1: \section{Overview of Analysis Method}
2: \label{sec:strategy}
3:
4: The analysis strategy follows closely that of Ref. \cite{ichep04}.
5: The starting point is the selection of a sample of events
6: in which the hadronic decay of one of the two \B mesons (\breco) is fully
7: reconstructed.
8: About 1000 different $B \rightarrow D Y$ decay chains are selected,
9: where $D$ refers to a
10: charm meson and $Y$ represents a collection of hadrons with total charge
11: $\pm 1$ and composed of $n_1 \pi^{\pm} + n_2 K^{\pm} + n_3 \KS + n_4\piz$,
12: with $n_1 + n_2 < 6$, $n_3 < 3$ and $n_4 < 3$.
13:
14: We reconstruct $D^-\ra K^+\pi^-\pi^-$, $K^+\pi^-\pi^-\piz$, $\KS\pi^-$,
15: $\KS\pi^-\piz$, $\KS\pi^-\pi^-\pi^+$;
16: $D^{*-}\ra\Dzb\pi^-$;
17: $\Dzb\ra K^+\pi^-$,
18: $K^+\pi^-\piz$, $K^+\pi^-\pi^-\pi^+$, $\KS\pi^+\pi^-$;
19: and $\Dstarzb \ra\Dzb\piz, \Dzb\gamma$.
20: Then we use $D^-$ and $D^{*-}$ ($\Dzb$ and $\Dstarzb$) decays as a ``seed'' to
21: reconstruct $B^0$ ($B^+$) decays.
22: Overall, we correctly reconstruct one $B$ candidate in
23: 0.3\% (0.5\%) of the \BzBzb\ (\BpBm) events.
24: The kinematic consistency of a \breco candidate with a \B meson decay is
25: checked using two variables: the beam-energy substituted mass
26: $\mes = \sqrt{s/4 - \vec{p}^{\,2}_B}$
27: and the energy difference $\Delta E = E_B - \sqrt{s}/2$.
28: Here $\sqrt{s}$ refers to the total
29: energy in the \FourS center of mass frame, and $\vec{p}_{B}$ and
30: $E_B$ denote
31: the momentum and energy of the \breco\ candidate in the same frame.
32: For signal events the \mes\ and $\Delta E $ distributions peak at the \B meson mass
33: and at zero, respectively.
34:
35: The combinatorial background from \BB\ events and
36: $e^+e^-\to q\bar{q}$ ($q=u,d,s,c$) production, in the \breco\ sample, is subtracted by performing an unbinned likelihood fit
37: to the \mes\ distribution, using the following threshold function \cite{argus}
38:
39: \begin{equation}
40: \label{eq:argus}
41: \frac{dN}{d\mes} = N\cdot \mes \cdot \sqrt{1-x^2} \cdot \exp \left(- \xi \cdot (1 - x^2)\right)\,
42: \end{equation}
43:
44: \noindent for the background (where $x = \mes/m_{\rm max}$ and $m_{\rm max}$ is the endpoint of the curve)
45: and a Gaussian function corrected for radiation losses
46: \cite{crystallball} peaked at the \B meson mass for the signal (Fig. \ref{fig:mesfit}).
47:
48: \begin{figure}[!b]
49: \begin{centering}
50: \epsfig{file=ps/fig01a.eps,width=6.7cm}
51: \epsfig{file=ps/fig01b.eps,width=6.7cm}
52: \epsfig{file=ps/fig01c.eps,width=6.7cm}
53: \epsfig{file=ps/fig01d.eps,width=6.7cm}
54: \caption{ Fit to the \mes distribution for events with a fully reconstructed \Bz (left) or \Bp (right) decay, after the request of a prompt energetic lepton (top) and after all selection criteria for \bpigen decays (bottom). The fitted curve (black line) to the data points (black dots) is the sum of a radiation loss corrected gaussian and a threshold function described by Eq.~\ref{eq:argus} (dashed line).
55: \label{fig:mesfit}}
56: \end{centering}
57: \end{figure}
58:
59: Once a \breco meson has been reconstructed in the event, the selection of
60: the other \B meson (\bsig) decay follows in two steps. First, a ``semileptonic selection'' is applied to obtain
61: a sample of semileptonic decays. Then, a refined selection is applied to
62: build a sample of $\bpigen$ decays, as described in Sec.~\ref{sec:selection}. In order to minimize the systematic uncertainties, the exclusive branching fractions are measured relative to the inclusive semileptonic one.
63:
64: The semileptonic selection is based essentially on the presence of a charged (electron or muon) lepton
65: with its momentum $p^{*}_{lep}$ in the \bsig rest frame
66: greater than a given $p^{*}_{cut}$. The number of \Bxlnu\ events, $N_{sl}^{meas}$, and the number of
67: remaining background events, $BG_{sl}$, peaking at the \B mass in the \mes distribution and estimated with the Monte Carlo
68: simulation, are related to the true number of semileptonic
69: decays $N_{sl}^{true}$ as
70: \begin{equation}
71: N_{sl}^{meas} - BG_{sl} = \epsilon_l^{sl} \epsilon_t^{sl}N_{sl}^{true}.
72: \end{equation}
73: Here $\epsilon_l^{sl}$ refers to the efficiency for selecting a lepton
74: from a semileptonic \B decay with a momentum above $p^{*}_{cut}$ in an
75: event with a hadronic \B decay, reconstructed with tag efficiency $\epsilon_t^{sl}$.
76: The background normalization is
77: taken from the Monte Carlo scaled to the luminosity of the data sample.
78:
79: The full reconstruction of the \breco meson allows for a precise determination
80: of the neutrino four-momentum $p_{\nu}$,
81: estimated from the missing four-momentum in the event:
82:
83: \begin{equation}
84: \label{eq:pmiss}
85: p_\nu = p_{miss} = p_{\Upsilon(4S)} - p_{\breco} - p_{\pi} - p_\ell,
86: \end{equation}
87:
88: \noindent where all momenta are measured in the laboratory frame: $p_{\Upsilon(4S)}$ is the sum of
89: the four-momenta of the colliding beams, $p_{\breco}$ is the measured four-momentum of the \breco,
90: $p_{\pi}$ is the measured four-momentum of the $\pi^-$ or $\piz$ and $p_{\ell}$ is the
91: measured four-momentum of the lepton. For signal events the only missing particle should
92: be a single undetected neutrino, while for background events the missing momentum and energy in the event
93: are due to other undetected or poorly measured particles.
94: Then, in signal events the resulting missing mass, defined as $\mmiss$ = $p_{miss}^2$, peaks at zero
95: and for background events
96: it tends to have larger values, and allows for a powerful discrimination of
97: signal and background.
98:
99: The number of $\bpigen$ events after the combinatoric subtraction
100: in a given $q^2$ range, $N_{excl}^{meas}$, and the number of peaking background events, $BG_{excl}$,
101: are related to the true number of
102: $\bpigen$ events $N_{excl}^{true}$ as
103: \begin{equation}
104: N_{excl}^{meas} - BG_{excl} = \epsilon_{sel}^{excl} \epsilon_l^{excl}
105: \epsilon_t^{excl} N_{excl}^{true} ,
106: \label{eq:NBGexcl}
107: \end{equation}
108: \noindent
109: where the signal efficiency $\epsilon_{sel}^{excl}$ accounts for all selection
110: criteria applied to the sample after the requirement of
111: a charged lepton with momentum $p^*_{lep} > p^*_{cut}$.
112:
113: The ratio between the partial branching fractions of the signal in a particular $q^2$ region
114: and the branching fraction of $\Bxlnu$ decays is
115: \begin{equation}
116: R_{excl/sl}=
117: \frac{\Delta\BR(\bpigen)}{\BR(\Bxlnu)}=
118: \frac{N_{excl}^{true}}{N_{sl}^{true}} =
119: \frac{(N_{excl}^{meas}- BG_{excl})/(\epsilon_{sel}^{excl})}
120: {(N_{sl}^{meas}-BG_{sl})}
121: \times \frac{\epsilon_l^{sl} \epsilon_t^{sl} }
122: {\epsilon_l^{excl} \epsilon_t^{excl} }.
123: \label{eq:ratioBR}
124: \end{equation}
125:
126: \noindent The ratio of efficiencies for $\Bxlnu$ and signal
127: events in Eq. \ref{eq:ratioBR} is expected to be close to, but not equal to
128: unity. Due to the difference in multiplicity and the different
129: lepton momentum spectra, we expect the tag efficiencies $\epsilon_t^{sl,excl}$
130: and the charged lepton efficiencies $\epsilon_l^{sl,excl}$ to be slightly different for the
131: two classes of events.
132: Finally using the semileptonic branching ratio $\BR(\Bxlnu) = (10.73 \pm 0.28)\%$ \cite{PDG2004}
133: and the ratio of the \Bz and \Bp lifetimes $\tau_{B^+}/\tau_{B^0}=1.086 \pm 0.017$
134: \cite{PDG2004}, the partial branching ratios $\Delta\BR(\bpi)$ and $\Delta\BR(\bpiz)$ can be derived.\\
135: