (2.5) SM events, |
| 13 |
+ |
dominated by dilepton $t\bar{t}$; the expected LM1 yield is 17 (14) and the |
| 14 |
+ |
expected LM3 yield is 4.3 (4.3). The signal regions are indicated in Fig.~\ref{fig:met_ht}. |
| 15 |
+ |
|
| 16 |
|
We use three independent methods to estimate from data the background in the signal region. |
| 17 |
< |
The first method is a novel technique based on the ABCD method used in our 2010 analysis~\cite{ref:ospaper}, |
| 17 |
> |
The first method is a novel technique based on the ABCD method, which we used in our 2010 analysis~\cite{ref:ospaper}, |
| 18 |
|
and exploits the fact that \HT\ and $y$ are nearly uncorrelated for the $t\bar{t}$ background; |
| 19 |
|
this method is referred to as the ABCD' technique. First, we extract the $y$ and \Ht\ distributions |
| 20 |
|
$f(y)$ and $g(H_T)$ from data, using events from control regions which are dominated by background. |
| 21 |
< |
Because $y$ and \Ht\ are weakly-correlated, we can predict the distribution of events in the $y$ vs. \Ht\ plane as: |
| 21 |
> |
Because $y$ and \Ht\ are weakly-correlated, the distribution of events in the $y$ vs. \Ht\ plane is described by: |
| 22 |
|
|
| 23 |
|
\begin{equation} |
| 24 |
|
\frac{\partial^2 N}{\partial y \partial H_T} = f(y)g(H_T), |
| 27 |
|
allowing us to deduce the number of events falling in any region of this plane. In particular, |
| 28 |
|
we can deduce the number of events falling in our signal regions defined by requirements on \MET\ and \Ht. |
| 29 |
|
|
| 30 |
< |
We measure the $f(y)$ and $g(H_T)$ distributions using events in the regions indicated in Fig.~\ref{fig:abcdprime} |
| 30 |
> |
We measure the $f(y)$ and $g(H_T)$ distributions using events in the regions indicated in Fig.~\ref{fig:abcdprimedata} |
| 31 |
|
Next, we randomly sample values of $y$ and \Ht\ from these distributions; each pair of $y$ and \Ht\ values is a pseudo-event. |
| 32 |
|
We generate a large ensemble of pseudo-events, and find the ratio $R_{S/C}$, the ratio of the |
| 33 |
|
number of pseudo-events falling in the signal region to the number of pseudo-events |
| 34 |
|
falling in a control region defined by the same requirements used to select events |
| 35 |
|
to measure $f(y)$ and $g(H_T)$. We then |
| 36 |
< |
multiply this ratio by the number of \ttbar\ MC events which fall in the control region |
| 36 |
> |
multiply this ratio by the number events which fall in the control region in data |
| 37 |
|
to get the predicted yield, ie. $N_{pred} = R_{S/C} \times N({\rm control})$. |
| 38 |
|
To estimate the statistical uncertainty in the predicted background, we smear the bin contents |
| 39 |
|
of $f(y)$ and $g(H_T)$ according to their uncertainties. We repeat the prediction 20 times |
| 40 |
|
with these smeared distributions, and take the RMS of the deviation from the nominal prediction |
| 41 |
|
as the statistical uncertainty. We have studied this technique using toy MC studies based on |
| 42 |
< |
similar event samples of similar size to the expected yield in data for 1 fb$^{-1}$. |
| 42 |
> |
event samples of similar size to the expected yield in data for 1 fb$^{-1}$. |
| 43 |
|
Based on these studies we correct the predicted backgrounds yields by factors of 1.2 $\pm$ 0.5 |
| 44 |
|
(1.0 $\pm$ 0.5) for the high \MET\ (high \Ht) signal region. |
| 45 |
|
|
| 54 |
|
reliably accounted for. We then use the observed |
| 55 |
|
$\pt(\ell\ell)$ distribution to model the $\pt(\nu\nu)$ distribution, |
| 56 |
|
which is identified with \MET. Thus, we use the number of observed |
| 57 |
< |
events with $\HT > 300\GeV$ and $\pt(\ell\ell) > 275\GeV^{1/2}$ |
| 57 |
> |
events with $\HT > 300\GeV$ and $\pt(\ell\ell) > 275\GeV$ |
| 58 |
|
($\HT > 600\GeV$ and $\pt(\ell\ell) > 200\GeV^{1/2}$ ) |
| 59 |
|
to predict the number of background events with |
| 60 |
< |
$\HT > 300\GeV$ and $\MET = > 275\GeV^{1/2}$ ($\HT > 600\GeV$ and $\MET = > 200\GeV^{1/2}$). |
| 60 |
> |
$\HT > 300\GeV$ and $\MET > 275\GeV$ ($\HT > 600\GeV$ and $\MET > 200\GeV$). |
| 61 |
|
In practice, two corrections must be applied to this prediction, as described below. |
| 62 |
|
|
| 63 |
|
% |
| 68 |
|
rescale the prediction by a factor equal to the inverse of the |
| 69 |
|
fraction of events passing the preselection which also satisfy the |
| 70 |
|
requirement $\pt(\ell\ell) > 50\GeVc$. |
| 71 |
< |
For the \Ht $>$ 300 GeV requirement corresponding to the high \MET\ signal region, |
| 71 |
> |
For the \Ht\ $>$ 300 GeV requirement corresponding to the high \MET\ signal region, |
| 72 |
|
we determine this correction from data and find $K_{50}=1.5 \pm 0.3$. |
| 73 |
< |
For the \Ht $>$ 600 GeV requirement corresponding to the high \Ht\ signal region, |
| 73 |
> |
For the \Ht\ $>$ 600 GeV requirement corresponding to the high \Ht\ signal region, |
| 74 |
|
we do not have enough events in data to determine this correction with statistical |
| 75 |
< |
precisions, so we instead extract it from MC and find $K_{50}=1.3 \pm 0.2$. |
| 75 |
> |
precision, so we instead extract it from MC and find $K_{50}=1.3 \pm 0.2$. |
| 76 |
|
The second correction ($K_C$) is associated with the known polarization of the $W$, which |
| 77 |
|
introduces a difference between the $\pt(\ell\ell)$ and $\pt(\nu\nu)$ |
| 78 |
|
distributions. The correction $K_C$ also takes into account detector effects such as the hadronic energy |
| 82 |
|
in the extraction of $K_C$ and the 5\% uncertainty in the hadronic energy scale~\cite{ref:jes}. |
| 83 |
|
|
| 84 |
|
Our third background estimation method is based on the fact that many models of new physics |
| 85 |
< |
produce an excess of SF with respect to OF lepton pairs. In SUSY, such an excess may produced |
| 85 |
> |
produce an excess of SF with respect to OF lepton pairs. In SUSY, such an excess may be produced |
| 86 |
|
in the decay $\chi_2^0 \to \chi_1^0 \ell^+\ell^-$ or in the decay of $Z$ bosons produced in |
| 87 |
|
the cascade decays of heavy, colored objects. In contrast, for the \ttbar\ background the |
| 88 |
|
rates of SF and OF lepton pairs are the same, as is also the case for other SM backgrounds |
| 94 |
|
\Delta = R_{\mu e}N(ee) + \frac{1}{R_{\mu e}}N(\mu\mu) - N(e\mu), |
| 95 |
|
\end{equation} |
| 96 |
|
|
| 97 |
< |
where $R_{\mu e} = 1.13 \pm 0.05$ is the ratio of muon to electron selection efficiencies. |
| 98 |
< |
This quantity is evaluated by taking the square root of the ratio of the number of observed |
| 99 |
< |
$Z \to \mu^+\mu^-$ to $Z \to e^+e^-$ events, in the mass range 76-106 GeV with no jets or |
| 97 |
> |
where $R_{\mu e} = 1.13 \pm 0.05$ is the ratio of muon to electron selection efficiencies, |
| 98 |
> |
evaluated by taking the square root of the ratio of the number of |
| 99 |
> |
$Z \to \mu^+\mu^-$ to $Z \to e^+e^-$ events in data, in the mass range 76-106 GeV with no jets or |
| 100 |
|
\met\ requirements. The quantity $\Delta$ is predicted to be 0 for processes with |
| 101 |
|
uncorrelated lepton flavors. In order for this technique to work, the kinematic selection |
| 102 |
|
applied to events in all dilepton flavor channels must be the same, which is not the case |
| 103 |
< |
for our default selection because the $Z$ mass veto is applied only to same-flavor channels.Therefore when applying the OF subtraction technique we also apply the $Z$ mass veto also |
| 103 |
> |
for our default selection because the $Z$ mass veto is applied only to same-flavor channels. |
| 104 |
> |
Therefore when applying the OF subtraction technique we also apply the $Z$ mass veto also |
| 105 |
|
to the $e\mu$ channel. |
| 106 |
|
|
| 107 |
|
All background estimation methods based on data are in principle subject to signal contamination |
| 114 |
|
in the different control regions for the two methods. |
| 115 |
|
For example, in the extreme case of a |
| 116 |
|
BSM signal with identical distributions of $\pt(\ell \ell)$ and \MET, an excess of events might be seen |
| 117 |
< |
in the ABCD method but not in the $\pt(\ell \ell)$ method. |
| 117 |
> |
in the ABCD' method but not in the $\pt(\ell \ell)$ method. |
| 118 |
|
|
