1: %% $Id: analysis.tex,v 1.10 2006/07/24 20:33:19 chcheng Exp $
2: \section{TIME-DEPENDENT ANALYSIS}
3: \label{sec:analysis}
4:
5: We model the time-dependent Dalitz plot distribution in a
6: PDF that consists of four components: signal, background with a real
7: \Dz, background with a fake \Dz and background that peaks in both \mES
8: and \mDz. The $(\mES,\mDz)$ model for the first three components has
9: been described in Section~\ref{sec:selection}. The peaking background
10: component shares the same $(\mES,\mDz)$ shape with the signal component.
11: The background fractions are determined from
12: a fit to $(\mES,\mDz)$ distributions and from generic Monte Carlo
13: samples (for peaking background) for each \Bz mode group and each
14: tagging category. Each event is assigned signal and background
15: probabilities based on the two-dimensional PDF.
16:
17: The time-dependent Dalitz model for signal is based on
18: Equation(\ref{eq:Probpm}). We modify the equation to take into account
19: mistagging and imperfect \dt reconstruction, following the methods
20: used in our other time-dependent analyses~\cite{tagging}, i.e.,
21: an additional factor $(1-2 w)$ is added to the
22: $\cos(\dm\dt)$ and $\sin(\dm\dt)$ terms, and the equation is convolved
23: with a three-Gaussian \dt resolution function. There are six tagging
24: categories with different mistag fractions $w$. We also allow
25: the $w$ of each category to be different for \Bz and \Bzb tags.
26: The means and widths of the core and the
27: second Gaussian are parameterized with scale factors multiplied by
28: \sigmadt. The mean and width of the third (outlier) Gaussian are fixed
29: at 0~ps and 8~ps, respectively. The mistag rates and the resolution
30: function are determined from control samples of $\Bz\ra
31: D^{(*)}\pi,\rho, a_1 $ decays. Most of the resolution function
32: parameters are consistent among the six tagging
33: categories, except for the core Gaussian mean and scale
34: factor, where the Lepton tagged sample is significantly different from
35: others. We allow these two parameters to be different for Lepton tag.
36:
37: The model for the background with a fake \Dz is determined from the
38: \Dz sideband data. The \dt model consists of a prompt component and a
39: exponential decay component with an effective lifetime. The resolution
40: function is a Gaussian whose mean and width are scaled by \sigmadt,
41: plus an outlier Gaussian. The mean of the core Gaussian and
42: the fraction of the prompt component are allowed to be different
43: between the Lepton tag and the other tags.
44:
45: The Dalitz distribution for background is modeled by an incoherent
46: mixture of several
47: resonances and a phase-space distribution,
48: \begin{equation}
49: \label{eq:dalitzbkg}
50: {\cal P}(m_+^2,m_-^2) = |a_\mathrm{NR}|^2 + \sum_r |a_r|^2 |A_r(m_+^2,m_-^2)|^2
51: \;.
52: \end{equation}
53: We find that the model describes the \Dz sideband data well if we
54: include $\Kstar(892)^-$, $\Kstar(892)^+$,
55: $\Kstar_0(1430)^-$, $\rho(770)$, $\rho(1450)$ and $\sigma$ resonances
56: in the model. We also check that the Dalitz distribution is
57: independent of the tagging category, the flavor tag, and \dt.
58:
59: Based on a study using the generic Monte Carlo samples, the background
60: with a real
61: \Dz comes mostly from \ccbar continuum events. We therefore model the
62: \dt distribution with a prompt component convolved with a core Gaussian plus
63: outlier resolution function. The parameters are determined from the
64: generic Monte Carlo sample.
65:
66: The Dalitz model for this background is either $A_D(m_+^2,m_-^2)$ or
67: $A_{\Dbar}(m_+^2,m_-^2)$ based on the flavor of the tagging side \Btag.
68: If \Btag is tagged as \Bz (\Bzb), the $D$ in the reconstructed
69: candidate is more likely to be a \Dz (\Dzb).
70: Since they are not \BB events, the mistag rates are not the same as
71: those for the signal. However, we don't have reliable mistag values
72: for continuum events. We therefore use the mistag rates for signal in the
73: nominal fit and vary them to estimate the systematic uncertainty.
74:
75: The peaking background being mostly from charged \B decays, the \dt model is an
76: exponential decay with the lifetime fixed at the
77: \Bp lifetime.
78: The Dalitz model is identical to that for
79: combinatorial background with real \Dz, except that the mistag rates
80: can be different. Again we fix the mistag rates to those for the signal, and
81: vary them for systematic uncertainty.
82:
83:
84: In the nominal fit, we allow \cosbb, \sinbb and \abslambda to float
85: and fit to all data samples and tagging categories simultaneously.
86: The \Bz lifetime and mixing frequency are fixed at the PDG values.
87: The fit results are shown in Table~\ref{tab:result}, where the result
88: for which the \sinbb is fixed at the world average and \abslambda at
89: one is also included. We also allow the
90: \mES shape and
91: background fractions to float in the fit. We find no significant
92: difference in either the central values or the statistical
93: uncertainties. The projections on the Dalitz plot variables are shown in
94: Figure~\ref{fig:datatoyDproj1} and are compared with the distributions
95: described by the model. Figure~\ref{fig:d0rhoasym} shows the
96: time-dependent \CP asymmetry for events in $\Dz\ra\KS\rho(770)$ region
97: ($|m(\pip\pim)-0.77|<0.25~\gevcc$), where the \CP asymmetry is
98: expected to be enhanced.
99: The apparent asymmetry in Figure~\ref{fig:d0rhoasym} is small compared
100: to \sinbb due to
101: the dilution factors from mistagging, background and contributions
102: from non-resonance and resonances other than $\rho$.
103:
104: \begin{table}[htb]
105: \caption{Results of the fits to data. Errors are statistical only.}
106: \begin{center}
107: \begin{tabular}{lccc}
108: \hline\hline
109: Final state & \cosbb & \sinbb & \abslambda \\
110: \hline
111: $\Dz\piz$ & $1.1^{+0.8}_{-0.9}$ & $1.0\pm0.5$ &
112: $1.13^{+0.17}_{-0.14}$ \\
113: $\Dz\eta^{(\prime)}$ & $0.4\pm1.1$ & $-0.1^{+0.9}_{-1.0}$ &
114: $0.96^{+0.19}_{-0.16}$ \\
115: $\Dz\omega$ & $-0.4^{+1.3}_{-1.4}$ & $0.7\pm 1.0$ &
116: $0.61^{+0.17}_{-0.15}$ \\
117: $\Dstarz\piz/\eta$ & $0.3\pm1.4$ & $-0.8^{+1.0}_{-0.9}$ &
118: $1.05^{+0.35}_{-0.25}$ \\
119: \hline
120: All & $0.54\pm0.54$ & $0.45\pm0.35$ & $0.98\pm 0.09$ \\
121: All & $0.55 \pm 0.52$ & $0.685$ (fixed) & 1 (fixed) \\
122: \hline\hline
123: \end{tabular}
124: \end{center}
125: \label{tab:result}
126: \end{table}
127:
128:
129:
130: \begin{figure}[htb]
131: \begin{center}
132: \includegraphics[width=0.98\textwidth]{sc2b-nominal-DalitzProj-1.eps}
133: \end{center}
134: \caption{Projection on (a,d) \msqKsp, (b,e) \msqpm, and (c,f) \msqKsm
135: for (a,b,c)
136: \Bz-tagged and (d,e,f) \Bzb-tagged events separately, in the signal region
137: ($\mES>5.27~\gevcc$, $|\mDz-\mDz^\mathrm{PDG}|<20~\mevcc$).
138: Points with error bars are data; histograms are from the PDF.}
139: \label{fig:datatoyDproj1}
140: \end{figure}
141:
142:
143: \begin{figure}[htb]
144: \begin{center}
145: \includegraphics[width=0.5\textwidth]{D0RhoAsym.eps}
146: \end{center}
147: \caption{Asymmetry distribution for the events in
148: $\Dz\ra\KS\rho(770)$ region. The curve is the result of the
149: PDF.}
150: \label{fig:d0rhoasym}
151: \end{figure}
152:
153:
154:
155:
156:
157:
158: