ViewVC Help
View File | Revision Log | Show Annotations | Root Listing
root/cvsroot/UserCode/Vuko/Notes/WZCSA07/AppendixFitTest.tex
Revision: 1.9
Committed: Tue Jul 29 02:38:23 2008 UTC (16 years, 9 months ago) by ymaravin
Content type: application/x-tex
Branch: MAIN
Changes since 1.8: +143 -198 lines
Log Message:
Modifications from YM mostly on new stuff to answer ARC questions.

File Contents

# User Rev Content
1 ymaravin 1.9 \subsection{Further cross-checks}
2     The test described in the previous Section illustrate the robustness of the
3     matrix method to estimate the misidentification background correctly for varied
4     jet flavor composition in the $\Z+jet$ sample.
5    
6     In the following we further scrutinize the details of the background estimation.
7     We provide detailed information on the background extraction using the
8     default selection criteria, that without the requirement on the \W candidate
9     transverse mass, and finally using default selection criteria without subtracting
10     backgrounds that have no genuine \Z bosons.
11    
12     \subsubsection{Details on extracting background with matrix method}
13    
14     In Fig.~\ref{fig:AllFits} we provide dilepton mass fit information for each channel
15     for loose and tight \W lepton requirements using the full statistics of the CSA07 samples.
16     The fit is performed using an addition of a convolution of a Gaussian and Breit-Wigner function
17     and a line in order to fit the background. It has to be noticed that due to a lack
18     of statistics in ``Chowder soup'' sample, all bins with 0 events from the sample
19     have been modified in order to avoid to have a null uncertainty. The
20     corresponding uncertainty in the bin with no events correspond to the weight
21     of each process in the ``Chowder soup''. One can see that the uncertainties are
22     large, and the fit is not really constrained.
23 beaucero 1.1
24     \begin{figure}[hbt]
25     \begin{center}
26     \scalebox{0.3}{\includegraphics{figs/Fit3eLoose.eps}\includegraphics{figs/Fit3eTight.eps}}\\
27     \scalebox{0.3}{\includegraphics{figs/Fit2e1muLoose.eps}\includegraphics{figs/Fit2e1muTight.eps}}\\
28     \scalebox{0.3}{\includegraphics{figs/Fit2mu1eLoose.eps}\includegraphics{figs/Fit2mu1eTight.eps}}\\
29     \scalebox{0.3}{\includegraphics{figs/Fit3muLoose.eps}\includegraphics{figs/Fit3muTight.eps}}\\
30 ymaravin 1.9 \caption{Invariant mass of \Z boson candidate for $3e$, $2e1\mu$, $2\mu1e$, and $3\mu$
31     signatures (from top to bottom) for the lepton passing loose (left) and tight (right) identification
32     criteria.}
33 beaucero 1.1 \label{fig:AllFits}
34     \end{center}
35     \end{figure}
36    
37 ymaravin 1.9 The linear fit takes into account not only the background with non-genuine
38     \Z boson but it also accounts for some part of the \Z+jets and
39     $Zb\bar{b}$ background as the $\gamma^*$ processes populate the sidebands as well.
40     However, the results are still consistent within errors, as it can be seen by comparison
41     of the last two columns in Table~\ref{tab:CompFit}.
42 beaucero 1.1 \begin{table}[h]
43     \begin{center}
44     \begin{tabular}{|l|c|c|c|c|c|c|c|} \hline
45     & \multicolumn{2}{c|}{Background with genuine \Z} & \multicolumn{4}{c|}{Background without
46     genuine \Z boson} \\
47 ymaravin 1.9 Channel & $\Z+jets$ & $\Z b\bar{b}$ & $t\bar{t}$ & $\W+jets$ & Combined & Fit result \\ \hline
48 beaucero 1.6 $3e$ Loose &7.1 & 2.9 & 1.1 & 0.4 & 1.5 & 1.5$ \pm $3.0 \\\hline
49     $3e$ Tight &2.0 & 1.2 & 0.6 & 0.4 & 1.0 & 1.1$ \pm $2.8 \\\hline
50 ymaravin 1.9 $2e1\mu$ Loose &4.0 & 4.7 & 6.2 & 0.0 & 6.2 & 6.0$ \pm $4.1 \\\hline
51     $2e1\mu$ Tight &0.0 & 0.1 & 0.7 & 0.0 & 0.7 & 1.0$ \pm $2.8 \\\hline
52     $2\mu 1e$ Loose &10.1 & 2.9 & 0.8 & 0.0 & 0.8 & 1.6$ \pm $3.1 \\\hline
53     $2\mu 1e$ Tight &1.8 & 1.3 & 0.6 & 0.0 & 0.6 & 1.0$ \pm $2.7 \\\hline
54     $3\mu$ Loose &4.5 & 4.2 & 5.9 & 0.0 & 5.9 & 3.1$ \pm $3.5 \\\hline
55     $3\mu$ Tight &0.1 & 0.3 & 0.3 & 0.0 & 0.3 & 0.5$ \pm $2.5 \\\hline
56 beaucero 1.1 \end{tabular}
57     \end{center}
58 ymaravin 1.9 \caption{Comparison between Monte Carlo truth information and the results of the fit for the background without genuine \Z boson. Number of events are obtained in the invariant mass range between 81 and 101 GeV. The ``Loose'' and ``Tight'' selection criteria applied on the \W lepton candidate.}
59 beaucero 1.1 \label{tab:CompFit}
60     \end{table}
61    
62 ymaravin 1.9 The comparison between estimated background and the MC truth information is provided
63     in Table~\ref{tab:FinalXCLoose} for ``Loose'' and \ref{tab:FinalXC} for ``Tight'' lepton candidates.
64     Within uncertainties the results agree with each other for every signature and every category of
65     \W lepton identification. The agreement between predicted and MC truth background
66     as function of the dilepton mass is given in Figs.~\ref{fig:FinalMatrix3e}-\ref{fig:FinalMatrix3mu}
67     for all four signal categories respectively.
68 beaucero 1.1
69     \begin{table}[h]
70     \begin{center}
71     \begin{tabular}{lcccc} \hline \hline
72     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
73 ymaravin 1.9 $N$ - \ZZ -\Z$\gamma$ &19.9$\pm$1.0&23.6$\pm$0.0&23.4$\pm$1.0&25.5$\pm$0.0\\ \hline
74     $N^{non genuine \Z}$ (Fit)&1.5$\pm$3.0&6.0$\pm$4.1&1.6$\pm$3.1&3.1$\pm$3.5\\ \hline
75     $N^{genuine \Z}$ (matrix method)&10.1 $\pm$6.8&15.8 $\pm$8.4&14.5 $\pm$6.8&15.8 $\pm$5.7\\ \hline
76     $N^{\WZ}$ & 8.8$\pm$7.5&7.8$\pm$11.1&7.9$\pm$7.6&9.8 $\pm$7.4\\ \hline
77     \WZ from MC &8.1&9.0& 9.2 &11.3\\
78 beaucero 1.1 \hline
79     \end{tabular}
80     \caption{Expected number of selected events for an integrated luminosity of 300
81 ymaravin 1.9 pb$^{-1}$ for the signal and estimated background with 81 GeV $< M_Z < $ 101 GeV
82     with the full selection criteria applied but the requirement on the \W lepton which is
83     required to pass only ``Loose'' criteria.}
84     \label{tab:FinalXCLoose}
85 beaucero 1.1 \end{center}
86     \end{table}
87    
88 beaucero 1.5 \begin{table}[h]
89     \begin{center}
90     \begin{tabular}{lcccc} \hline \hline
91     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
92 ymaravin 1.9 $N$ - ZZ -Zgamma &12.4$\pm$1.0 &8.7$\pm$0.1 &13.1$\pm$0.9&10.6$\pm$0.0\\ \hline
93     $N^{non genuine Z}$ (Fit)&1.1$\pm$2.8&1.0$\pm$2.8&1.0$\pm$2.7&0.5$\pm$2.5\\ \hline
94     $N^{genuine Z}$ (matrix method)&3.2 $\pm$1.8&0.9 $\pm$1.1&4.6 $\pm$2.1&0.9 $\pm$1.1\\ \hline
95     $N^{WZ}$ & 8.2$\pm$3.5&7.7 $\pm$3.2 &7.6$\pm$3.5&9.6 $\pm$2.8\\ \hline
96     \WZ from MC &7.9&8.1& 9.0 &10.1\\ \hline
97 beaucero 1.5 \end{tabular}
98 ymaravin 1.9 \caption{Expected number of selected events for an integrated luminosity of 300
99     pb$^{-1}$ for the signal and estimated background with 81 GeV $< M_Z < $ 101 GeV for
100     the full selection criteria applied.}
101     \label{tab:FinalXC}
102 beaucero 1.5 \end{center}
103     \end{table}
104    
105 beaucero 1.4 \begin{figure}[hbt]
106     \begin{center}
107 ymaravin 1.9 \scalebox{0.62}{\includegraphics{figs/MatrixMethod3eLooseTightZmassMWtCut.eps}}
108     \caption{Comparison between background predicted with matrix method and MC truth information for the
109     $3e$ channel as a function of the \Z boson candidate invariant mass for loose (a, c) and tight (b, d) requirements
110     on the \W lepton for background (a, b) and signal (c, d).}
111 beaucero 1.4 \label{fig:FinalMatrix3e}
112     \end{center}
113     \end{figure}
114    
115     \begin{figure}[hbt]
116     \begin{center}
117 ymaravin 1.9 \scalebox{0.62}{\includegraphics{figs/MatrixMethod2e1muLooseTightZmassMWtCut.eps}}
118     \caption{
119     Comparison between background predicted with matrix method and MC truth information for the
120     $2e1\mu$ channel as a function of the \Z boson candidate invariant mass for loose (a, c) and tight (b, d) requirements
121     on the \W lepton for background (a, b) and signal (c, d).}
122     \label{fig:FinalMatrix2e1mu}
123 beaucero 1.4 \end{center}
124     \end{figure}
125    
126     \begin{figure}[hbt]
127     \begin{center}
128 ymaravin 1.9 \scalebox{0.62}{\includegraphics{figs/MatrixMethod2mu1eLooseTightZmassMWtCut.eps}}
129     \caption{Comparison between background predicted with matrix method and MC truth information for the
130     $2\mu 1e$ channel as a function of the \Z boson candidate invariant mass for loose (a, c) and tight (b, d) requirements
131     on the \W lepton for background (a, b) and signal (c, d).}
132 beaucero 1.4 \label{fig:FinalMatrix2mu1e}
133     \end{center}
134     \end{figure}
135 beaucero 1.2
136 beaucero 1.4 \begin{figure}[hbt]
137     \begin{center}
138 ymaravin 1.9 \scalebox{0.62}{\includegraphics{figs/MatrixMethod3muLooseTightZmassMWtCut.eps}}
139     \caption{Comparison between background predicted with matrix method and MC truth information for the
140     $3\mu$ channel as a function of the \Z boson candidate invariant mass for loose (a, c) and tight (b, d) requirements
141     on the \W lepton for background (a, b) and signal (c, d).}
142 beaucero 1.4 \label{fig:FinalMatrix3mu}
143     \end{center}
144     \end{figure}
145 beaucero 1.2
146    
147 ymaravin 1.9 \clearpage
148     \subsubsection{Background estimation without the \W boson transverse mass requirement}
149 beaucero 1.2
150 ymaravin 1.9 We also estimate background to the \WZ signal for the selection criteria without
151     the requirement on the transverse \W boson candidate mass. The results of
152     extraction of the background sources without real \Z boson are given in Table~\ref{tab:FitNoMWt}.
153 beaucero 1.2 \begin{table}[h]
154     \begin{center}
155     \begin{tabular}{|l|c|c|c|c|c|c|c|} \hline
156     & \multicolumn{2}{c|}{Background with genuine \Z} & \multicolumn{4}{c|}{Background without
157     genuine \Z boson} \\
158     Channel & $\Z+jets$ & $\Z b\bar{b}$ & $t\bar{t}$ & $\W+jets$ & $t\bar{t}$ + $\W+jets$ & Fit result \\ \hline
159 beaucero 1.6 $3e$ Loose &44.6 & 12.7 & 1.6 & 0.4 & 2.0 & 6.6$\pm$4.2 \\\hline
160     $3e$ Tight &13.9 & 5.0 & 0.8 & 0.4 & 1.2 & 3.8$\pm$3.5 \\\hline
161     $2e1mu$ Loose &41.5 & 78.9 & 12.6 & 0 & 12.6 & 16.9$\pm$5.5 \\\hline
162     $2e1mu$ Tight &1.0 & 2.0 & 0.9 & 0 & 0.9 & 1.5$\pm$3.2 \\\hline
163     $2mu1e$ Loose &56.3 & 15.4 & 1.9 & 0 & 1.9 & 6.9$\pm$4.4 \\\hline
164     $2mu1e$ Tight &17.3 & 5.6 & 0.8 & 0 & 0.8 & 4.1$\pm$2.5 \\\hline
165     $3mu$ Loose &43.7 & 84.9 & 12.0 & 0 & 12.0 & 11.0$\pm$5.0 \\\hline
166     $3mu$ Tight &0.8 & 2.3 & 0.3 & 0 & 0.3 & 0.8$\pm$2.8 \\\hline
167 beaucero 1.2 \end{tabular}
168     \end{center}
169 ymaravin 1.9 \caption{Comparison between Monte Carlo truth information and the results of the fit for the background
170     without genuine \Z boson. Number of events are obtained in the invariant mass range between 81 and 101 GeV. The
171     ``Loose'' and ``Tight'' selection criteria applied on the \W lepton.}
172 beaucero 1.2 \label{tab:FitNoMWt}
173     \end{table}
174    
175 ymaravin 1.9 The comparison between the estimated and MC truth backgrounds is given
176     in Tables~\ref{tab:FinalNoMWtCutLoose} and {tab:FinalNoMWtCut} for ``Loose''
177     and ``Tight'' requirements on the \W lepton. The results agree with each other
178     within one sigma of uncertainty.
179 beaucero 1.2
180     \begin{table}[h]
181     \begin{center}
182     \begin{tabular}{lcccc} \hline \hline
183     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
184 ymaravin 1.9 $N$ - \ZZ -\Z$\gamma$ &75.3$\pm$7.3&146.6$\pm$0.0&90.4$\pm$5.9&156.9$\pm$0.0\\ \hline
185     $N^{non genuine \Z}$ (Fit)&6.6$\pm$4.2&16.9$\pm$5.5&6.9$\pm$4.4&11.0$\pm$5.0\\ \hline
186     $N^{genuine \Z}$ (matrix method)&58.5 $\pm$14.4&139.2 $\pm$20.3&73.2 $\pm$14.9& 147.7 $\pm$14.7\\ \hline
187     $N^{\WZ}$ &9.5$\pm$16.4&7.4 $\pm$27.0&11.3 $\pm$17.0& 9.2 $\pm$19.0\\\hline
188     \WZ from MC &12.0&14.2& 13.6 &17.2\\
189 beaucero 1.2 \hline
190     \end{tabular}
191 ymaravin 1.9 \caption{Expected number of selected events for an integrated luminosity of 300 \invpb
192     for the signal and estimated background for 81 GeV $< M_Z < $ 101 GeV and for ``Loose''
193     \W lepton.}
194     \label{tab:FinalNoMWtCutLoose}
195 beaucero 1.2 \end{center}
196     \end{table}
197 beaucero 1.4
198 beaucero 1.6 \begin{table}[h]
199     \begin{center}
200     \begin{tabular}{lcccc} \hline \hline
201     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
202 ymaravin 1.9 $N$ - \ZZ -\Z$\gamma$ &36.0$\pm$7.1&15.9$\pm$0.0&40.2$\pm$5.7&18.0$\pm$0.0\\ \hline
203     $N^{non genuine \Z}$ (Fit)&3.8$\pm$3.5&1.5$\pm$3.2&4.1$\pm$2.5&0.8$\pm$2.8\\ \hline
204     $N^{genuine \Z}$ (matrix method)&18.7 $\pm$6.1&8.4 $\pm$6.6&23.4 $\pm$7.5& 8.9 $\pm$7.1\\ \hline
205     $N^{\WZ}$ &10.1 $\pm$8.0&7.6 $\pm$7.5&11.0 $\pm$8.9& 9.1 $\pm$7.6\\ \hline
206     \WZ from MC &11.6&12.3& 13.3 &14.9\\
207 beaucero 1.6 \hline
208     \end{tabular}
209 ymaravin 1.9 \caption{Expected number of data events for an integrated luminosity of 300 \invpb for the signal and estimated background for 81 GeV $< M_Z < $ 101 GeV and for ``Tight'' \W lepton.}
210     \label{tab:FinalNoMWtCut}
211 beaucero 1.6 \end{center}
212     \end{table}
213    
214 ymaravin 1.9 \subsubsection{Performance of the matrix method without background categorization}
215     The performance of the matrix method depends on the validity of the following two assumptions:
216     the $p_{fake}$ should describe the probability of misidentified jets passing loose criteria to also
217     pass tight lepton requirements, and there must be only one misidentified lepton. If the former
218     challenge can be tackled by estimating $p_{fake}$ in the sample with the jet composition similar
219     to that of the major \WZ background (\W+jets in this case), then the latter can be safely assumed
220     if the sources of backgrounds with multiple leptons are suppressed. The latter can be checked
221     with data by measuring comparing \W+jets cross-section measured in data with MC truth information
222     and estimating the \W+jets background to the \WZ signal. As total $\W+jets$ background to
223     \WZ signal is very small, we can neglect background contribution with multiple misidentified leptons
224     to the signal.
225    
226     Therefore, with small data sample, it might be a good approximation not to divide instrumental
227     background into genuine \Z boson and fake \Z boson categories, but apply the matrix method
228     directly to the dilepton
229     invariant mass after the physics backgrounds are subtracted. This results in smaller
230     systematic uncertainties associated with the fit. We follow this procedure and provide the
231     comparisons between predicted and true MC backgrounds
232     in Tables~\ref{tab:FinalNoFitLoose} and \ref{tab:FinalNoFit} for ``Loose'' and ``Tight''
233     \W lepton, respectively.
234 beaucero 1.4
235     \begin{table}[h]
236 beaucero 1.6 \begin{center}
237     \begin{tabular}{lcccc} \hline \hline
238     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
239 ymaravin 1.9 $N$ - \ZZ -\Z$\gamma$ &19.9$\pm$1.0&23.6$\pm$0.0&23.4$\pm$1.0&25.5$\pm$0.0\\ \hline
240     $N^{genuine \Z}$ (matrix method)&10.1 $\pm$0.6&0.9 $\pm$1.0&14.5 $\pm$0.9&15.8 $\pm$0.7\\ \hline
241     $N^{\WZ}$ &8.8 $\pm$0.6&7.7 $\pm$1.0&7.9 $\pm$0.9&9.8 $\pm$0.7\\ \hline
242     \WZ from MC &8.1&9.0& 9.2 &11.3\\
243 beaucero 1.6 \hline
244     \end{tabular}
245 ymaravin 1.9 \caption{Expected number of events for an integrated luminosity of 300 \invpb for the signal
246     and estimated background for 81 GeV $< M_Z < $ 101 GeV with ``Loose'' \W lepton criteria.}
247     \label{tab:FinalNoFitLoose}
248 beaucero 1.6 \end{center}
249     \end{table}
250    
251     \begin{table}[h]
252     \begin{center}
253     \begin{tabular}{lcccc} \hline \hline
254     & 3e &2e1$\mu$ &2$\mu$1e &3$\mu$\\ \hline
255 ymaravin 1.9 $N$ - \ZZ -\Z$\gamma$ &12.4$\pm$1.0 &8.7$\pm$0&13.1$\pm$0.9&10.6$\pm$0.0\\ \hline
256     $N^{genuine \Z}$ (matrix method)&3.2 $\pm$1.6&15.8 $\pm$0.7&4.6 $\pm$1.9&0.9 $\pm$1.1\\ \hline
257     $N^{\WZ}$ &8.2 $\pm$1.6&7.8 $\pm$0.7&7.6 $\pm$1.9&9.6$\pm$1.1\\ \hline
258     \WZ from MC &7.9&8.1& 9.0 &10.1\\ \hline
259 beaucero 1.4 \end{tabular}
260 ymaravin 1.9 \caption{Expected number of events for an integrated luminosity of 300 \invpb for the signal
261     and estimated background for 81 GeV $< M_Z < $ 101 GeV and ``Tight'' \W lepton requirement.}
262     \label{tab:FinalNoFit}
263 beaucero 1.4 \end{center}
264     \end{table}
265 ymaravin 1.9 The agreement between estimated and MC true backgrounds is excellent.