| 1 |
claudioc |
1.8 |
\section{Acceptance and efficiency systematics}
|
| 2 |
claudioc |
1.1 |
\label{sec:systematics}
|
| 3 |
|
|
|
| 4 |
|
|
This is a search for new physics contributions to
|
| 5 |
|
|
events with high \met and lots of jet activity.
|
| 6 |
|
|
As seen in Section~\ref{sec:results}, there is no
|
| 7 |
|
|
evidence for a contribution beyond SM expectations.
|
| 8 |
|
|
|
| 9 |
|
|
Strictly speaking it is impossible to talk about
|
| 10 |
claudioc |
1.8 |
``acceptance and efficiency systematics'' because these kinds of
|
| 11 |
claudioc |
1.1 |
systematics only apply to a well defined final state.
|
| 12 |
claudioc |
1.8 |
Nevertheless, we can make general statements about the
|
| 13 |
|
|
systematic uncertainties, including quantitative
|
| 14 |
benhoob |
1.17 |
estimates of the systematic uncertainties associated with
|
| 15 |
benhoob |
1.18 |
a few specific processes. Note that we have used Spring10
|
| 16 |
|
|
MC for the studies of systematic uncertainties described in this section,
|
| 17 |
|
|
and we are currently checking if any of the reported values
|
| 18 |
|
|
change after switching to Fall10 MC.
|
| 19 |
claudioc |
1.1 |
|
| 20 |
benhoob |
1.4 |
The systematic uncertainty on the lepton acceptance consists
|
| 21 |
claudioc |
1.1 |
of two parts: the trigger efficiency uncertainty and the
|
| 22 |
claudioc |
1.12 |
ID and isolation uncertainty. We discuss these in turn.
|
| 23 |
claudioc |
1.1 |
|
| 24 |
|
|
The trigger efficiency
|
| 25 |
|
|
for two leptons of $P_T>10$ GeV, with one lepton of
|
| 26 |
|
|
$P_T>20$ GeV is very high, except in some corners
|
| 27 |
claudioc |
1.13 |
of phase space, see Section~\ref{sec:trgeffsum}.
|
| 28 |
claudioc |
1.1 |
We estimate the efficiency uncertainty to be a few percent,
|
| 29 |
benhoob |
1.21 |
mostly in the low $P_T$ region. For $t\bar{t}$, LM0 and LM1
|
| 30 |
|
|
we find trigger efficiency uncertainties of less than 1\%, evaluated
|
| 31 |
|
|
by taking the difference in yields in the signal region between
|
| 32 |
|
|
assuming 100\% trigger efficiency and using the trigger efficiency model.
|
| 33 |
|
|
% trigger efficiency uncertainties: ttbar 0.3%, LM0 0.6%, LM1 0.6%
|
| 34 |
claudioc |
1.1 |
|
| 35 |
claudioc |
1.3 |
\begin{figure}[tbh]
|
| 36 |
|
|
\begin{center}
|
| 37 |
benhoob |
1.19 |
\includegraphics[width=1.0\linewidth]{ttdilD6T_eff_Dec02_38X.png}
|
| 38 |
|
|
\includegraphics[width=1.0\linewidth]{lm_eff_Dec02_38X.png}
|
| 39 |
claudioc |
1.3 |
\caption{\label{fig:effttbar}\protect
|
| 40 |
benhoob |
1.24 |
Identification and isolation efficiencies for leptons from $t \to W \to \ell$ and
|
| 41 |
|
|
$t \to W \to \tau \to \ell$ in $t\bar{t}$ events (top). Isolation efficiency
|
| 42 |
|
|
for $t\bar{t}$, LM0 and LM1 (bottom).}
|
| 43 |
claudioc |
1.3 |
\end{center}
|
| 44 |
|
|
\end{figure}
|
| 45 |
|
|
|
| 46 |
|
|
|
| 47 |
claudioc |
1.7 |
\begin{table}[hbt]
|
| 48 |
|
|
\begin{center}
|
| 49 |
|
|
\caption{\label{tab:tagandprobe} Tag and probe results on $Z \to \ell \ell$
|
| 50 |
|
|
on data and MC. We quote ID efficiency given isolation and
|
| 51 |
benhoob |
1.22 |
the isolation efficiency given ID. }
|
| 52 |
claudioc |
1.7 |
\begin{tabular}{|l||c|c|}
|
| 53 |
|
|
\hline
|
| 54 |
benhoob |
1.22 |
& Data T\&P & MC T\&P \\
|
| 55 |
|
|
\hline
|
| 56 |
|
|
$\epsilon(id|iso)$ electrons & $0.925 \pm 0.007$ & $0.934 \pm 0.004$ \\
|
| 57 |
|
|
$\epsilon(iso|id)$ electrons & $0.991 \pm 0.002$ & $0.987 \pm 0.002$ \\
|
| 58 |
|
|
$\epsilon(id|iso)$ muons & $0.962 \pm 0.005$ & $0.984 \pm 0.002$ \\
|
| 59 |
|
|
$\epsilon(iso|id)$ muons & $0.987 \pm 0.003$ & $0.982 \pm 0.002$ \\
|
| 60 |
claudioc |
1.7 |
\hline
|
| 61 |
|
|
\end{tabular}
|
| 62 |
|
|
\end{center}
|
| 63 |
|
|
\end{table}
|
| 64 |
|
|
|
| 65 |
|
|
|
| 66 |
claudioc |
1.3 |
The ID efficiencies in MC are shown in
|
| 67 |
|
|
Figures~\ref{fig:effttbar}
|
| 68 |
|
|
for the leptons from $t \to W \to \ell$ and $t \to W \to \tau \to \ell$.
|
| 69 |
claudioc |
1.7 |
Tag and probe studies show that these are correct to about 2\%,
|
| 70 |
|
|
see Table~\ref{tab:tagandprobe}.
|
| 71 |
|
|
Note that the isolation efficiency depends on the jet activity in
|
| 72 |
claudioc |
1.1 |
the final state. For example, in MC we find that the
|
| 73 |
|
|
lepton isolation efficiency differs by $\approx 4\%$
|
| 74 |
|
|
{\bf per lepton} between $Z$ events and $t\bar{t}$ events\cite{ref:top}.
|
| 75 |
benhoob |
1.23 |
{\bf \color{red} This difference has been evaluated with Spring10 MC, currently checking with Fall10 MC. }
|
| 76 |
claudioc |
1.11 |
%\noindent {\bf This figure should be cut off at 100 GeV, and
|
| 77 |
|
|
%the y-axis should be zero-suppressed}
|
| 78 |
claudioc |
1.1 |
|
| 79 |
|
|
Another significant source of systematic uncertainty is
|
| 80 |
|
|
associated with the jet and $\met$ energy scale. The impact
|
| 81 |
claudioc |
1.8 |
of this uncertainty is final-state dependent. Final
|
| 82 |
|
|
states characterized by lots of hadronic activity and \met are
|
| 83 |
claudioc |
1.1 |
less sensitive than final states where the \met and SumJetPt
|
| 84 |
|
|
are typically close to the requirement. To be more quantitative,
|
| 85 |
|
|
we have used the method of Reference~\cite{ref:top} to evaluate
|
| 86 |
|
|
the systematic uncertainties on the acceptance for $t\bar{t}$
|
| 87 |
|
|
and two benchmark SUSY points. The uncertainties are calculated
|
| 88 |
|
|
assuming a 5\% uncertainty to the hadronic energy scale in CMS.
|
| 89 |
|
|
|
| 90 |
benhoob |
1.23 |
For $t\bar{t}$ we find uncertainties of 8\% (baseline
|
| 91 |
|
|
selection) and 27\% (signal region D) ({\bf \color{red} This number
|
| 92 |
|
|
needs to be double-checked, Derek finds 33\%.}); for LM0 and LM1 we find
|
| 93 |
|
|
14\% and 6\% respectively for signal region D.
|
