hep-ex0408068/analysis.tex
1: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
2: \section{Common Analysis Aspects}
3: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
4: \label{sec:strategy}
5: The event selection and reconstruction and the measurements of branching fractions 
6: follow closely the strategy described in Ref.~\cite{Aubert:2003zw} and represents the common 
7: base for the analyses presented here. However, each analysis may differ in some details. 
8: This is particularly true for the measurement of exclusive branching fractions. 
9: This section describes aspects in common to all studies. Details specific to each 
10: analysis are provided in the following sections. 
11: 
12: \subsection{Event Reconstruction and Selection}
13: 
14: In this paper  we study the recoil of   fully reconstructed $B$ 
15: in hadronic decay modes (\breco), 
16: which is a moderately pure 
17: sample with known flavor and four-momentum.
18: We select \breco\ decays of the type  $B \rightarrow \Db  Y$,
19: where $D$ refers to a charm  meson, and $Y$ represents a collection of
20: hadrons with  a total  charge of $\pm  1$, composed  of $n_1\pi^{\pm}+
21: n_2K^{\pm}+ n_3\KS +  n_4\piz$, where $n_1 + n_2 < 6$,  $n_3 < 3$, and
22: $n_4 <  3$.  Using $D^-$  and $D^{*-}$ ($\Dzb$ and  $\Dstarzb$) as
23: seeds for  $B^0$ ($B^+$) decays,  we reconstruct about  1000 different
24: decay chains. Overall, we correctly reconstruct one $B$ candidate in 0.3\%  (0.5\%) of the 
25: \BzBzb\ (\BpBm) events.
26: The kinematic consistency of a $B_{reco}$ candidate with a $B$ meson
27: decay is checked using two variables, the beam-energy-substituted mass
28: $\mes = \sqrt{s/4 - \vec{p}^{\,2}_B}$
29:  and the energy difference, 
30: $\Delta E  = E_B  - \sqrt{s}/2$. Here  $\sqrt{s}$ refers to  the total
31: energy in the \FourS center of  mass frame, and $\overrightarrow{p_B}$ and $E_B$ denote
32: the momentum and energy of the $B_{reco}$ candidate in the same frame. 
33: For signal events the \mes\ distribution peaks at the $B$ meson mass, while
34: $\Delta E $ is consistent with zero. We  require $\Delta E  = 0$ within  approximately three
35: standard deviations. 
36: 
37: 
38: A semileptonic decay of the other $B$ meson (\brecoil) is identified by the
39: presence of a charged lepton with momentum in the \brecoil\ rest frame ($p^*$) higher than 1\gevc\ . 
40: In addition, the detection of missing
41: energy and momentum in the event is taken as evidence for the presence
42: of a neutrino. 
43: The hadronic system $X$  is reconstructed from charged
44: tracks  and energy depositions
45: in the calorimeter that are not associated with
46: the \breco\ candidate or the identified lepton.  Care is taken to eliminate 
47: fake charged tracks, as well as low-energy beam-generated photons and energy depositions in the calorimeter from charged and
48: neutral hadrons.  The
49: neutrino four-momentum $p_{\nu}$ is estimated from the
50: missing momentum four-vector $p_{miss} = p_{\Upsilon(4S)}-p_{\breco} -p_X-p_\ell$, 
51: where all momenta are measured in the laboratory frame, and
52: $p_{\Upsilon(4S)}$ refers to the  \FourS\ meson. 
53: 
54: Undetected particles and measurement uncertainties affect the determination 
55: of the four-momenta of the  $X$ system and neutrino, 
56:  and lead to a large leakage of \Bxclnu\ background from the high \mx\
57:  into the low \mx\ region.
58: To improve the resolution on these four-momenta this analysis
59: exploits the well-known kinematics of the $e^+e^-$\to\FourS\to\BB\ process 
60: and performs a two-constraint kinematic fit to the whole event.  
61: 
62: In the sample of reconstructed $B$ decays two backgrounds need
63: to be considered: the combinatorial background from \BB\ and continuum events, 
64: due to random association of tracks and neutral clusters, which does not peak
65: in \mes, and the \BB\ background whose \mes\ distribution has the same shape as the signal.
66: After applying all selection criteria, the remaining combinatorial
67: background is subtracted by performing an unbinned likelihood fit
68: to the \mes\ distribution. In this fit, the combinatorial
69: background originating from $e^+e^-\to q\bar{q}$ ($q=u,d,s,c$) continuum
70: and \BB\ events is described by an empirical threshold function~\cite{Albrecht:1993pu}, and 
71: the signal is described by a modified Gaussian~\cite{cry} peaked at the $B$ meson mass. 
72: 
73: To reject the background in the sample of semileptonic decays we require exactly one
74: charged lepton with $p^*>$1\gevc, 
75: a total event charge of zero, and a missing mass consistent with zero
76: ($\mmiss < 0.5 \gev^2/c^4$).  These criteria partly suppress the dominant
77: \Bxclnu\ decays, many of which contain an additional neutrino or
78: an undetected $\KL$ meson.  We explicitly veto
79: the $\Bzb\to\Dstarp\ell^-\overline{\nu}$ background by searching candidates for such a decay with a partial reconstruction technique, that is 
80: only identifying
81: the $\pi^+_s$ from the $\Dstarp\to \Dz\pi_s^+$ decay and the
82: lepton: since the momentum of the $\pi^+_s$ is almost collinear with the
83: \Dstarp\ momentum in the laboratory frame, we can approximate the
84: energy of the \Dstarp\ as $E_{\Dstarp}
85: \simeq m_{\Dstarp} \cdot E_{\pi_s} /145 \mevcc$ and estimate the neutrino mass as
86:   $  m_{\nu}^2  =  (p_B   -  p_{\Dstarp}  -  p_{\ell})^2 $. Events with $  m_{\nu}^2 >- 3  \gev^2/c^4$
87: are likely to be background events and are rejected.
88: Finally, we  veto events  with charged or  neutral kaons  in the
89: recoil  \Bb\ to reduce  the peaking background  from \Bxclnu\  decays.  Charged
90: kaons are identified \cite{Aubert:2001tu}
91: with an efficiency varying between 60\% at the highest and almost 100\% at the
92: lowest  momenta.  The  pion   misidentification rate is  about  2\%.   The
93: $\KS\to\pi^+\pi^-$  decays  are reconstructed  with  an efficiency  of
94: $80\%$ from pairs of oppositely  charged tracks with an invariant mass
95: between 486  and   510  \mevcc. 
96: 
97: 
98: \subsection{Measurement of Charmless Semileptonic Branching Ratios}
99: 
100: 
101: To  reduce the systematic uncertainties in the 
102: derivation of branching fractions,  the observed
103: number  of  signal events,  corrected   for  peaking background  and  efficiency,  is
104: normalized to the total number of semileptonic decays \Bxlnu\
105:  in the recoil of the $B_{reco}$ candidates. 
106: The number of observed $B_{reco}$ events which contain a charged
107: lepton  with $p^*>$ 1\gevc is denoted as
108: $N_{sl}^{meas}$. It can be related to the true number of semileptonic
109: decays, $N_{sl}^{true}$ and the remaining peaking
110: background $BG_{sl}$, estimated with Monte Carlo simulation, by
111: $N_{sl}^{true} = (N_{sl}^{meas} - BG_{sl})/\epsilon_l^{sl} \epsilon_t^{sl}=
112: N_{sl}/\epsilon_l^{sl} \epsilon_t^{sl}$ .
113: Here $\epsilon_l^{sl}$ refers to the efficiency for selecting a lepton
114: from a semileptonic B decay with a momentum above $p_{cut}$ in an
115: event with a reconstructed $B$ with efficiency $\epsilon_t^{sl}$. Figure \ref{fig:mxhadfit} shows the 
116: result of the \mes\ fit used to determine $N_{sl}^{meas}$. 
117: \begin{figure}[htb]
118:  \begin{centering}
119: \epsfig{file=plots/mes-nsl.eps,width=0.5\linewidth}
120: \caption{ Fit to the $\mes$ distribution for the sample with high momentum lepton. 
121: \label{fig:mxhadfit}}
122:  \end{centering}
123: \end{figure} 
124: 
125: 
126: 
127: If we denote as $N_u^{meas}$ the number of  events fitted in the
128: sample  after all requirements, and  with $BG_u$ the peaking background
129: coming from semileptonic decays other than the signal, the true number of signal events $N_u^{true}$
130: is related to them by
131: \begin{equation}
132: N_u= N_u^{meas} - BG_u = \epsilon_{sel}^u  \epsilon_l^u \epsilon_t^u N_u^{true} ,
133: \end{equation}
134: \noindent
135: where the signal efficiency $\epsilon_{sel}^u $ accounts for all selection
136: criteria applied on the sample after the requirement of 
137: a high momentum lepton.  
138: 
139: To measure $BG_u$ in the inclusive studies, the peaking background ($BG_u$) is estimated by performing a $\chi^2$ fit on 
140: the \mx\ or \mx-\Q\ distributions, resulting from \mes\ fits in individual \mx\ or \mx-\Q\ bins, 
141: with the shape of the background estimated from Monte Carlo simulation,
142: and its normalization free to vary. An 
143: example of such a $\chi^2$ fit for the \mx\ analysis is shown in Fig.~\ref{mxfit}. 
144: For the exclusive analysis the background normalization is 
145: taken from Monte Carlo scaled to the data luminosity. 
146: \begin{figure}
147:  \begin{centering}
148:  \epsfig{figure=plots/prlPlotsPlot-3a.eps ,height=7cm}
149:  \epsfig{figure=plots/prlPlotsPlot-3b.eps ,height=7cm}
150: \caption{ $\chi^2$ fit to the \mX\ distribution: a)
151:      data (points) and fit components, and b) 
152:      data and signal MC after subtraction of the $b\to c\ell\nu$ and the ``other'' backgrounds.  
153: \label{mxfit}}
154: \end{centering}
155: \end{figure}
156: 
157: 
158: The ratio between the branching fractions for the signal and $\Bxlnu$ decays is 
159:  \begin{equation}
160:  R_{u/sl} = 
161:  \frac{\BR({\rm { signal)}}}{\BR(\Bxlnu)}=
162:  \frac{N_u^{true}}{N_{sl}^{true}} = 
163:  \frac{(N_u^{meas}- BG_u)/(\epsilon_{sel}^u)}{(N_{sl}^{meas}-BG_{sl})} 
164:  \times \frac{\epsilon_l^{sl} \epsilon_t^{sl} } {\epsilon_l^u \epsilon_t^u }.
165:  \label{eq:ratioBR}
166:  \end{equation}
167: 
168: 
169: 
170: \noindent
171: The efficiency ratio is expected to be close to, but not equal to
172: unity.  Due to the difference in multiplicity and the different
173: lepton momentum spectra, we expect the tag efficiency $\epsilon_t$
174: and lepton efficiency $\epsilon_l$ to be slightly different for the
175: two classes of events.  
176: The signal branching fraction is then obtained from $R_{u/sl}$ using  
177: the \Bxclnu\ branching ratio measured in \babar, 
178: \begin{equation}
179: \label{eq:semilep}
180: \BR(\Bxclnu)= (10.61 \pm 0.16 (\rm exp.) \pm 0.06 (\rm theo.))\% ~\cite{Aubert:2004aw}.
181: \end{equation}
182: which, given the previous \babar\ measurement for the 
183: charmless semileptonic branching ratio \cite{Aubert:2003zw}, corresponds to a total 
184: semileptonic branching fraction of $\BR(\Bxlnu)=(10.83\pm 0.19)\%$.
185: 
186: \subsection{Systematic Uncertainties}
187: \label{sec:sys}
188: Most of the systematic uncertainties are common to all analyses.
189: Uncertainties related to the reconstruction of charged tracks
190: are  determined  by  removing   randomly  a  fraction  of  tracks
191: corresponding  to  the uncertainty  in  the  track finding  efficiency
192: ($1.3\%$ for tracks  with transverse momentum
193: $p_\perp > 0.2\gevc$ and  $2.5\%$ for tracks
194: with  $p_\perp  <  0.2\gevc$).   The  systematic  error  due  to  the
195: reconstruction of neutral  particles in the EMC is studied by varying the 
196: resolution and efficiency to match
197: those found in control samples in data.
198: 
199: We  estimate the systematic  error due  
200: to particle  identification by
201: varying the  electron and kaon identification efficiencies  by $\pm 2\%$
202: and   the   muon   identification   efficiency   by   $\pm3\%$.    The
203: misidentification probabilities are varied  by $15\%$ for all particles. 
204: 
205: The uncertainty of the $B_{reco}$ combinatorial background subtraction
206: is estimated by changing the signal shape function to a Gaussian function instead 
207: of the empirical function of Ref.~\cite{cry}. 
208: We evaluated the effect of cross-feed between \Bz\ and \Bp\ decays 
209: by repeating the analysis with only the \Bz\Bzb or the \Bp\Bm Monte Carlo. 
210: 
211: The impact of the charm semileptonic branching fractions has been estimated by varying each of
212: the exclusive  branching fractions within  one standard deviation  of the
213: current world average~\cite{Hagiwara:2002fs}.  Similarly, the branching fractions
214: of charm mesons have been varied, both for exclusive decays and
215: for inclusive kaon production. To study the cross-feed among the 
216: charmless modes we also varied the number of 
217: charmless exclusive semileptonic decays by $30\%$ for $\Bbar \ra \pi \ell \bar{\nu}$
218: and $\Bbar \ra \rho \ell \bar{\nu}$, by $40\%$ for $\Bbar \ra \omega \ell \bar{\nu}$ and by 
219: $100\%$ for the remaining exclusive charmless semileptonic B decays.
220: 
221: In presenting the results the  systematic effects are divided into the following
222: categories: uncertainties related to detector and reconstruction simulation
223: (tracking reconstruction, neutral reconstruction, lepton identification,
224: charged Kaon identification), $\sigma_{det}$; uncertainties related to the theoretical model 
225: of the non-resonant \btoulnu\ decays, $\sigma_{theo}$;
226: uncertainties related to other aspects of the \btoulnu\  simulation ( 
227: resonant signal decay branching fractions, 
228: $s\bs$ production, hadronization, etc.), $\sigma_{ul\nu}$; 
229: uncertainties due to the limited available Monte Carlo  statistics ($\sigma_{MCstat}$);
230: uncertainties related to background simulation and subtraction ($\Bbar \ra
231: D^{(*,**)} \ell \bar{\nu}$ and $D$ branching fractions, fit to the \mx\ distribution), $\sigma_{bkg}$; 
232: uncertainties related to the subtraction of the combinatorial background and cross-feed,  
233: $\sigma_{breco}$. 
234: