1: The basic analysis technique is based on \BB\ events in which one of the $B$ mesons decays hadronically and is fully reconstructed ($B_{reco}$) and the recoiling $B$ ($B_{recoil}$) decays semileptonically.
2: The requirement of such a decay topology allows the \B\ momentum to be determined, so that transformations to the rest frame of the recoiling \B\ meson are possible. It also results in a clean sample of \BB\ events, with the flavor of the reconstructed \B\ meson determined.
3:
4: \subsection{Selection of Hadronic $B$ Decays, $B \ra D Y$}
5:
6:
7: The reconstruction of $B$ decays is designed
8: to identify samples of decays of the type $B \rightarrow D Y$. Here $D$ refers to a charm meson and $Y$ represents a collection of hadrons of total charge $\pm 1$, composed of $\pi^{\pm}, K^{\pm}, \KS$ and $\piz$ mesons. The charm meson serves as a ``seed'' for the selection of candidates. We use four different seeds: $D^+$ and $D^{*+}$ for $\Bzb$ and $D^0$ and $D^{*0}$ for $B^-$ (charge conjugate modes are implied throughout this document). Several different decay modes are used, each of which is characterized by a signal-to-background ratio dependent on the multiplicity and on the composition of the $Y$ system. In events with more than one reconstructed $B$ decay, the decay mode with the highest {\it a priori} signal-to-background ratio is selected.
9:
10: $B_{reco}$ candidates are identified by two kinematic variables,
11: the beam energy-substituted mass, $M_{ES}=\sqrt{E^{*2}_{beam} - p_B^{*2}}$,
12: and the energy difference in the \FourS\ rest frame,
13: $\Delta E = E_B^* - E^*_{beam}$. Here $E^{*}_{beam}$ is the beam energy,
14: and $p_B^{*}$ and $E_B^*$ refer to the $B$ momentum and energy in the center-of-mass frame. $B_{reco}$ candidates are required to have $\Delta E$ within three standard deviations of zero and $M_{ES} > 5.27 \gevcc$. A sideband region in $M_{ES}$ used for background subtraction is defined by $5.20 < M_{ES} < 5.27 \gevcc$.
15: An example for the $M_{ES}$ distribution can be found in Figure~\ref{fig:mes}.
16:
17:
18: \subsection{Selection of Semileptonic Decays, $B \ra X \ell \nu$}
19:
20: Semileptonic $B$ decays are identified by the presence of one and only one electron or muon above a minimum momentum $P^*_{min}$ measured in the rest frame of the $B$ meson recoiling against the $B_{reco}$. $P^*_{min}$ is
21: varied for the different measurements in the range of $0.9 -1.6 \gevc$.
22: This requirement on the lepton momentum reduces backgrounds from secondary charm or $\tau^{\pm}$ decays, and from hadrons faking leptons.
23:
24: To further improve the background rejection, we require that the lepton charge and the flavor of the reconstructed $B^{\pm}$ be consistent with a prompt semileptonic decay of the $B_{recoil}$. For a reconstructed \Bz we do not impose this restriction, so as not to lose events due to $\Bz - \Bzb$ mixing.
25:
26:
27: Electrons are identified using a likelihood-based algorithm, combining the track momentum with the energy, position, and shape of the shower measured in the EMC, the Cherenkov angle and the number of photons measured in the DIRC, and the specific energy loss in the DCH.
28: The efficiency of these selection requirements has been measured with radiative Bhabha events as a function of the laboratory momentum $p_{lab}$ and polar angle $\theta_{lab}$, and has been corrected for the higher multiplicity of \BB events using Monte Carlo simulations. The average electron selection efficiency is ($90 \pm 2$)\%.
29:
30: Muons are identified using a cut-based selection for minimum ionizing tracks, relying on information from the finely segmented instrumented flux return of the magnet. The number of interaction lengths traversed by the track, the spatial width of the observed signals, and the match between the IFR hits and the extrapolated charged track are used in the selection.
31: %
32: The muon identification efficiency has been measured with $\mu^+\mu^-
33: (\gamma)$ events and two-photon production of $\mu^+\mu^-$ pairs.
34: The average muon selection efficiency is $(70\pm4)\%$.
35:
36: The misidentification probabilities for pions, kaons, and protons have been extracted from selected control samples in data. They vary between
37: 0.05\% and 0.1\% (1.0\% and 3\%) for the electron (muon) selection.
38:
39: To further reduce backgrounds we require the total charge of the event to be $|Q_{tot}|= |Q_{Breco} + Q_{Brecoil}| \leq 1$, and restrict the total missing mass to $|M_{miss}^2| < 1.0~[\gevcc]^2$.
40: We explicitly allow for a charge imbalance to reduce the dependence on the exact modeling of charged particle tracking in the Monte Carlo simulation, especially at low momenta, and the production of tracks from photon conversions.
41:
42: Figure ~\ref{fig:mes} shows the $M_{ES}$ distribution for events with a $B_{reco}$ candidate and a charged lepton candidate from the recoiling $B$ meson. A sideband subtraction is performed in bins of $M_X$ to remove the background contribution under the peak at the $B$ mass. For this purpose a sum of background (approximated by a function first introduced by the ARGUS Collaboration~\cite{Albrecht:1987nr}) and signal (described by a function first used by the Crystal Ball experiment~\cite{Skwarnicki:1986xj}) is fitted to the $M_{ES}$ distribution of the data. For $M_{ES} > 5.27 \gevcc$ we find $5819$ signal events and $3585$ background events. The fit to the $M_{ES}$ distribution is shown in Figure ~\ref{fig:mes}.
43: \begin{figure}[t]
44: \begin{center}
45: % \mbox{\epsfig{file=mes_0.9.eps,height=10cm,width=15.5cm}}
46: \mbox{\epsfig{file=mes_vera.eps,height=10cm,width=15.5cm}}
47: %\vspace{+0.5cm}
48: \caption{\em The $M_{ES}$ distribution for
49: selected events with one reconstructed $B$ decay and a lepton with $P^*_{min}=0.9 \gevc$.
50: \label{fig:mes} }
51: \end{center}
52: \end{figure}
53: \subsection{Reconstruction of the Hadronic Mass $M_X$}
54:
55: \label{subsec:kfit}
56:
57: The hadron system in the decay $B \ra X \ell \nu$ is formed from $K^{\pm}$, $\pi^{\pm}$, and photons that are not associated with the $B_{reco}$ candidate or identified as leptons. Specifically, we select charged tracks in the fiducial volume
58: $0.41 < \theta_{lab} < 2.54~{\rm rad}$ with a minimum transverse momentum $p_t > 120 \mevc$. For photons we require an energy $E_{lab}>100 \mev$ and $0.41 < \theta_{lab} < 2.409~{\rm rad}$.
59:
60: To improve the resolution on the measurement of the hadronic mass $M_X$, we exploit the kinematic constraints of the $\BB$ state by performing a kinematic fit to the full event using the measured momenta and energies of all particles.
61: The measured four-momentum $P_X^m$ of the $X$ system can be written as
62:
63: \begin{equation}
64: P_X^m = \sum_{i=1}^{N_{ch}} P^{ch}_i
65: + \sum_{j=1}^{N_{\gamma}} P^{\gamma}_j
66: \end{equation}
67: \noindent
68: where $P$ are four-momenta and the indices $ch$ and $\gamma$
69: refer to the selected charged tracks and photons, respectively. Depending on particle identification the charged tracks are assigned either the $K^{\pm}$ or $\pi^{\pm}$ mass.
70:
71:
72:
73: The missing momentum in the event is reconstructed by relating the sum of the four-momenta of the colliding beams, $Q_{CM}$,
74: to the measured four-momenta of the $B_{reco}$ candidate, the $X$ hadrons, and the charged lepton,
75: \begin{equation}
76: P_{miss}=Q_{CM}-P_{reco}^m-P_X^m - P_{\ell}^m .
77: \end{equation}
78: The measured invariant mass squared, $M_{miss}^2 = P_{miss}^2$, is an important discriminant on the quality of the reconstruction of the total recoil system.
79: Any secondary particles from the decay of the hadronic $X$ system that are undetected or poorly measured will impact the measurement of both $M_X$ and $M_{miss}^2$. Likewise any sizable energy loss of the leptons via bremsstrahlung or internal radiation will impact the measurement of these two quantities. The effect of initial state radiation is rather small, due to the narrow width of the \FourS\ resonance.
80: \begin{figure}[t]
81: \begin{center}
82: %\begin{picture}(15.,15.)
83: %\mbox{\epsfig{file=reso_mmiss_vcb_mod.eps,height=12.4cm,width=15.5cm}}
84: \mbox{\epsfig{file=reso_mmiss_vcb_mod2.eps,scale=1.5}}
85: %\put(0.0,0.0){\mbox{\epsfig{file=reso_test.eps,height=13cm,width=15.5cm}}}
86: %\mbox{\epsfig{file=reso_test.eps,height=13cm,width=15.5cm}}
87: %\end{picture}
88: %\vspace{+0.1cm}
89: \caption{\em Monte Carlo comparison of the reconstructed and fitted mass $M_X$: Mean and r.m.s. of the distribution $\Delta M_X= M_X^{True}-M_X^{Reco}$ (open circles) and $\Delta M_X= M_X^{True}-M_X^{Fit}$ (dots) as a function of the measured missing mass squared $M_{miss}^2$.
90: \label{scanMmiss_vcb} }
91: \end{center}
92: \vspace{-0.7cm}
93: \end{figure}
94: We exploit the available kinematic information from the full event, namely the $B_{reco}$ and the $B_{recoil}$ candidate, by performing a 2C kinematic fit that imposes four-momentum conservation, the equality of the masses of the two $B$ mesons, $M_{recoil}=M_{reco}$, and forces $M_{miss}^2 = M_{\nu}^2 = 0$.
95: The fit takes into account event-by-event the measurement errors of all individual particles and the measured missing mass.
\unitlength1.0cm % coordinates in cm
96: \begin{figure}[t]
97: \begin{center}
98: \begin{picture}(15.,10.)
99: \put(-1.2,0.0){\mbox{\epsfig{file=datafit_1.0_0.9_1.eps, scale=.8}}}
100: \end{picture}
101: \end{center}
102: \vspace{-0.5cm}
103: \caption{\em Sideband-subtracted $M_X$ distribution for $P^*_{min}=0.9 \gevc$ (left) and $P^*_{min}=1.5\gevc$ (right). The hatched histograms show the contributions from $B \rightarrow D^* \ell \nu$, $B \rightarrow D \ell \nu$ and $B \rightarrow X_H \ell \nu$ decays as determined by the fit, as well as the background distribution. The white histogram represents the sum of all the aforementioned distributions.
104: The histograms extend over the range in $M_X$ that was used in the fit to extract $\langle M_X^2 - \overline{m}_D^2 \rangle$.}
105: \label{fig:mxdist}
106: \end{figure}
107:
108: Figure \ref{scanMmiss_vcb} shows a comparison for $B\ra X_c \ell \nu$ events between the reconstructed and the kinematically fitted mass $M_X$ as a function of the measured $M_{miss}^2$, in terms of the mean and r.m.s. of the distributions for $\Delta M_X= M_X(true)-M_X$.
109:
110: The constraints of energy and momentum conservation and the
111: equal-mass hypothesis lead not only to a significant improvement
112: in the mass resolution of the $X$ system but also provide an almost unbiased estimator of the mean and a resolution that is largely independent of $M_{miss}^2$.\\
113: