hep-ex0604037/prd.tex
1: \documentclass[twocolumn,showpacs,aps,prd]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{epsf}  
4: 
5: \newcommand{\BaBarYear}    {06}
6: \newcommand{\BaBarNumber}  {014}
7: \newcommand{\SLACPubNumber} {11821}
8: \newcommand{\LANLNumber} {XXXXXXX}
9: 
10: \newcommand{\BaBarType}      {PUB}  
11: \input pubboard/babarsym
12: 
13: \input DDsymbols
14: 
15: \setlength{\textwidth}{17.9cm}
16: \setlength{\textheight}{23.9cm}
17: 
18: \long\def\inst#1{\par\nobreak\kern 4pt\nobreak
19:     {\it #1}\par\vskip 10pt plus 3pt minus 3pt}
20: 
21: \begin{document}
22: 
23: \preprint{\babar-PUB-\BaBarYear/\BaBarNumber} 
24: \preprint{SLAC-PUB-\SLACPubNumber} 
25: 
26: \begin{flushleft}
27: \babar-PUB-\BaBarYear/\BaBarNumber \\
28: SLAC-PUB-\SLACPubNumber \\
29: \end{flushleft}
30: 
31: 
32: \title{ 
33: \Large \bf\boldmath 
34: Measurement of Branching Fractions and \CP-Violating Charge Asymmetries
35: for \B\ Meson Decays to $D^{(*)}\Dbar^{(*)}$, and Implications for the CKM Angle $\gamma$ 
36: }
37: 
38: \input pubboard/authors_feb2006.tex
39: \begin{abstract}
40: We present measurements of the branching fractions and charge
41: asymmetries of \B\ decays to all $D^{(*)}\Dbar^{(*)}$ modes.
42: Using 232 million \BB\ pairs recorded on the $\Upsilon (4S)$ resonance
43: by the \babar\ detector at the \epem asymmetric $B$ factory \pep2 at the Stanford Linear Accelerator Center,
44: we measure the branching fractions 
45: \begin{eqnarray}
46: \mathcal{B}(\Bz \to D^{*+}D^{*-}) & = & \BRbztodstdstNum, \nonumber\\
47: \mathcal{B}(\Bz \to D^{*\pm}D^{\mp}) & = & \BRbztodstdNum, \nonumber\\
48: \mathcal{B}(\Bz \to D^{+}D^{-}) & = & \BRbztoddNum, \nonumber\\
49: \mathcal{B}(\Bp \to D^{*+}\Dbar^{*0}) & = & \BRbptodstdstzNum, \nonumber\\
50: \mathcal{B}(\Bp \to D^{*+}\Dbar^{0}) & = & \BRbptodstdzNum, \nonumber\\
51: \mathcal{B}(\Bp \to D^{+}\Dbar^{*0}) & = & \BRbptoddstzNum, \nonumber\\
52: \mbox{and}\; \mathcal{B}(\Bp \to D^{+}\Dbar^{0}) & = & \BRbptoddzNum, \nonumber
53: \end{eqnarray}
54: where in each case the first uncertainty is statistical and the second systematic.  We also determine the limits
55: \begin{eqnarray}
56: \mathcal{B}(\Bz \to D^{*0}\Dbar^{*0}) & < & \LIMbztodstzdstzNum,\nonumber\\ 
57: \mathcal{B}(\Bz \to D^{*0}\Dbar^{0}) & < & \LIMbztodstzdzNum,\nonumber\\ 
58: \mbox{and}\; \mathcal{B}(\Bz \to D^{0}\Dbar^{0}) & < & \LIMbztodzdzNum,\nonumber
59: \end{eqnarray}
60: each at 90\% confidence level.  All decays above denote either member of a charge
61: conjugate pair.  We also determine the \CP-violating charge asymmetries 
62: \begin{eqnarray}
63: \mathcal{A}(\Bz \to D^{*\pm}D^{\mp}) & = & \ACPbztodstdNum,\nonumber\\
64: \mathcal{A}(\Bp \to D^{*+}\Dbar^{*0}) & = & \ACPbptodstdstzNum, \nonumber\\
65: \mathcal{A}(\Bp \to D^{*+}\Dbar^{0}) & = & \ACPbptodstdzNum,\nonumber\\
66: \mathcal{A}(\Bp \to D^{+}\Dbar^{*0}) & = & \ACPbptoddstzNum,\nonumber\\ 
67: \mbox{and}\; \mathcal{A}(\Bp \to D^{+}\Dbar^{0}) & = & \ACPbptoddzNum.\nonumber
68: \end{eqnarray}
69: Additionally, when we combine these results with information from time-dependent \CP asymmetries in
70: $\Bz \to D^{(*)+}D^{(*)-}$ decays and
71: world-averaged branching fractions of $B$ decays to $D_s^{(*)}\Dbar^{(*)}$ modes,
72: we find the CKM phase $\gamma$ is favored to lie in the range $[0.07-2.77]$ radians (with a +0
73: or $+\pi$ radians ambiguity) at 68\% confidence level.
74: \end{abstract}
75: 
76: \pacs{13.25.Hw, 12.15.Hh, 11.30.Er}
77: 
78: \maketitle
79: 
80: \newpage
81: 
82: 
83: \setcounter{footnote}{0}
84: 
85: \section{Introduction}\label{sec:intro}
86: 
87: We report on measurements of branching fractions of neutral and charged $B$-meson decays to the ten double-charm
88: final states $D^{(*)}\Dbar^{(*)}$.  For the four charged $B$ decays to $D^{(*)}\Dbar^{(*)}$ and for neutral
89: $B$ decays to $\Dstarpm\Dmp$, we also measure the direct 
90: \CP-violating time-integrated charge asymmetry
91: \begin{equation}
92: \mathcal{A}_{\CP} \equiv \frac{\Gamma^{-} - \Gamma^{+}}{\Gamma^{-} + \Gamma^{+}},
93: \label{eq:acp}
94: \end{equation}
95: where in the case of the charged $B$ decays, the superscript on $\Gamma$ corresponds to the sign of the \Bpm meson, and
96: for $\Dstarpm\Dmp$, $\Gamma^{+}$ refers to $\Dstarm\Dp$ and $\Gamma^{-}$ to $\Dstarp\Dm$.
97: 
98: In the neutral $B \to D^{(*)+}D^{(*)-}$ decays, the interference of the dominant 
99: tree diagram (see Fig.~\ref{fig:bddFeynman}a) with the neutral $B$ mixing diagram is sensitive to the Cabibbo-Kobayashi-Maskawa (CKM) phase 
100: $\beta \equiv \arg \left[\, -V_{\rm cd}^{}V_{\rm cb}^* / V_{\rm td}^{}V_{\rm tb}^*\, \right]$, where $V$ is the CKM quark mixing matrix~\cite{CKM}.
101: However, the theoretically uncertain contributions of penguin diagrams (Fig.~\ref{fig:bddFeynman}b) with different weak phases are potentially
102: significant and may shift both the observed \CP asymmetries and the branching fractions by amounts that depend on the ratios of the penguin to 
103: tree contributions and their relative phases.  A number of theoretical estimates exist for the resulting values of the branching fractions and \CP
104: asymmetries~\cite{gronauetc,rosner,sandaxing,phamxing,xing2}.
105: 
106: 
107: The penguin-tree interference in neutral and charged $B \to D^{(*)} \Dbar^{(*)}$ decays can provide sensitivity to the angle
108: $\gamma = \arg \left[\, -V_{\rm ud}^{}V_{\rm ub}^* / V_{\rm cd}^{}V_{\rm cb}^*\, \right]$~\cite{DL,ADL}.  With additional information on the branching fractions of
109: $B \to D^{(*)}_s \Dbar^{(*)}$ decays,
110: the weak phase may be extracted, assuming SU(3) flavor symmetry between $B \to D^{(*)} \Dbar^{(*)}$ and $B \to D^{(*)}_s \Dbar^{(*)}$.
111: For this analysis, we assume that the breaking of SU(3) can be parametrized via the ratios of decay
112: constants $f_{D_s^{(*)}}/f_{D^{(*)}}$, which are quantities that can be
113: determined either with lattice QCD or from experimental measurements~\cite{lattice}.
114: 
115: \begin{figure}[h]  
116: \includegraphics{TreeDiagNew2.epsi}
117: 
118: \vspace*{0.9cm}
119: 
120: \scalebox{1.05}{\includegraphics{PenguinDiagNew2.epsi}}
121: 
122: \vspace*{0.9cm}
123: 
124: \scalebox{1.05}{\includegraphics{ExchangeDiag2.epsi}}
125: 
126: \vspace*{0.9cm}
127: 
128: \scalebox{1.05}{\includegraphics{AnnihilationDiag2.epsi}}
129: \caption{
130: \label{fig:bddFeynman}
131: Feynman graphs for $B \to D^{(*)} \Dbar^{(*)}$ decays:
132: the tree (a) and penguin (b) diagrams are the leading terms for both \Bz $\to D^{(*)+}D^{(*)-}$ and \Bp $\to D^{(*)+}\Dbar^{(*)0}$ 
133: decays, whereas the exchange (c) and annihilation (d) diagrams (the latter of which is OZI-suppressed) are the lowest-order terms 
134: for \Bz $\to D^{(*)0}\Dbar^{(*)0}$ decays.}
135: \end{figure}
136: 
137: In addition to presenting measurements of the \Bz $\to D^{(*)+}D^{(*)-}$ and \Bp $\to D^{(*)+} \Dbar^{(*)0}$ branching fractions, and
138: the \CP-violating charge asymmetries for the latter modes and for \Bz $\to D^{*\pm}D^{\mp}$,
139: we search for the color-suppressed decay modes \Bz $\to D^{(*)0}\Dbar^{(*)0}$,
140: which have not been previously measured, and determine limits on those branching fractions~\cite{CC}.  If observed, the decays \Bz $\to D^{(*)0}\Dbar^{(*)0}$
141: would provide evidence of $W$-exchange or annihilation contributions (see Fig.~\ref{fig:bddFeynman}c,\ref{fig:bddFeynman}d).
142: In principle, these decays could also provide sensitivity to the CKM phase $\beta$ if sufficient data were 
143: available.
144: By combining all of these results with information from time-dependent \CP asymmetries in
145: $\Bz \to D^{(*)+}D^{(*)-}$ decays and
146: world-averaged branching fractions of $B$ decays to $D_s^{(*)}\Dbar^{(*)}$ modes, we determine the implications for $\gamma$ using the method of Refs.~\cite{DL,ADL}.
147: 
148: \section{Detector and Data} \label{sec:detector}
149: 
150: The results presented in this paper are based on data collected
151: with the \babar\ detector~\cite{BABARNIM}
152: at the PEP-II asymmetric-energy $e^+e^-$ collider~\cite{pep}
153: located at the Stanford Linear Accelerator Center. 
154: The integrated luminosity
155: is 210.5~\invfb, corresponding to 231.7 million \BB\ pairs, 
156: recorded at the $\Upsilon (4S)$ resonance
157: (``on-peak'', at a center-of-mass (c.m.) energy $\sqrt{s}=10.58\ \gev$).
158: 
159: The asymmetric beam configuration in the laboratory frame
160: provides a boost of $\beta\gamma = 0.56$ to the $\Upsilon(4S)$.  
161: Charged particles are detected and their momenta measured by the
162: combination of a silicon vertex tracker (SVT), consisting of five layers of double-sided detectors,
163: and a 40-layer central drift chamber (DCH),
164: both operating in the 1.5-T magnetic field of a solenoid.  
165: For tracks with transverse momentum greater than 120~\mevc, the DCH provides the
166: primary charged track finding capability.
167: The SVT provides complementary standalone track finding for tracks of lower momentum, allowing for
168: reconstruction of charged tracks with transverse momentum $p_T$
169: as low as 60 \mevc, with efficiencies in excess of 85\%.  
170: This ability to reconstruct tracks with low $p_T$ efficiently is  
171: necessary for reconstruction of the slow charged pions from
172: \Dstarp $\to \Dz\pip$ decays in $B \to D^{(*)} \Dbar^{(*)}$ signal events. 
173: The transverse momentum resolution for the combined tracking system is 
174: $\sigma_{p_T}/p_T=0.0013p_T + 0.0045$, where 
175: $p_T$ is measured in \gevc.  
176: Photons are detected and their energies measured by a CsI(Tl) electromagnetic
177: calorimeter (EMC).  The photon energy resolution is 
178: $\sigma_{E}/E = \left\{2.3 / E(\gev)^{1/4} \oplus 1.4 \right\}\%$, 
179: and their angular resolution with respect to the interaction point is
180: $\sigma_{\theta} = \mbox{(4.2 mrad)}/\sqrt{E(\gev)}$. 
181: The measured $\pi^0$ mass resolution for $\piz$'s with
182: laboratory momentum in excess of 1 \gevc\ is approximately 6 \mevcc.
183: 
184: Charged-particle identification (PID) is provided by 
185: an internally reflecting ring-imaging
186: Cherenkov light detector (DIRC) covering the central region,
187: and the most probable energy loss (\dedx) in the tracking devices.
188: The Cherenkov angle resolution of the DIRC is measured to be 2.4~mrad, which
189: provides over $5\sigma$ separation between charged kaons and pions at
190: momenta of less than $2~\gevc$.  The $\dedx$ resolution 
191: from the drift chamber is typically about $7.5\%$ for pions.  
192: Additional information
193: to identify and reject electrons and muons is provided
194: by the EMC and detectors embedded between the steel plates of
195: the magnetic flux return (IFR).
196: 
197: \section{Candidate Reconstruction and {\boldmath $B$} Meson Selection}
198: \label{sec:eventsel}
199: 
200: Given the high multiplicity of the final states studied, very high
201: combinatorial background levels are expected. Selection criteria (described
202: in Sec.~III A--E)  are designed to minimize the expected statistical error on the 
203: $B$ branching fractions (as described in Sec.~III F).
204: A {\tt GEANT4}-based~\cite{GEANT4} Monte Carlo (MC) simulation
205: of the material composition and the instrumentation response of the \babar\ detector is
206: used to optimize signal selection criteria and evaluate signal detection efficiency.
207: We retain sufficient sidebands in the discriminating
208: variables to characterize the background in subsequent fits.
209: 
210: \subsection{Charged track and {\boldmath \KS} selection}
211: \label{sec:track_sele}
212: 
213: Charged particle tracks are selected via pattern recognition algorithms
214: using measurements from the SVT and DCH detectors.
215: We additionally require all charged-particle tracks (except for those from 
216: $\KS\ra\pip\pim$ decays) to
217: originate within 10 cm along the beam axis and 1.5 cm in the plane perpendicular to the beam axis of the center of the
218: beam crossing region.  To ensure a well-measured momentum, all charged-particle tracks except
219: those from $\KS\ra\pip\pim$ decays and \pip from $\Dstarp \to \Dz\pip$ decays must also
220: be reconstructed from at least 12 measurements in the DCH.
221: All tracks that meet these criteria are considered as charged pion candidates.
222: 
223: Tracks may be identified as kaons based on a likelihood selection developed from 
224: Cherenkov angle and \dedx information from the DIRC and tracking detectors respectively.
225: For the typical laboratory momentum spectrum of the signal kaons, this  
226: selection has an efficiency of about 85\% and a purity of 
227: greater than 98\%, as determined from control samples of $\Dstarp \to \Dz\pip$,
228: $D^0\to K^-\pip$ decays.  
229: 
230: We require $\KS \to \pip\pim$ candidates to have an invariant mass within 15 \mevcc of the nominal \KS
231: mass~\cite{PDG2004}.  The probability that the two daughter tracks originate from the same 
232: point in space must be greater than 0.1\%.  The
233: transverse flight distance of the \KS from the primary event vertex 
234: must be both greater than $3\sigma$ from zero (where $\sigma$ is the measured uncertainty on the transverse flight length) and
235: also greater than 2 mm.
236: 
237: \subsection{Photon and {\boldmath \piz} selection}
238: \label{sec:gampiz_sele}
239: 
240: Photons are reconstructed from energy deposits in the electromagnetic
241: calorimeter which are not associated with a charged track.  To reject backgrounds
242: from electronics noise, machine background, and hadronic interactions in the EMC, we require that all photon
243: candidates have an energy greater than 30 \mev in the laboratory frame and to have a 
244: lateral shower shape consistent with that of a photon.
245: Neutral pions are reconstructed from pairs of photon candidates whose energies in the laboratory frame sum to more than 200 \mev.
246: The \piz candidates must have an invariant mass between 115 and 150~\mevcc.  The 
247: \piz candidates that meet these criteria, when combined with other tracks or neutrals to form $B$ candidates, 
248: are then constrained to originate from their expected decay points,
249: and their masses are constrained to the nominal value~\cite{PDG2004}.
250: This procedure improves the mass and energy resolution of the parent
251: particles. 
252: 
253: 
254: \begin{figure*}[t]
255: \begin{center}
256: \scalebox{0.85}{\includegraphics{data_Deltam.eps}}
257: \end{center}
258: \vspace*{-0.6cm}
259: \caption{
260: \label{fig:deltam}
261: Distributions of $\Delta m$ in the full data sample for three \Dstar decay modes.
262: Plot a) shows $\Delta m(\Dstarp - \Dz)$ for $\Dstarp \to \Dz\pip$ decays where
263: \Dz decays to $\Km\pip$.  Plot b) shows $\Delta m(\Dstarp - \Dp)$ for
264: $\Dstarp \to \Dp\piz$ decays where \Dp decays to $\Km\pip\pip$.  Plot c)
265: shows $\Delta m(\Dstarz - \Dz)$ for $\Dstarz \to \Dz\piz$ decays where \Dz decays
266: to $\Km\pip$.  
267: Nominal values for $\Delta m$ are 145.4 \mevcc, 140.6 \mevcc, and 
268: 142.1 \mevcc for the three cases respectively~\cite{PDG2004}.
269: }
270: \end{figure*}
271: 
272: \subsection{Event selection}
273: \label{sec:event_sele}
274: 
275: We select \BB events by applying criteria on the
276: track multiplicity and event topology. 
277: At least three reconstructed
278: tracks, each with transverse momentum greater than 100 \mevc, are required in the
279: laboratory polar angle region $0.41 < \theta_{\rm lab} < 2.54$.  
280: The event must have a total measured energy in the laboratory frame greater than 4.5~\gev to reject beam-related background.  
281: The ratio of Fox-Wolfram moments $H_2/H_0$~\cite{FoxW} is a parameter between 0 (for ``perfectly spherical'' events) and 1 (for ``perfectly jet-like'' events), 
282: and we require this ratio to be less than 0.6 for each event, in order to help reject non-\BB background.
283: This criterion rejects between 30 and 50 percent of non-\BB background (depending on the decay mode), while keeping almost all of the signal decays.
284: 
285: \subsection{{\boldmath $D$} and {\boldmath \Dstar} meson selection}
286: \label{sec:dmeson_sele}
287: 
288: We reconstruct \Dz mesons in the four decay modes $\Dz \to \Km\pi^{+}$, $\Dz \to \Km\pip\piz$, $\Dz \to \Km\pip\pim\pip$,
289: and $\Dz \to \KS\pip\pim$, and \Dp mesons in the two decay modes $\Dp \to \Km\pip\pip$ and $\Dp \to \KS\pip$.
290: We require \Dz and \Dp candidates to have 
291: reconstructed masses 
292: within $\pm20$ \mevcc of their nominal masses~\cite{PDG2004}, except for 
293: $\Dz \to \Km\pip\piz$, for which we require $\pm40$ \mevcc due to the poorer resolution for modes containing
294: \piz's.  
295: These criteria correspond to approximately $2.5 \sigma$ of the respective mass resolutions.
296: The $\Dz \to \Km\pip\piz$ decays must also satisfy a criterion on the reconstructed invariant
297: masses of the $\Km\pip$ and $\Km\piz$ pairs: the combination of reconstructed invariant masses must lie at a point 
298: in the $\Km\pip\piz$ Dalitz plot~\cite{Dalitz} for  
299: which the expected density normalized to the maximum density 
300: (``Dalitz weight'') is at least 6\%.
301: Additionally, the daughters of \Dz and \Dp candidates 
302: must have a probability of originating from a common point in space greater than 0.1\%, and are then
303: constrained both to originate from that common spatial point and to have their respective nominal invariant masses.
304: 
305: \begin{figure*}[t]
306: \begin{center}
307: \includegraphics[width=0.8\textwidth]{cut_var_MC.eps}
308: \end{center}
309: \vspace*{-0.6cm}  
310: \caption{
311: \label{fig:selvars}
312: Distributions of signal selection variables: a) the likelihood variable $-\ln(\masslik)$,
313: b) the $\Delta E$ variable,
314: c) the Fisher discriminant $\mathcal{F}$, and
315: d) the $D$-meson flight length variable $L$,
316: each for the representative signal mode \Bz $\to \Dz\Dzb$, and
317: for the corresponding combinatorial background from \BzBzb, \BpBm, \ccbar, and (\uubar+\ddbar+\ssbar) 
318: MC simulated decays respectively.  In each plot, the component distributions are normalized to have the same area below the curves.
319: }
320: \end{figure*}
321: 
322: Candidate \Dstarp and \Dstarz mesons are reconstructed in the decay modes $\Dstarp \to \Dz\pip$, $\Dstarp \to \Dp\piz$,
323: $\Dstarz \to \Dz\piz$, and $\Dstarz \to \Dz\gamma$, using pairs of selected \Dz, \Dp, \piz, \pip, and $\gamma$ candidates.
324: The \pip from $\Dstarp \to \Dz\pip$ decays is additionally required to have a c.m.~momentum of less than 450 \mevc.  
325: Candidate \piz mesons from $\Dstarp \to \Dp\piz$ and $\Dstarz \to \Dz\piz$ are required to have c.m.~momenta 
326: $p^{*}$ in the range $70 < p^{*} < 450$~\mevc.  Photons from $\Dstarz \to \Dz\gamma$ decays are required
327: to have energies in the laboratory frame greater than 100 \mev and c.m.~energies less than 450 \mev.  The \Dstar
328: daughter particles are constrained to originate from a common point in space.  After this constraint is applied, the mass differences
329: $\Delta m$ of the reconstructed masses of the \Dstar and $D$ candidates are required to be within 
330: the ranges shown in Table~\ref{tab:deltaMcuts}.
331: \begin{table}[h]
332: \caption{\label{tab:deltaMcuts}
333: Allowed $\Delta m$(\Dstar - $D$) ranges for the four \Dstar decay modes.
334: }
335: \begin{center}
336: \begin{tabular}{lcc}
337: \hline \hline
338:                         & Minimum             & Maximum             \\
339: Mode                    & $\Delta m$ (\mevcc) & $\Delta m$ (\mevcc) \\
340: \hline
341: $\Dstarp \to \Dz\pip$   &  $139.6$                    &  $151.3$                    \\
342: $\Dstarp \to \Dp\piz$   &  $135.0$                    &  $146.3$                    \\
343: $\Dstarz \to \Dz\piz$   &  $135.0$                    &  $149.3$                    \\
344: $\Dstarz \to \Dz\gamma$ &  $100.0$                    &  $170.0$                    \\
345: \hline \hline
346: \end{tabular}
347: \end{center}
348: \end{table}
349: As shown in Fig.~\ref{fig:deltam}, the excellent resolution in $\Delta m$ for signal candidates makes the
350: $\Delta m$ requirement a very powerful 
351: criterion to reject background (see next section), especially for decay modes containing a $\Dstarp \to \Dz\pip$.
352: 
353: \subsection{Variables used for {\boldmath $B$} meson selection}
354: \label{sec:bmesonvar_sele}
355: 
356: A $B$-meson candidate is constructed by combining two $D^{(*)}$ candidates that have both passed the selection
357: criteria described previously.  The pairs of $D^{(*)}$ candidates are constrained to originate from the
358: same point in space.  We form a likelihood variable, $\masslik$, that is defined by a product
359: of Gaussian distributions for each $D$ mass and $\Dstar-D$ mass difference.
360: \begin{table*}[t]
361: \caption{\label{tab:globalcuts}Expected values of the branching fractions $\mathcal{B}$ for each $B \to D^{(*)} \Dbar^{(*)}$ decay mode, which are used for the purpose of determining selection
362:                                criteria that minimize the expected uncertainty on the measured branching fraction for each mode; also, optimized 
363:                                $\mathcal{F}$ and $L$ selection criteria for each mode.  
364:                                An ``---'' indicates no cut is made in $\mathcal{F}$ or $L$ for that 
365:                                decay mode.}
366: \begin{center}
367: \scalebox{1.0}{
368: \begin{tabular}{lccc}
369: \hline \hline
370: Mode                         &  Expected $\mathcal{B}$ & $\mathcal{F}_{\rm min}$ &  $L_{\rm min}$   \\
371: \hline 
372: \Bz $\to D^{*+}D^{*-}$       &  $8.3 \times 10^{-4}$   &    ---        &  ---   \\
373: \Bz $\to D^{*\pm}D^{\mp}$    &  $8.8 \times 10^{-4}$   &    ---        &  ---   \\
374: \Bz $\to D^{+}D^{-}$         &  $3.0 \times 10^{-4}$   &      0.62     &  $1.3$  \\
375: \Bz $\to D^{*0}\Dstarzb$ &  $1.0 \times 10^{-5}$   &      0.60     &  $-1.6$  \\
376: \Bz $\to D^{*0}\Dzb$  &  $1.0 \times 10^{-5}$   &      0.53     &  $-0.4$  \\
377: \hline
378: \end{tabular}
379: }
380: \hspace{1cm}
381: \scalebox{1.0}{
382: \begin{tabular}{llccc}
383: \hline \hline
384: Mode                         &  Expected $\mathcal{B}$ & $\mathcal{F}_{\rm min}$ &  $L_{\rm min}$   \\
385: \hline 
386: \Bz $\to D^{0}\Dzb$   &  $1.0 \times 10^{-5}$   &      0.47     &  $-0.4$  \\
387: \Bp $\to D^{*+}\Dstarzb$ &  $1.0 \times 10^{-3}$  &      0.60     &  ---   \\
388: \Bp $\to D^{*+}\Dzb$  &  $4.4 \times 10^{-4}$   &      0.53     &  $-1.3$  \\
389: \Bp $\to D^{+}\Dstarzb$  &  $4.4 \times 10^{-4}$   &      0.53     &  $ 0.0$  \\
390: \Bp $\to D^{+}\Dzb$   &  $3.0 \times 10^{-4}$   &      0.53     &  $ 0.5$  \\
391: \hline
392: \end{tabular}
393: }
394: \end{center}
395: \end{table*}
396: 
397: For example, in the decay $\Bz \to \Dstarp\Dstarm$, $\masslik$ is the product of four terms:
398: Gaussian distributions for each $D$ mass and double Gaussian (\textit{i.e.}~the sum of two Gaussian distributions) terms
399: for each $\Delta m$ term (the $\Dstar - D$ mass difference).  Defining $G(x;\mu,\sigma)$ as a normalized
400: Gaussian distribution where $x$ is the independent variable, $\mu$ is the mean, and $\sigma$ is the resolution,
401: $\masslik$ for $\Bz \to \Dstarp\Dstarm$ decays is defined as:
402: \begin{widetext}
403: \begin{eqnarray}
404: \label{eq:chi2likedstd}
405: \masslik & = & G(m_D;m_{D_{\rm PDG}},\sigma_{m_D}) \times
406:                       G(m_{\Dbar};m_{\Dbar_{\rm PDG}},\sigma_{m_{\Dbar}}) \times
407:                   \nonumber \\
408: &  & \qquad \left[ f_{\rm core} G(\Delta m_{\Dstarp};\Delta m_{\Dstarp_{\rm PDG}},\sigma_{\Delta m_{\rm core}})
409:                      + (1 - f_{\rm core}) G(\Delta m_{\Dstarp};\Delta m_{\Dstarp_{\rm PDG}},\sigma_{\Delta m_{\rm tail}})  \right]
410:            \times  \nonumber \\
411: &  & \qquad \left[ f_{\rm core} G(\Delta m_{\Dstarm};\Delta m_{\Dstarm_{\rm PDG}},\sigma_{\Delta m_{\rm core}})
412:                      + (1 - f_{\rm core}) G(\Delta m_{\Dstarm};\Delta m_{\Dstarm_{\rm PDG}},\sigma_{\Delta m_{\rm tail}})  \right],
413: \end{eqnarray}
414: \end{widetext}
415: where the subscript ``PDG'' refers to the nominal value~\cite{PDG2004}, and all reconstructed masses and uncertainties are determined before mass constraints are applied.  
416: For $\sigma_{m_D}$, we use errors calculated 
417: candidate-by-candidate.  The parameter $f_{\rm core}$ is the ratio of the area of the core Gaussian to the total area of the double
418: Gaussian distribution.  This, along with $\sigma_{\Delta m_{\rm core}}$ and $\sigma_{\Delta m_{\rm tail}}$, is determined separately 
419: for each of the four \Dstar decay modes given above, using MC simulation of signal events that is calibrated to inclusive samples
420: of the \Dstar decay modes in data.  
421: For each of the $B$ decay modes, a higher value of \masslik tends to indicate a greater signal likelihood.  
422: The distributions of $-\ln(\masslik)$ for the representative signal mode \Bz $\to \Dz\Dzb$ and
423: for the corresponding combinatorial background from generic \BzBzb, \BpBm, \ccbar, and (\uubar+\ddbar+\ssbar) decays, are shown in 
424: Fig.~\ref{fig:selvars}a.
425: We use  
426: \masslik in selecting signal candidates, as will be described in the upcoming section.
427: 
428: We also use the two variables for fully-reconstructed $B$ meson selection at the \Y4S energy: the beam-energy-substituted mass 
429: \mes $\equiv [(s/2 + \vec{p}_i \cdot \vec{p}_B)^2/E_i^2 - \vec{p}_B^2]^{1/2}$, where the initial total \epem four-momentum $(E_i,\vec{p}_i)$
430: and the $B$ momentum $\vec{p}_B$ are defined in the laboratory frame; and 
431: $\Delta E \equiv E_B^{\rm cm} - \sqrt{s}/2$ is
432: the difference between the reconstructed $B$ energy in the c.m.~frame
433: and its known value.
434: The normalized distribution of $\Delta E$ for the representative signal mode \Bz $\to \Dz\Dzb$, and
435: for the corresponding combinatorial background components, 
436: is shown in Fig.~\ref{fig:selvars}b.
437: 
438: 
439: In addition to \masslik, \mes, and $\Delta E$, a Fisher discriminant $\mathcal{F}$~\cite{cleofisher} and a $D$-meson flight length variable $L$ are used
440: to help separate signal from background.  The Fisher discriminant assists in the suppression of background from continuum events by incorporating 
441: information from the topology of the event.  The discriminant is formed from the momentum flow into nine polar angular intervals of $10^\circ$ centered on the
442: thrust axis of the $B$ candidate, the angle of the event thrust axis 
443: with respect to the beam 
444: axis ($\theta_{T}$),
445: and the angle of the $B$ candidate momentum with respect to the beam axis ($\theta_B$):
446: \begin{equation}
447: \mathcal{F} \equiv \sum_{i=1}^{11} \alpha_i x_i.
448: \end{equation}
449: The values $x_i~(i = 1, ..., 9)$ are the scalar sums of the momenta of all charged
450: tracks and neutral showers in the polar angle interval $i$, $x_{10}$ is
451: $|\text{cos}\theta_{T}|$, and $x_{11}$ is $|\text{cos}\theta_B|$.
452: The coefficients $\alpha_i$ are determined from MC simulation to maximize
453: the separation between signal and background~\cite{cleofisher}.
454: The normalized distribution of $\mathcal{F}$ for the representative signal mode \Bz $\to \Dz\Dzb$, and
455: for the corresponding background components,
456: is shown in Fig.~\ref{fig:selvars}c.
457: 
458: The flight length variable $L$ that we consider is defined as 
459: $(\ell_1 + \ell_2)/\sqrt{\sigma_1^2 + \sigma_2^2}$, with the decay lengths $\ell_i$ of the two $D$ mesons defined as
460: \begin{equation}
461: {\vec x}_{D_i} = {\vec x}_B + (\ell_i \times {\vec p}_{D_i}) 
462: \end{equation}
463: where $\vec{x}_D$ and $\vec{x}_B$ are the measured decay vertices of the $D$ and $B$, respectively, and $\vec{p}_D$ is
464: the momentum of a $D$.
465: The $\sigma_i$ are the measured uncertainties on $\ell_i$.
466: This observable exploits the ability
467: to distinguish the long $D$ lifetime.  Thus, background events
468: have an $L$ distribution centered around zero, while
469: events with real $D$ mesons have a distribution favoring
470: positive values.
471: The normalized distribution of $L$ for the representative signal mode \Bz $\to \Dz\Dzb$, and
472: for the corresponding background components,
473: is shown in Fig.~\ref{fig:selvars}d.
474: 
475: 
476: 
477: \subsection{Analysis optimization and signal selection}
478: \label{sec:analopti_sele}
479: 
480: We combine information from the \masslik, $\Delta E$, $\mathcal{F}$, and $L$ variables to select signal candidates in each decay mode.
481: The fractional statistical uncertainty on a measured branching fraction is proportional to $\sqrt{(N^s + N^b)}/N^s$, where $N^s$ is the 
482: number of reconstructed signal events and $N^b$ is the number of background events within the selected signal region for a mode. 
483: The values $N^s$ and $N^b$ are calculated,
484: using detailed MC simulation of the signal decay modes as well as of \BB and continuum background decays,
485: by observing the number of simulated $B$ decay candidates
486: that satisfy the selection criteria for $-\ln(\masslik)$, $|\Delta E|$, $\mathcal{F}$, and $L$.
487: We choose criteria which minimize the expected $\sqrt{(N^s + N^b)}/N^s$ for each mode.  Note that to calculate the expected number of signal events $N^s$,
488: one must assume an expected branching fraction, as well as the ratios of
489: \BB and continuum events using their relative cross-sections.  These are given, along with the requirements on $\mathcal{F}$ and $L$, in Table~\ref{tab:globalcuts}.  
490: 
491: 
492: For each possible combination of \Dstarp, \Dstarz, \Dp, and \Dz decay modes, we determine the combination of selection criteria on
493: $-\ln(\masslik)$ and $|\Delta E|$ that minimizes the overall expected $\sqrt{(N^s + N^b)}/N^s$ for each $B$ decay mode (see Tables~\ref{tab:modekey}, ~\ref{tab:masslikcuts}, 
494: and~\ref{tab:deltaecuts}).  
495: The selection criteria for $\mathcal{F}$ and $L$ are chosen, however, 
496: only for each $B$ decay mode and not separately for each $D^{(*)}$ mode combination.  The restrictiveness of the kaon identification selection
497: is also optimized separately for each charged and neutral $D^{(*)}$ mode.  
498: 
499: Between 1\% and 34\% of selected $B \to D^{(*)} \Dbar^{(*)}$ events have more than one reconstructed $B$ candidate
500: that passes all selection criteria in
501: \masslik, $\Delta E$, $\mathcal{F}$, and $L$, with the largest percentages
502: occurring in the decay modes \Bz $\to \Dstarz\Dstarzb$ and \Bz $\to \Dz\Dzb$, and the smallest occurring in \Bz $\to \Dstarpm\Dmp$ and \Bz $\to \Dp\Dm$.
503: In such events, we choose the reconstructed $B$ with the largest value of \masslik as the
504: signal $B$ candidate.
505: 
506: 
507: 
508: 
509: \begin{table*}[p]
510: \caption{\label{tab:modekey} Key to mode numbers used in Tables~\ref{tab:masslikcuts} and~\ref{tab:deltaecuts} below.}
511: \begin{center}
512: \scalebox{0.90}{
513: \begin{tabular}{lc}
514: \hline \hline
515: Mode                                    &  \#  \\
516: \hline
517: \Dstarp $\to (K^-\pi^+)\pi^+$           &  1 \\
518: \Dstarp $\to (K^-\pi^+\piz)\pi^+$       &  2 \\
519: \Dstarp $\to (K^-\pi^+\pi^-\pi^+)\pi^+$ &  3 \\
520: \Dstarp $\to (\KS\pi^+\pi^-)\pi^+$      &  4 \\
521: \Dstarp $\to (K^-\pi^+\pi^+)\piz$       &  5 \\
522: \Dstarz $\to (K^-\pi^+)\piz$            &  6 \\
523: \Dstarz $\to (K^-\pi^+\piz)\piz$        &  7 \\
524: \Dstarz $\to (K^-\pi^+\pi^-\pi^+)\piz$  &  8 \\
525: \hline \hline
526: \end{tabular}
527: }
528: \hspace{1cm}
529: \scalebox{0.90}{
530: \begin{tabular}{lc}
531: \hline \hline
532: Mode                                     & \# \\
533: \hline
534: \Dstarz $\to (\KS\pi^+\pi^-)\piz$        &  9 \\
535: \Dstarz $\to (K^-\pi^+)\gamma$           & 10 \\
536: \Dp $\to K^-\pip\pip$                    & 11 \\
537: \Dz $\to \Km\pip$                        & 12 \\
538: \Dz $\to \Km\pip\piz$                    & 13 \\
539: \Dz $\to \Km\pip\pim\pip$                & 14 \\
540: \Dz $\to \KS\pi^+\pi^-$                  & 15 \\
541:                                          &    \\
542: \hline \hline
543: \end{tabular}
544: }
545: \vspace*{-0.3cm}
546: \end{center}
547: \end{table*}
548: \begin{table*}[p]
549: \caption{\label{tab:masslikcuts}Optimized $-\ln(\masslik)$ selection criteria used for all $B \to D^{(*)} \Dbar^{(*)}$ modes.  Selected events in a given mode must have 
550:                                 $-\ln(\masslik)$ less than the given value.  The $D^{(*)}$ decay modes $1-15$ are defined in Table~\ref{tab:modekey} above.
551:                                 Elements with ``---'' above and on the diagonal are modes that are 
552:                                 unused since, due to high backgrounds, they do not help to increase signal sensitivity.}
553: \begin{center}
554: \scalebox{0.83}{
555: \begin{tabular}{ll|ccccc|ccccc|c|cccc}
556: \hline \hline
557:           &                    &  \multicolumn{5}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $D^{*-}$}} & 
558:                                   \multicolumn{5}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $\Dstarzb$}} &
559:                                   \multicolumn{1}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $D^{-}$}}  &
560:                                   \multicolumn{4}{|c}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $\Dzb$}} \\
561:           &                    &   1   &   2   &   3   &   4   &   5   &   6   &   7   &   8   &   9   &  10   &  11   &    12  &   13  &   14   &   15  \\
562: \hline
563:           &          1         & 13.0  & 12.0  & 17.3  & 19.8  & 10.5  & 14.6  & 17.5  & 9.2   &  ---  & 8.9   &  8.2  &   8.6  &  8.5  &  8.2   &  8.0  \\
564:           &          2         &       & 10.6  & 11.0  & 18.3  &  9.5  & 11.5  &  9.8  & 10.7  &  ---  & 8.7   &  8.4  &   7.8  &  ---  &  8.8   & ---   \\
565: {\Large $D^{*+}$}&   3         &       &       & 11.7  & 11.0  &  9.8  & 11.7  &  9.6  & 10.4  &  ---  & 9.0   &  8.8  &   9.3  &  9.4  &  9.0   & ---   \\
566:           &          4         &       &       &       &  ---  &  ---  &  ---  &  ---  &  ---  &  ---  & ---   &  9.6  &  15.1  &  9.2  &  ---   & ---   \\
567:           &          5         &       &       &       &       &  ---  &  8.2  &  ---  &  ---  &  ---  & ---   &  ---  &   6.6  &  ---  &  ---   & ---   \\
568: \hline
569:           &          6         &       &       &       &       &       & 12.2  &  8.4  & 9.6   & 7.6   & ---   &  9.9  &   7.6  &  6.7  &  7.2   & ---   \\
570:           &          7         &       &       &       &       &       &       &  ---  &  ---  &  ---  & ---   &  7.5  &   ---  &  ---  &  ---   & ---   \\
571: {\Large $D^{*0}$}&   8         &       &       &       &       &       &       &       &  ---  &  ---  & ---   &  9.2  &   ---  &  ---  &  ---   & ---   \\
572:           &          9         &       &       &       &       &       &       &       &       &  ---  & ---   &  ---  &   5.8  &  ---  &  ---   & ---   \\
573:                  &  10         &       &       &       &       &       &       &       &       &       & ---   &  ---  &   ---  &  ---  &  ---   & ---   \\
574: \hline
575: {\Large $D^{+}$} &  11         &       &       &       &       &       &       &       &       &       &       &  6.0  &   7.3  &  5.8  &  6.5   &  6.2  \\
576: \hline
577:                  &  12         &       &       &       &       &       &       &       &       &       &       &       &   ---  &  5.2  &  6.8   & ---   \\
578:                  &  13         &       &       &       &       &       &       &       &       &       &       &       &        &  ---  &  6.2   & ---   \\
579: \raisebox{1.5ex}[0pt]{\Large $D^{0}$} &  14         &       &       &       &       &       &       &       &       &       &       &       &        &       &  6.9   & ---   \\
580:                  &  15         &       &       &       &       &       &       &       &       &       &       &       &        &       &        & ---   \\
581: \hline \hline
582: \end{tabular}
583: }
584: \vspace*{-0.3cm}
585: \end{center}
586: \end{table*}
587: \begin{table*}[p]
588: \caption{\label{tab:deltaecuts}Optimized $\Delta E$ selection criteria used for all $B \to D^{(*)} \Dbar^{(*)}$ modes.  Selected events in a given mode must have 
589:                                $|\Delta E|$ (in \mev) less than the given value.  The $D^{(*)}$ decay modes $1-15$ are defined in Table~\ref{tab:modekey} above.
590:                                Elements with ``---'' above and on the diagonal are modes that are unused since, due to high 
591:                                backgrounds, they do not help to increase signal sensitivity.}
592: \begin{center}
593: \scalebox{0.83}{
594: \begin{tabular}{ll|ccccc|ccccc|c|cccc}
595: \hline \hline
596:           &                    &  \multicolumn{5}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $D^{*-}$}} & 
597:                                   \multicolumn{5}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $\Dstarzb$}} &
598:                                   \multicolumn{1}{|c|}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $D^{-}$}}  &
599:                                   \multicolumn{4}{|c}{\rule[0mm]{0mm}{5mm}\raisebox{0.5mm}[0mm]{\Large $\Dzb$}} \\
600:           &                    &   1   &   2   &   3   &   4   &   5   &   6   &   7   &   8   &   9   &  10   &  11   &    12  &   13  &   14   &   15  \\
601: \hline
602:           &          1         & 35.5  & 33.8  & 30.4  & 35.2  & 25.5  & 35.7  & 21.0  & 26.0  &  ---  & 43.6  & 18.0  &  18.1  & 20.2  & 17.1   & 19.0  \\
603:           &          2         &       & 34.5  & 29.6  & 23.5  & 27.4  & 40.9  & 23.9  & 21.4  &  ---  & 29.3  & 19.4  &  25.9  &  ---  & 19.5   &  ---  \\
604: {\Large $D^{*+}$}&   3         &       &       & 23.5  & 23.7  & 18.2  & 34.0  & 30.6  & 20.6  &  ---  & 27.3  & 18.6  &  19.0  & 20.4  & 17.1   &  ---  \\
605:           &          4         &       &       &       &  ---  &  ---  &  ---  &  ---  &  ---  &  ---  &  ---  & 21.9  &  16.9  & 19.7  &  ---   &  ---  \\
606:           &          5         &       &       &       &       &  ---  &  19.1 &  ---  &  ---  &  ---  &  ---  &  ---  &  16.4  &  ---  &  ---   &  ---  \\
607: \hline
608:           &          6         &       &       &       &       &       &  35.1 & 23.0  & 27.3  & 25.5  &  ---  & 23.9  &  17.4  & 19.6  & 17.4   &  ---  \\
609:           &          7         &       &       &       &       &       &       &  ---  &  ---  &  ---  &  ---  & 20.0  &   ---  &  ---  &  ---   &  ---  \\
610: {\Large $D^{*0}$}&   8         &       &       &       &       &       &       &       &  ---  &  ---  &  ---  & 16.6  &   ---  &  ---  &  ---   &  ---  \\
611:           &          9         &       &       &       &       &       &       &       &       &  ---  &  ---  &  ---  &  24.5  &  ---  &  ---   &  ---  \\
612:                  &  10         &       &       &       &       &       &       &       &       &       &  ---  &  ---  &   ---  &  ---  &  ---   &  ---  \\
613: \hline
614: {\Large $D^{+}$} &  11         &       &       &       &       &       &       &       &       &       &       & 15.1  &  15.5  & 19.2  & 15.4   & 15.5  \\
615: \hline
616:                  &  12         &       &       &       &       &       &       &       &       &       &       &       &   ---  & 18.7  & 16.1   &  ---  \\
617:                  &  13         &       &       &       &       &       &       &       &       &       &       &       &        &  ---  & 19.0   &  ---  \\
618: \raisebox{1.5ex}[0pt]{\Large $D^{0}$} &  14         &       &       &       &       &       &       &       &       &       &       &       &        &       & 15.9   &  ---  \\
619:                  &  15         &       &       &       &       &       &       &       &       &       &       &       &        &       &        &  ---  \\
620: \hline \hline
621: \end{tabular}
622: }
623: \end{center}
624: \end{table*}
625: 
626: \begin{table*}[t!]
627: \caption{\label{tab:effs}
628: Elements of the efficiency and crossfeed matrix $\epsilon_{ij}$, and their respective uncertainties, used to calculate the branching fractions and charge asymmetries, as
629: described in the text.  {\bf\boldmath All values are in units of $10^{-4}$.}  
630: Uncertainties on the last digit(s) are given in parentheses.  
631: Elements with ``---'' correspond to values that are zero (to three digits after the decimal point).  
632: The column corresponds to the generated mode and the row corresponds to the reconstructed mode.
633: }
634: \begin{center}
635: \scalebox{1.0}{
636: \begin{tabular}{lcccccccccc}
637: \hline \hline
638: Mode  &$\Dstarp\Dstarm$ &$\Dstarpm\Dmp$  & $\Dp\Dm$   &  $\Dstarz\Dstarzb$   & $\Dstarz\Dzb$   & $\Dz\Dzb$   & $\Dstarp\Dstarzb$   & $\Dstarp\Dzb$  & $\Dp\Dstarzb$   &  $\Dp\Dzb$ \\
639: \hline
640: $\Dstarp\Dstarm$   & 14.24(6) & 0.010(3) &  ---  &  ---  &  ---  &  ---  & 0.18(1) & ---   &  ---  &   --- \\
641: $\Dstarpm\Dmp$     & 0.020(3) & 11.52(6) &  ---  &  ---  &  ---  &  ---  & 0.010(3) & 0.040(3) & 0.08(1) &   --- \\
642: $\Dp\Dm$           &   ---  &   --- & 9.51(8) &  ---  &  ---  &   --- & ---   & ---   &  ---  & 0.010(3) \\
643: $\Dstarz\Dstarzb$  & 0.080(3) & ---   & ---   & 2.60(2) & 0.030(3) & ---   & 0.42(1) & 0.010(3) &  ---  &   --- \\
644: $\Dstarz\Dzb$      &   --- &   --- &   --- & 0.020(3) & 3.40(2) &  ---  & 0.010(3) & 0.46(1) & 0.010(3) &   --- \\
645: $\Dz\Dzb$          &   --- &   --- &   --- &  ---  & 0.010(3) & 12.02(10) &  ---  & 0.010(3) & 0.020(3) &  ---  \\
646: $\Dstarp\Dstarzb$  & 2.60(2) & ---   & ---   & 0.23(1) & 0.010(3) &   --- & 7.52(4) & 0.07(1) &  ---  &   --- \\
647: $\Dstarp\Dzb$      & 0.040(3) & 0.06(2) & ---   &  ---  & 0.11(5) &   --- & 0.03(2) & 13.51(25) & 0.040(3) &  ---  \\
648: $\Dp\Dstarzb$      &   --- & 0.41(1) &   --- & 0.010(3) & 0.010(3) &  ---  &  ---  & 0.070(3) & 3.70(3) &  --- \\
649: $\Dp\Dzb$          &   --- & 0.020(3) & 0.06(1) &  ---  &  ---  & 0.050(3) &  ---  & 0.010(3) & 0.020(3) & 14.93(9) \\
650: \hline \hline
651: \end{tabular}
652: }
653: \end{center}
654: \end{table*}
655: 
656: 
657: \section{Efficiency and crossfeed determination}\label{sec:effs}
658: 
659: The efficiencies are determined using fits to \mes distributions of signal MC events that pass all selection criteria in $\masslik$, $|\Delta E|$, 
660: $\mathcal{F}$, and $L$.
661: There is a small, but non-negligible probability that a signal $B$ decay
662: of mode $i$ is reconstructed as a different signal decay mode $j$.
663: We refer to this as crossfeed.
664: Thus, efficiencies can be represented as a matrix $\epsilon_{ij}$.  
665: where each contributing generated event is weighted by the $D$ and $\Dbar$ decay mode branching fractions. 
666: To determine the elements of $\epsilon_{ij}$, we fit the \mes distributions of signal MC events generated as $B$ decay mode $i$
667: and reconstructed as $B$ decay mode $j$.  The distributions are modeled as the sum of signal and background probability distribution functions (PDFs), where the PDF for 
668: the signal is a Gaussian distribution centered around the $B$ mass, and the PDF for background is an 
669: empirical function~\cite{argus} of the form
670: \begin{equation}
671: \label{eq:argus}
672: f(x) \propto x\sqrt{1-x^2}\exp[-\kappa(1-x^2)],
673: \end{equation}
674: where we define $x \equiv 2\mes/\sqrt{s}$, and $\kappa$ is a parameter determined by the fit.  In \BB MC samples containing signal and background decays, 
675: we find that the \mes distribution is well-described 
676: by adding a simple Gaussian function to the empirical shape in Eq.~\ref{eq:argus}.  We fit the \mes distributions of signal MC events generated as mode $i$ and passing
677: selection criteria in mode $j$ to the above distribution
678: by minimizing the $\chi^2_{ij}$ of each fit with respect to $\kappa_{ij}$ (the $\kappa$ parameter for each mode $(i,j)$), the number of signal events $N^s_{ij}$, 
679: and the number of background 
680: events $N^b_{ij}$.
681: We determine the efficiencies $\epsilon_{ij}$ as $N^s_{ij}/N^g_i$, where $N^g_i$ is the total number of signal MC events that were generated in mode $i$.
682: The diagonal elements of the $\epsilon_{ij}$ matrix (\textit{i.e.} the numbers typically denoted as ``efficiencies'') are in the range (0.2 -- 1.5)$ \times 
683: 10^{-3}$.
684: The main crossfeed source is misidentification between $D^{*0}$ and $D^{*\pm}$ candidates.  
685: The matrix $\epsilon_{ij}$ and the uncertainties on the elements of this matrix are given in Table~\ref{tab:effs}.
686: Crossfeed between different $D$ submodes (\textit{i.e.} mode numbers 12--15 in Table~\ref{tab:modekey}) is negligible. 
687: 
688: \section{Branching fraction results}\label{sec:bfs}
689: 
690: In order to determine the number of signal events in each mode, one must not only account for background which is distributed according to
691: combinatorial phase space, but also for background which can have a different distribution in \mes.
692: It is possible for a component of the background to have an \mes distribution
693: with a PDF that is more similar to signal (\textit{i.e.} a Gaussian distribution centered around 
694: the $B$ mass) than to a phase-space distribution.  Such a component
695: is known as ``peaking'' background and typically derives from background events that have the 
696: same or similar final state particles as the signal decay mode.  
697: For example, in \Bztodd, peaking background primarily comes from the decays $\Bz \rightarrow DKX$ or
698: $\Bz \rightarrow D\pi X$, where $D \rightarrow \Kpipi$ and $X$ is $K^0$,
699: $\rho$, $a_1$ or $\omega$, and
700: the light mesons ($KX$) or ($\pi X$) fake a $D \rightarrow \Kpipi$ decay.
701: The optimization procedure that was detailed in Sec.~\ref{sec:analopti_sele} eliminates decay submodes that have a 
702: large enough amount of peaking (in addition to combinatorial) background to decrease, rather than increase, the sensitivity for a particular decay;
703: the final selection was detailed in Tables~\ref{tab:globalcuts}, \ref{tab:masslikcuts}, and~\ref{tab:deltaecuts}.
704: We determine the
705: amount of peaking background $P_i$ in each $B$ decay mode $i$ via fitting the \mes distributions of \BB MC simulated events. 
706: We minimize the $\chi^2_{i}$ of each fit, allowing the variables
707: $\kappa^P_{i}$ (representing the ``ARGUS parameter'' described earlier), the number of expected peaking background events in data $P_{i}$, and the number of 
708: phase-space background events $N^{\rm MC bkg}_{i}$, to float.  The fitted number of peaking background events $P_i$ 
709: is compatible with zero, within two standard deviations, for all 
710: modes $i$.
711: 
712: \begin{figure*}[p]
713: \begin{center}
714: \scalebox{0.52}{\includegraphics{dstdst_data_mes.eps}}
715: \hspace*{0.5cm}
716: \scalebox{0.52}{\includegraphics{dstd_data_mes.eps}}
717: 
718: \vspace*{-0.2cm}
719: 
720: \hspace*{0.04cm}
721: \scalebox{0.52}{\includegraphics{dd_data_mes.eps}}
722: \hspace*{0.65cm}
723: \scalebox{0.52}{\includegraphics{dstzdstz_data_mes.eps}}
724: 
725: \vspace*{-0.2cm}
726: 
727: \hspace*{-0.49cm}
728: \scalebox{0.52}{\includegraphics{dstzdz_data_mes.eps}}
729: \hspace*{0.65cm}
730: \scalebox{0.52}{\includegraphics{dzdz_data_mes.eps}}
731: 
732: \vspace*{-0.2cm}
733: 
734: \hspace*{-0.12cm}
735: \scalebox{0.52}{\includegraphics{dstdstz_data_mes.eps}}
736: \hspace*{0.65cm}
737: \scalebox{0.52}{\includegraphics{dstdz_data_mes.eps}}
738: 
739: \vspace*{-0.2cm}
740: 
741: \hspace*{0.25cm}
742: \scalebox{0.52}{\includegraphics{dstzd_data_mes.eps}}
743: \hspace*{0.32cm}
744: \scalebox{0.52}{\includegraphics{dzd_data_mes.eps}}
745: 
746: \vspace*{-0.2cm}
747: \end{center}
748: \caption{
749: \label{fig:data_dstdst}
750: Distributions of \mes for selected candidates in each $D^{(*)}\Dbar^{(*)}$ mode.
751: The error bars represent the statistical errors only.
752: The solid lines represent the fits to the data, and the shaded areas
753: the fitted background.
754: }
755: \end{figure*}
756: 
757: \begin{table*}[t]
758: \caption{\label{tab:bfsacps}
759: Results of the fits for the ten signal decay modes: the number of events for fitted signal $N^{\rm sig}$,
760: the peaking background $P$, and the crossfeed $C$, the branching fractions
761: $\mathcal{B}$, 90\% C.L.~upper limits on
762: branching fractions, previous measurements of branching fractions
763: (for modes that have previous measurements), and charge asymmetries.  The uncertainties are statistical.  For
764: the final branching fraction and charge asymmetry results,
765: the systematic errors are also given.
766: }
767: \begin{center}
768: \begin{tabular}{lr@{$\pm$}lr@{$\pm$}lr@{$\pm$}lr@{}c@{}lccc}
769: \hline\hline
770: \textbf{Mode}                            & \multicolumn{2}{c}{$N^{\rm sig}$} 
771:                                                          & \multicolumn{2}{c}{$P$}  & \multicolumn{2}{c}{$C$}   
772:                                                                            &  \multicolumn{3}{c}{{\boldmath $\mathcal{B}$} $(10^{-4})$}        &  
773:                                                                    $\begin{array}{c} \mbox{\textbf{U.L.}} \\ (10^{-4}) \end{array}$  & 
774:                                                                    $\begin{array}{c} \mbox{Previous $\mathcal{B}$} \\ \mbox{results $(10^{-4})$} \end{array}$ 
775:                                    & {\boldmath $\mathcal{A}_{\CP}$} \\
776: \hline
777: {\boldmath \Bz $\to D^{*+}D^{*-}$}       & 270 & 19  & $-1$ &  2 &  4  &  1  &  {\boldmath \BRbztodstdstCV}   & {\boldmath \BRbztodstdstStat}   & {\boldmath \BRbztodstdstSyst}   &                      
778: & 
779:       $\begin{array}{c} 8.1 \pm 0.8 \pm 1.1\;[21] \\ 8.3 \pm 1.6 \pm 1.2\;[22] \\ 9.9^{+4.2}_{-3.3} \pm 1.2\;[23] \end{array}$                            & \\
780: \hline
781: {\boldmath \Bz $\to D^{*\pm}D^{\mp}$}    & 156 & 17  &  1 &  3 &  2  &  1  &  {\boldmath \BRbztodstdCV}     & {\boldmath \BRbztodstdStat}     & {\boldmath \BRbztodstdSyst}     
782: &                      &  
783:       $\begin{array}{c} 8.8 \pm 1.0 \pm 1.3\;[24] \\ 11.7 \pm 2.6^{+2.2}_{-2.5}\;[25] \\ 6.7^{+2.0}_{-1.7} \pm 1.1\;[26] \end{array}$  & 
784:                                                      {\boldmath $\quad\ACPbztodstdNum\;\mbox{[27]}$} \\
785: \hline
786: {\boldmath \Bz $\to D^{+}D^{-}$}         &  63 &  9  &  1 &  2 &  0  &  0  &  {\boldmath \BRbztoddCV}       & {\boldmath \BRbztoddStat}       & {\boldmath \BRbztoddSyst}       &   & 
787: $1.91 \pm 0.51 \pm 0.30\;[28]$   & \\
788: \hline
789: {\boldmath \Bz $\to D^{*0}\Dstarzb$} &  0 &  6  & $-2$ &  2 &  0  &  0  &  \BRbztodstzdstzCV & \BRbztodstzdstzStat & \BRbztodstzdstzSyst & {\boldmath \LIMbztodstzdstzMant} & 
790: $< 270$~\cite{Aleph1} & \\
791: \hline
792: {\boldmath \Bz $\to D^{*0}\Dzb$}  & 10 &  8  & $-2$ &  3 &  1  &  1  &  \BRbztodstzdzCV   & \BRbztodstzdzStat   & \BRbztodstzdzSyst   & {\boldmath \LIMbztodstzdzMant}   &                  
793: & \\
794: \hline
795: {\boldmath \Bz $\to D^{0}\Dzb$}   & $-11$ & 12  & $-8$ & 4 &  0  &  0  &  \BRbztodzdzCV     & \BRbztodzdzStat     & \BRbztodzdzSyst     & {\boldmath \LIMbztodzdzMant}     &                  
796: & \\
797: \hline\hline
798: {\boldmath \Bp $\to D^{*+}\Dstarzb$} & 185 & 20  & $-5$ & 4 & 34  &  4  &  {\boldmath \BRbptodstdstzCV}  & {\boldmath \BRbptodstdstzStat}  & {\boldmath \BRbptodstdstzSyst}  &                      
799: &  $\begin{array}{c} 10.5^{+3.3}_{-2.8} \pm 2.0\;[26] \\ < 110\;[29] \end{array}$ & 
800:                                              {\boldmath $\ACPbptodstdstzNum$ } \\
801: \hline
802: {\boldmath \Bp $\to D^{*+}\Dzb$}  & 115 & 16  &  1 & 4 &  3  &  1  &  {\boldmath \BRbptodstdzCV}    & {\boldmath \BRbptodstdzStat}    & {\boldmath \BRbptodstdzSyst} & & 
803: $\begin{array}{c} 4.57 \pm 0.71 \pm 0.56\;[28] \\ < 130\;[29,30] \end{array}$ & 
804:                                              {\boldmath $\ACPbptodstdzNum$ }   \\
805: \hline
806: {\boldmath \Bp $\to D^{+}\Dstarzb$}  &  63 & 11  & 3 &  3 &  9  &  2  &  {\boldmath \BRbptoddstzCV}    & {\boldmath \BRbptoddstzStat}    & {\boldmath \BRbptoddstzSyst} & & 
807: $< 130$~\cite{Aleph1,Aleph1N} & 
808:                                              {\boldmath $\ACPbptoddstzNum$ }   \\
809: \hline
810: {\boldmath \Bp $\to D^{+}\Dzb$}   & 129 & 20  & $-2$ &  5 &  1  &  1  &  {\boldmath \BRbptoddzCV}      & {\boldmath \BRbptoddzStat}      
811: & {\boldmath \BRbptoddzSyst}      
812: & & 
813:  $\begin{array}{c} 4.83 \pm 0.78 \pm 0.58\;[28] \\ < 67\;[29] \end{array}$  & 
814:                                              {\boldmath $\ACPbptoddzNum$ }    \\
815: \hline\hline
816: \end{tabular}
817: \end{center}  
818: \end{table*}
819: 
820: We then fit the actual data to determine the number of reconstructed signal events in each mode.
821: We fit the \mes distributions of reconstructed $B$ decays that pass all selection criteria in each mode $i$ to a sum of a Gaussian distribution and a phase space
822: distribution (Eq.~\ref{eq:argus}), similar to the PDFs used for efficiency and peaking background fits described above.  We minimize the
823: $\chi^2_{i}$ of each data fit, allowing the parameter $\kappa_{i}$, the number of signal events in data $N^{\rm sig}_{i}$, and the number of background events in data 
824: $N^{\rm bkg}_{i}$, each to float. 
825: The \mes distributions and the results of the fits are shown in Fig.~\ref{fig:data_dstdst}.
826: The branching fractions $\mathcal{B}_i$ are then determined via the equation
827: \begin{equation}
828: \label{eq:bfs}
829: \sum_j \epsilon_{ij} \mathcal{B}_j N_B = N^{\rm sig}_i - P_i
830: \end{equation}
831: where $N_B = N_{\BB} = (231.7 \pm 2.6) \times 10^{6}$ is the total number of charged or neutral $B$ decays in the data sample,
832: assuming equal production rates of charged and neutral $B$ pairs.
833: 
834: 
835: We determine the branching fractions as
836: \begin{equation}
837: \mathcal{B}_i = \sum_j\epsilon^{-1}_{ij}(N^{\rm sig}_j - P_j)/N_B,
838: \end{equation}
839: (where $\epsilon^{-1}_{ij}$ is the inverse of matrix $\epsilon_{ij}$) yields
840: the branching fractions given in Table~\ref{tab:bfsacps}.  Note that the measured branching fractions for the three modes 
841: $\Bz \to D^{(*)0}\Dbar^{(*)0}$ are not significantly greater than zero.  Thus, we have determined upper limits on
842: the branching fractions for these modes.  The 90\% confidence level (C.L.) upper limits quoted in Table~\ref{tab:bfsacps} are 
843: determined using the Feldman-Cousins method~\cite{FeldmanCousins} and include all systematic uncertainties detailed below. 
844: Since the branching fractions can be correlated through the use of Eq.~\ref{eq:bfs}, we also provide the covariance matrix, with 
845: all systematic uncertainties included, in Table~\ref{tab:covBF}. The covariance matrix is obtained via the approximation given in~\cite{matinverr}.
846: 
847: \begin{table*}[t]
848: \caption{Covariances of $B \to D^{(*)}\bar{D}^{(*)}$ branching fractions (with all systematic uncertainties included), in units of $10^{-8}$.}
849: \begin{center}
850: \begin{tabular}{rcccccccccc}
851: \hline \hline
852: Mode  &$\Dstarp\Dstarm$ &$\Dstarpm\Dmp$  & $\Dp\Dm$   &  $\Dstarz\Dstarzb$   & $\Dstarz\Dzb$   & $\Dz\Dzb$   & $\Dstarp\Dstarzb$   & $\Dstarp\Dzb$  & $\Dp\Dstarzb$   &  $\Dp\Dzb$ \\
853: \hline		                                                                	          
854: $\Dstarp\Dstarm$  & 1.26 &    0.55 &    0.22 & $-$0.15 &    0.07 & $-$0.01 &    0.73 &    0.33 &    0.54 &    0.30 \\
855: $\Dstarpm\Dmp$    &      &    0.91 &    0.26 & $-$0.08 &    0.04 & $-$0.01 &    0.46 &    0.19 &    0.37 &    0.26 \\
856: $\Dp\Dm$          &      &         &    0.39 & $-$0.03 &    0.02 &    0.00 &    0.16 &    0.08 &    0.26 &    0.16 \\
857: $\Dstarz\Dstarzb$ &      &         &         &    1.27 & $-$0.04 &    0.00 & $-$0.53 & $-$0.06 & $-$0.13 & $-$0.05 \\
858: $\Dstarz\Dzb$     &      &         &         &         &    1.25 &    0.00 &    0.07 & $-$0.02 &    0.05 &    0.02 \\
859: $\Dz\Dzb$         &      &         &         &         &         &    0.22 & $-$0.01 &    0.00 & $-$0.01 &    0.00 \\
860: $\Dstarp\Dstarzb$ &      &         &         &         &         &         &    2.60 &    0.31 &    0.55 &    0.27 \\
861: $\Dstarp\Dzb$     &      &         &         &         &         &         &         &    0.43 &    0.19 &    0.11 \\
862: $\Dp\Dstarzb$     &      &         &         &         &         &         &         &         &    2.61 &    0.27 \\
863: $\Dp\Dzb$         &      &         &         &         &         &         &         &         &         &    0.53 \\
864: \hline \hline
865: \end{tabular}
866: \label{tab:covBF}
867: \end{center}
868: \end{table*}
869: 
870: 
871: \section{Branching fraction systematic uncertainties}\label{sec:bfsysts}
872: 
873: \begin{table*}[t]
874: \caption{Estimates of branching fraction systematic uncertainties (as percentages of the absolute values of the branching fraction central values) for all $B$ modes, after propagating the errors through Eq.~\ref{eq:bfs}.
875: The totals are the sums in quadrature of the uncertainties in each column.
876: Note that the term ``Dalitz weight'' refers to the selection on the reconstructed invariant masses of the $\Km\pip$ and $\Km\piz$ pairs
877: for $\Dz \to \Km\pip\piz$ decays that was described in Sec.~\ref{sec:dmeson_sele}.
878: }
879: \begin{center}
880: \begin{small}
881: \begin{tabular}{lccccccccccc}
882: \hline \hline
883: Mode  & $\Dstarp\Dstarm$ & $\Dstarpm\Dmp$  & $\Dp\Dm$   &  $\Dstarz\Dstarzb$   & $\Dstarz\Dzb$   & $\Dz\Dzb$   & $\Dstarp\Dstarzb$   & $\Dstarp\Dzb$  & $\Dp\Dstarzb$   &  $\Dp\Dzb$ \\
884: \hline
885: $D^{*+}$ BFs        & 1.4 & 0.7 & 0.0 & 0.9 & 0.1 & 0.0 & 0.7 & 0.7 & 0.0 & 0.0 \\
886: $D^{*0}$ BFs        & 0.0 & 0.0 & 0.0 & 4.9 & 1.6 & 0.0 & 2.1 & 0.0 & 4.4 & 0.0 \\
887: $D^{0}$ BFs         & 5.0 & 2.7 & 0.0 & 7.4 & 3.7 & 5.7 & 5.2 & 4.5 & 3.3 & 2.7 \\
888: $D^{+}$ BFs        & 1.4 & 6.5 & 13.2 & 0.1 & 0.2 & 0.4 & 0.1 & 0.3 & 6.5 & 6.5 \\
889: Tracking efficiency & 7.9 & 6.5 & 4.8 & 7.9 & 3.0 & 4.7 & 6.0 & 6.0 & 3.8 & 4.4 \\
890: \KS efficiency      & 0.3 & 0.2 & 0.0 & 0.0 & 0.1 & 0.0 & 0.1 & 0.2 & 0.3 & 0.2 \\
891: Neutrals efficiency & 2.5 & 1.0 & 0.0 & 8.4 & 2.9 & 1.9 & 4.6 & 1.6 & 4.3 & 1.0 \\
892: Kaon identification & 4.6 & 4.7 & 5.0 & 7.3 & 4.9 & 5.4 & 5.0 & 4.6 & 4.6 & 4.7 \\
893: \masslik cut        & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 \\
894: $\cal{F}$ cut       & 0.0 & 0.0 & 0.9 & 0.9 & 0.9 & 0.9 & 0.9 & 0.9 & 0.9 & 0.9 \\
895: $L$ cut             & 0.0 & 0.0 & 0.8 & 0.8 & 0.8 & 0.8 & 0.0 & 0.8 & 0.8 & 0.8 \\
896: \DeltaE cut         & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 \\
897: Dalitz weight cut   & 1.0 & 0.5 & 0.0 & 1.4 & 0.2 & 1.0 & 1.0 & 0.8 & 0.7 & 0.5 \\
898: P($\chi^2$) cut     & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 & 3.8 \\
899: Fit model          & 1.8 & 3.6 & 3.1 & 5.4 & 6.7 & 44.6 & 4.9 & 2.8 & 7.0 & 3.6 \\
900: Spin alignment      & 1.0 & 0.0 & 0.0 & 6.1 & 0.0 & 0.1 & 4.1 & 0.0 & 0.0 & 0.0 \\
901: Peaking background & 0.9 & 2.0 & 2.9 & 24.5 & 32.3 & 144.6 & 3.1 & 3.4 & 4.9 & 4.0 \\
902: Crossfeed           & 0.4 & 0.6 & 0.8 & 1.9 & 1.1 & 1.6 & 0.6 & 0.4 & 1.0 & 0.6 \\
903: $N_{\BB}$           & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 & 1.1 \\
904: \hline
905:   Total  & 12.0 & 12.3 & 16.1 & 31.0 & 34.2 & 151.7 & 13.6 & 11.0 & 14.8 & 11.9 \\
906: \hline \hline
907: \end{tabular}
908: \end{small}
909: \label{tab:brsysts}
910: \end{center}
911: \end{table*}
912: Table~\ref{tab:brsysts} shows the results of our evaluation of the systematic uncertainties on the branching fraction measurements.
913: \paragraph{Submode branching fractions} The central values and uncertainties on the branching fractions of the $D$ and \Dstar mesons are propagated into the calculation
914: of the branching fraction measurements.  The world average measurements~\cite{PDG2004} are used.
915: \paragraph{Charged track finding efficiency} From studies of absolute tracking efficiency, we assign a systematic uncertainty of 0.8\% 
916: per charged track on the efficiency of finding tracks 
917: other than slow pions from charged \Dstar decays and daughters of \KS decays.  For the slow pions, we assign a systematic uncertainty of 2.2\% each, 
918: as determined from a separate efficiency 
919: study (using extrapolation of slow tracks found in the SVT into the DCH tracking detector and vice-versa).
920: Track finding efficiency uncertainties are treated as 100\% correlated among the tracks in a candidate.
921: These uncertainties are weighted by the $D$ and \Dstar branching fractions.
922: \paragraph{\KS reconstruction efficiency} From a study of the \KS reconstruction efficiency (using an inclusive data sample of events containing one or more \KS, as well as 
923: corresponding MC samples), we assign a 2.5\% 
924: systematic uncertainty for all modes containing a \KS.
925: The value 2.5\% comes from the statistical uncertainty in the ratio of data to MC yields and the variation of this ratio over different selection criteria.
926: The uncertainty is weighted by the $D$ and \Dstar branching fractions.
927: \paragraph{\piz and $\gamma$ finding efficiency} From studies of the neutral particle finding efficiency through the ratios of $\tau^+ \to \rho^+(\pi^+\pi^0)\nu$ to $\tau^+ \to \pi^+\nu$ between data and MC, we assign a 3\% systematic uncertainty per \piz, including 
928: the slow \piz from \Dstar and \Dstarz decays.  For isolated photons from \Dstarz decays, we assign a 1.8\% systematic uncertainty, 100\% correlated with the \piz efficiency uncertainty.
929: These uncertainties are weighted by the $D$ and \Dstar branching fractions.
930: \paragraph{Charged kaon identification} We assign a systematic uncertainty of 2.5\% per charged kaon, according to a study of kaon particle identification efficiency
931: (using kinematically-reconstructed $\Dz \to \Km\pip$ candidates).
932: The uncertainty is weighted by the $D$ and \Dstar branching fractions.
933: \paragraph{Other selection differences between data and MC} Differences in momentum measurement, decay vertex finding efficiency, etc., can result in additional differences between 
934: efficiencies
935: in data and in MC.  We use a sample of the more abundant $B^0 \to D_s^{*+}\Dstarm$ events in data, selected in a similar manner as the
936: $B \to D^{(*)}\Dbar^{(*)}$ modes, to determine these uncertainties.
937: To estimate the systematic error arising from differences
938: between the data and MC $D$ and $D^*$ mass resolutions, we
939: calculate the number of $D_s^*D^*$ events seen in the data and MC
940: as a function of the \masslik cut, while fixing the other selection criteria to their nominal values. The number of observed events is
941: extracted from a fit to the \mes distribution. We then plot the ratio of the data yield
942: ($N_{\rm data}$) to the MC yield ($N_{\rm MC}$) as a function of the
943: \masslik cut over a range of values that gives the same efficiencies as
944: in the $D^{(*)}\Dbar^{(*)}$ analyses.  We find the rms of the
945: $N_{\rm data}/N_{\rm MC}$ ratio and assign this as a systematic
946: uncertainty for applying this cut.  The same technique is used to
947: determine the systematic uncertainties from all other selection criteria in Table~\ref{tab:brsysts}: the selections on 
948: $\cal{F}$, $L$, \DeltaE, the reconstructed invariant masses for $\Dz \to \Km\pi+\piz$ (``Dalitz weight''), and vertex P($\chi^2$).
949: \paragraph{Fit model} The data yield is obtained from an \mes fit where the mean ($\mu$) and
950: width ($\sigma$) of the $B$ mass and the end-point ($\sqrt{s}/2$) of the phase-space distribution (Eq.~\ref{eq:argus}) are fixed. These parameters are
951: estimated and have associated uncertainties.  The nominal value of $\sigma$ is determined from signal MC for each $B$ decay mode.
952: To estimate the systematic uncertainty due to possible differences
953: between the \mes resolutions in data and signal MC, we first look at this
954: difference ($\Delta\sigma = \sigma_{\rm data} - \sigma_{\rm MC}$) for those
955: modes with high purity, including our control sample.
956: These differences are consistent with zero, justifying our use of $\sigma_{\rm MC}$ in obtaining the data
957: yield. We then find the weighted average of $\Delta\sigma$, which is
958: given by $(0.11 \pm 0.08)$ \mevcc.
959: As a conservative estimate, we repeat the data yield determinations
960: by moving $\sigma$ up and down by 0.2 \mevcc, and take the average of the absolute values of the
961: changes in each data yield as the systematic uncertainty of
962: fixing $\sigma$ to the MC value for that $B$ mode.
963: A combined fit of common modes in data is used to determine the nominal values for $\mu$ and for the endpoint of the \mes distribution $\sqrt{s}/2$.
964: Hence, we move the parameters up and down by their fitted errors
965: (0.2 \mevcc for $\mu$ and 0.1 \mevcc for $\sqrt{s}/2$) to obtain their
966: corresponding systematic uncertainties.
967: The quadratic sum of the three uncertainties from $\mu$, $\sigma$ and
968: $\sqrt{s}/2$ gives the systematic uncertainty of the fit model for each $B$ mode.
969: \paragraph{Spin-alignment dependence} The $\Bztodstdst$, $\Bztodstzdstz$, and $\Bchtodstzdst$ decays
970: are pseudoscalar $\to$ vector vector (VV) transitions
971: described by three independent helicity amplitudes $A_0$, $A_{\parallel}$, and $A_{\perp}$~\cite{Dunietz}.  
972: The lack of knowledge of the true helicity amplitudes in
973: the $B \to VV$ final states contributes a systematic uncertainty to
974: the efficiency.  The dominant source of this effect originates
975: from the $p_T$-dependent inefficiency in reconstructing the low-momentum ``soft''
976: pions in the $\Dstarp$ and $\Dstarz$ decays, and the fact that the three helicity amplitudes contribute very
977: differently to the slow pion $p_T$ distributions.  
978: To estimate the size of this effect, MC samples are 
979: produced with a phase-space angular distribution model for the decay products.
980: Each event is then weighted by the
981: angular distribution for given input values of the helicity amplitudes
982: and phase differences.   The efficiency is then determined for a
983: large number of amplitude sets and the observed distributions in efficiencies
984: are used to estimate a systematic uncertainty.
985: For a given iteration, a random number, based on a uniform PDF,
986: is generated for each of the three parameters: $R_\perp, \alpha$, and $\eta$, where
987: \begin{equation}
988: \! R_\perp =  \frac{|A_{\perp}|^2}{|A_0|^2 + |A_{\parallel}|^2 + |A_{\perp}|^2}, \quad \alpha = \frac{|A_{0}|^2  - |A_{\parallel}|^2 }{|A_0|^2 + |A_{\parallel}|^2},
989: \end{equation}
990: and $\eta$ is the strong phase difference between $A_0$ and $A_{||}$. 
991: Since $R_\perp$ for \Bztodstdst has already been measured~\cite{babarDstDstCP}, 
992: a Gaussian PDF with mean and width fixed to the
993: measured values is used instead for that mode.
994: The events of the MC sample are weighted by the
995: corresponding angular distribution and the efficiency is determined
996: (after applying all selection cuts)
997: by fitting the \mes distribution and dividing by the number of generated events.
998: The procedure is repeated 1000 times for each $B \to VV$ sample.
999: The relative spread in efficiencies (rms divided by the mean)
1000: is used to estimate the systematic
1001: uncertainty due to a lack of knowledge of the true amplitudes.
1002: \paragraph{Peaking background and crossfeed} The uncertainties on the peaking background vector $P_i$ and on the
1003: efficiency matrix $\epsilon_{ij}$ are dominated by the available MC statistics.  The resulting uncertainties on each element
1004: of the vector and matrix are propagated through to the branching fraction results via the formalism of Eq.~\ref{eq:bfs}.
1005: \paragraph{Number of \BB} The number of \BB events in the full data sample, and the uncertainty on this number, are determined via
1006: a dedicated analysis of charged track multiplicity and event shape~\cite{FoxW}.  The uncertainty introduces a systematic uncertainty of
1007: 1.1\% on each of the branching fractions.
1008: 
1009: \section{Measurement of {\boldmath \CP}-violating charge asymmetries}\label{sec:acps}
1010: 
1011: \begin{figure*}[p]
1012: \begin{center}
1013: \scalebox{0.53}{\includegraphics{dstd_data_acp.eps}}
1014: 
1015: \vspace*{-1.2cm}
1016:       
1017: \scalebox{0.53}{\includegraphics{dstdstz_data_acp.eps}}
1018: 
1019: \vspace*{-1.2cm}
1020: 
1021: \scalebox{0.53}{\includegraphics{dstdz_data_acp.eps}}
1022: 
1023: \vspace*{-1.2cm}
1024:       
1025: \scalebox{0.53}{\includegraphics{dstzd_data_acp.eps}}
1026: 
1027: \vspace*{-1.2cm}
1028: 
1029: \scalebox{0.53}{\includegraphics{dzd_data_acp.eps}}
1030: 
1031: \vspace*{-0.5cm}
1032: 
1033: \end{center}
1034: \caption{
1035: \label{fig:data_acps}
1036: Fitted distributions of \mes for the two conjugate states of each of the five relevant modes.
1037: The error bars represent the statistical errors only.
1038: The solid lines represent the fits to the data, and the shaded areas
1039: the fitted background.
1040: The raw asymmetries $\mathcal{A}$ are the normalized differences in the amount of signal between the members of each conjugate pair.
1041: }
1042: \end{figure*}
1043: 
1044: To obtain the charge asymmetries $\mathcal{A}_{\CP}$ (defined in Eq.~\ref{eq:acp}), we perform unbinned extended maximum likelihood fits to
1045: the \mes distributions of the selected events in each of the four charged-$B$ decay modes $D^{*+}\Dstarzb$,
1046: $D^{*+}\Dzb$, $D^{+}\Dstarzb$, $D^{+}\Dzb$, and their respective charge conjugates, and in the neutral-$B$ decay mode $\Dstarpm\Dmp$, using
1047: Eq.~\ref{eq:argus} as the PDF for the combinatorial background for both charges in each pair.  The free parameters of each of the five fits individually
1048: are: 1) the combinatorial background shape parameter
1049: $\kappa$, 2) the total number of signal events, 3) the total number of background events,
1050: and 4) the ``raw'' charge asymmetry $\mathcal{A}$.  Parameters 1 and 3 are considered (and thus constrained to be)
1051: the same for both charge states in each mode; this assumption is validated in MC simulation of the background
1052: as well as in control samples of $\Bz \to \Dstarm\rho^+$ and $\Bz \to \Dstarm a_1^+$ decays in data.  
1053: The results of the fits are shown in Fig.~\ref{fig:data_acps}.
1054: Two potentially biasing effects must be considered: there can be a asymmetry in the efficiencies for reconstructing
1055: positively- and negatively-charged tracks, and peaking background and crossfeed between the modes can cause a small difference
1056: between the measured (``raw'') asymmetry and the true asymmetry.  The former of those two effects is discussed in Sec.~VIII below. 
1057: Regarding the latter, to obtain the charge asymmetries $\mathcal{A}_{\CP}$ from the ``raw''
1058: asymmetries $\mathcal{A}$, very small corrections for peaking background and crossfeed between modes must be made.
1059: Using the terminology of Eq.~\ref{eq:bfs}, and 
1060: considering the branching fractions $\mathcal{B}_i$ to be sums of a ``$+$'' mode (with a \Bz or \Bp, containing a $\bar{b}$ quark, as the initial state) 
1061: and a ``$-$'' mode (with a \Bzb or \Bm, which contain a $b$ quark, as the initial state): $\mathcal{B}_i \equiv \mathcal{B}_i^+ + \mathcal{B}_i^-$,
1062: we have the two equations
1063: \begin{equation}
1064: \sum_j \epsilon_{ij} \mathcal{B}_j^{\pm} N_B = N^{\rm sig \pm}_i - P_i^{\pm}
1065: \end{equation}
1066: for the ``$+$'' and ``$-$'' states respectively, which imply
1067: \setlength{\arraycolsep}{0.0mm}
1068: \begin{eqnarray}
1069: & & \mathcal{B}_i^- \pm \mathcal{B}_i^+ = \nonumber\\
1070: & & \qquad \sum_j \epsilon^{-1}_{ij} [(N^{\rm sig -}_j - P_j^-) \pm (N^{\rm sig +}_j - P_j^+)] / N_B. \qquad 
1071: \end{eqnarray}
1072: \setlength{\arraycolsep}{1.0mm}
1073: As 
1074: \begin{equation}
1075: \mathcal{A}_{\CP,i} \equiv \frac{\Gamma_i^{-} - \Gamma_i^{+}}{\Gamma_i^{-} + \Gamma_i^{+}} = \frac{\mathcal{B}_i^- - \mathcal{B}_i^+}{\mathcal{B}_i^- + \mathcal{B}_i^+}, 
1076: \end{equation}
1077: we have
1078: \begin{equation}
1079: \mathcal{A}_{\CP,i} = \frac{\sum_j \epsilon^{-1}_{ij} [(N^{\rm sig -}_j - P_j^-) - (N^{\rm sig +}_j - P_j^+)]}{\sum_j \epsilon^{-1}_{ij} [(N^{\rm sig -}_j - P_j^-) + (N^{\rm sig +}_j - P_j^+)]}.
1080: \end{equation}
1081: Since $N^{\rm sig}_j \equiv N^{\rm sig -}_j + N^{\rm sig +}_j$
1082: and the ``raw'' asymmetry in a mode $\mathcal{A}_j \equiv \frac{N^{\rm sig -}_j - N^{\rm sig +}_j}{N^{\rm sig -}_j + N^{\rm sig +}_j}$, we have
1083: \begin{equation}
1084: \label{eqn:acpxfeed}
1085: \mathcal{A}_{\CP,i} = \frac{\sum_j \epsilon^{-1}_{ij} [\mathcal{A}_j N^{\rm sig}_j - \mathcal{A}^P_j P_j]}{\sum_j \epsilon^{-1}_{ij} [N^{\rm sig}_j - P_j]}
1086: \end{equation}
1087: where $\mathcal{A}^P_j \equiv \frac{P_j^- - P_j^+}{P_j^- + P_j^+}$ are the charge asymmetries of the peaking backgrounds.
1088: The total yields $N^{\rm sig}_j$, peaking backgrounds $P_j$, and efficiency matrix $\epsilon_{ij}$ are identical to those used for the branching fraction
1089: measurements and are given in Tables~\ref{tab:bfsacps} and~\ref{tab:effs}.
1090: The values $\mathcal{A}^P_j$ are nominally set to 0 and
1091: are varied to obtain systematic uncertainties due to the uncertainty on the charge asymmetry of the peaking background (see Sec.~\ref{sec:acpsysts}).  
1092: Thus, Eq.~\ref{eqn:acpxfeed} is used to determine the
1093: final $\mathcal{A}_{\CP}$ values from the measured asymmetries, in order to account for the small effects due to 
1094: peaking background and crossfeed between modes.
1095: The measured $\mathcal{A}_{\CP}$ values 
1096: are given in Table~\ref{tab:bfsacps}.  They are all consistent with zero, and their errors are dominated by statistical uncertainty.
1097: 
1098: \begin{table*}[t]
1099: \caption{
1100: Summary of the systematic uncertainties estimated for the $\mathcal{A}_{\CP}$ asymmetries, in \%.  
1101: } 
1102: \begin{center}
1103: \begin{tabular}{lccccc}
1104: \hline\hline
1105: Systematics source                                 & $\Bz \to D^{*\pm}D^{\mp}$ & $\Bp \to \Dstarp\Dstarzb$ & $\Bp \to \Dstarp\Dzb$ & $\Bp \to \Dp\Dstarzb$ & $\Bp \to \Dp\Dzb$  \\
1106: \hline
1107: Slow pion charge asymmetry                         & $0.53$          &    $0.53$       &  $0.53$     &   ---       &   ---      \\
1108: Charge asymmetry from other tracks                 & $0.80$          &    $0.80$       &  $0.80$     &  $0.80$     &  $0.80$    \\
1109: Amount of peaking bkgd.                            & $0.06$          &    $0.42$       &  $0.19$     &  $0.64$     &  $0.53$    \\
1110: $\mathcal{A}_{\CP}$ of peaking bkgd.               & $0.42$          &    $0.09$       &  $0.58$     &  $3.36$     &  $0.85$    \\
1111: Crossfeed uncertainty                              & $0.00$          &    $0.00$       &  $0.01$     &  $0.31$     &  $0.00$    \\
1112: \mes resolution uncertainty                        & $0.20$          &    $0.04$       &  $0.04$     &  $0.01$     &  $0.14$    \\
1113: $B$ mass uncertainty                               & $0.20$          &    $0.37$       &  $1.38$     &  $1.38$     &  $0.53$    \\
1114: Uncertainty in $\sqrt{s}$                          & $0.00$          &    $0.03$       &  $0.08$     &  $0.05$     &  $0.05$    \\
1115: Potential fit bias                                 & $0.74$          &    $1.97$       &  $1.19$     &  $0.53$     &  $1.66$    \\
1116: \hline 
1117: TOTAL $\delta(\mathcal{A}_{\CP})$                  & $1.6$          &    $2.4$       &  $2.3$     &  $3.8$     &  $2.2$    \\
1118: \hline \hline
1119: \end{tabular}
1120: \vspace{0.3cm} 
1121: \label{tab:acpsysts}
1122: \end{center}
1123: \end{table*}
1124: 
1125: \setcounter{paragraph}{0}
1126: 
1127: \section{Systematic uncertainties on charge asymmetry measurements}\label{sec:acpsysts}
1128: 
1129: Table~\ref{tab:acpsysts} shows the results of our evaluation of the various sources of systematic uncertainty that are important for the $\mathcal{A}_{\CP}$ measurements.
1130: \paragraph{Slow \pipm charge asymmetry} A charge asymmetry in the reconstruction efficiency of the low-transverse-momentum charged pions
1131: from $\Dstarpm \to \Dz\pipm$ decays can cause a shift in $\mathcal{A}_{\CP}$ by biasing the rates of positively charged vs.~negatively charged decays for each
1132: mode.  We estimate this systematic uncertainty by using data control samples of $\Bz \to \Dstarm X^+$ and  $\Bzb \to \Dstarp X^-$ decays, where
1133: $X$ is either $\pi$, $\rho$, or $a_1$, and determining if there is an asymmetry in the number of \Dstarp vs.~\Dstarm reconstructed.  There
1134: are two potential biases of this technique: 1) a charge asymmetry in tracks other than the slow charged pions, and 2) the presence of doubly-Cabibbo-suppressed
1135: $\Bz(\Bzb) \to \Dstarmp X^{\pm}$ decays which could potentially introduce a direct-\CP-violating asymmetry between the two states in the control sample.  
1136: Discussion of 1) is detailed in the paragraph below, and the rate of 2) has been determined in analyses such as Refs.~\cite{exclS2bg} and~\cite{partialS2bg} to
1137: be of order $0.1\%$, well below the sensitivity for this measurement.  We combine the information from the control sample modes and determine
1138: an uncertainty of 0.5\% for each $\mathcal{A}_{\CP}$ measurement for modes with a charged slow pion.
1139: \paragraph{Charge asymmetry from tracks other than slow \pipm} Auxilliary track reconstruction studies place a stringent bound on 
1140: detector charge asymmetry effects at transverse momenta above 200 \mevc.  Such tracking and PID systematic effects were studied in detail in
1141: the analysis of $B \to \phi K^{*}$~\cite{phikstar}.  We assign a 0.2\% systematic per charged track, thus an overall systematic
1142: of 0.4\% per mode (as the positively charged and negatively charged decays for each mode have, on balance, one positive vs.~one negative track respectively).
1143: This systematic uncertainty is added linearly to the slow pion charge asymmetry systematic due to potential correlation.
1144: \paragraph{Amount of peaking background} Peaking background can potentially bias $\mathcal{A}_{\CP}$ measurements in two ways:
1145: 1) a difference in the total amount of peaking background from the expected total amount can, to second order, alter the measured
1146: asymmetry between the positively charged and negatively charged decays, 2) a more direct way for peaking background to alter the measured $\mathcal{A}_{\CP}$ would
1147: be if the peaking background itself were to have an asymmetry between the amount that is reconstructed as positively charged and the amount reconstructed as negative.  
1148: 1) is discussed here; 2) is discussed in the paragraph below.  The systematic uncertainty due to the uncertainty on the total amount of 
1149: peaking background in the five decays is determined via the formalism of Eq.~\ref{eqn:acpxfeed}.  Namely, the uncertainty is given by
1150: \begin{equation}
1151: \label{eqn:acppbsyst}
1152: \delta\mathcal{A}_{\CP,i} = \frac{(\sum_j \epsilon^{-1}_{ij} \mathcal{A}_j N^{\rm sig}_j) \times \sqrt{\sum_j (\epsilon^{-1}_{ij})^2 (\delta P)^2_j}}
1153:                                  {(\sum_j \epsilon^{-1}_{ij} [N^{\rm sig}_j - P_j])^2}
1154: \end{equation}
1155: where $(\delta P)_j$ are the uncertainties on the amount of peaking background (which are given, along with the other parameters in the equation, 
1156: in Table~\ref{tab:bfsacps}).
1157: \paragraph{$\mathcal{A}_{\CP}$ of peaking background} The systematic uncertainty due to the $\mathcal{A}_{\CP}$ of the peaking background is also determined using the formalism 
1158: of Eq.~\ref{eqn:acpxfeed}.  Namely, the uncertainty is given by
1159: \begin{equation}
1160: \label{eqn:acppbacpsyst}
1161: \!\!\!\!\delta^{\prime}\mathcal{A}_{\CP,i} = \frac{(\sum_j \epsilon^{-1}_{ij} \mathcal{A}_j N^{\rm sig}_j) \!\times\!\! \sqrt{\sum_j (\epsilon^{-1}_{ij})^2 (\delta A^P)^2_j P^2_j}}
1162: {(\sum_j \epsilon^{-1}_{ij} [N^{\rm sig}_j - P_j]) (\sum_j \epsilon^{-1}_{ij} \mathcal{A}_j N^{\rm sig}_j)}.
1163: \end{equation}
1164: Investigation of the sources of the peaking background in these modes motivates a conservative choice of 0.68 for the $(\delta A^P)_j$ values.
1165: \paragraph{Amount of crossfeed} The systematic error due to uncertainties in the amount of crossfeed between the modes is also 
1166: determined via the formalism of
1167: Eq.~\ref{eqn:acpxfeed}.  Namely, the uncertainty is given by
1168: \begin{widetext}
1169: \begin{eqnarray}
1170: \label{eqn:acpxfeedsyst}
1171: \delta^{\prime\prime}\mathcal{A}_{\CP,i} & = & \frac{\sqrt{\sum_{jk} \mathcal{A}_j N^{\rm sig}_j {\rm cov}(\epsilon^{-1}_{ij},\epsilon^{-1}_{ik}) \mathcal{A}_k N^{\rm sig}_k}}
1172:                                                     {\sum_j \epsilon^{-1}_{ij} [N^{\rm sig}_j - P_j]}\\
1173:  &   & \qquad - \quad \frac{(\sum_j \epsilon^{-1}_{ij} \mathcal{A}_j N^{\rm sig}_j) \times 
1174:                                                       (\sqrt{\sum_{jk} [N^{\rm sig}_j - P_j] {\rm cov}(\epsilon^{-1}_{ij},\epsilon^{-1}_{ik})  [N^{\rm sig}_k - P_k]})}
1175:                                                       {(\sum_j \epsilon^{-1}_{ij} [N^{\rm sig}_j - P_j])^2}.\nonumber
1176: \end{eqnarray}
1177: \end{widetext}
1178: The covariance between the elements of the inverse efficiency matrix is obtained using the method of Ref.~\cite{matinverr}. 
1179: The very small systematic uncertainty due to crossfeed is thus obtained using Eq.~\ref{eqn:acpxfeedsyst} and the amounts of crossfeed and their uncertainties that are given in 
1180: Table~\ref{tab:effs}.
1181: \paragraph{Uncertainty in \mes resolution, $B$ mass, and $\sqrt{s}$}
1182: The uncertainties in \mes resolution and the beam energy $\sqrt{s}$ are
1183: determined by varying these parameters within their fitted $\pm1 \sigma$ ranges and observing the resulting changes in $\mathcal{A}_{\CP}$.
1184: The uncertainty in the reconstructed
1185: $B$ mass can also have an impact on the fitted \mes distributions and thus on the fitted 
1186: $\mathcal{A}_{\CP}$ values.  Varying the $B$ mass between the fitted value and the $\pm1 \sigma$ range of the nominal \Bz or \Bp invariant mass
1187: allows the determination of the
1188: resulting effect on the $\mathcal{A}_{\CP}$ values.
1189: \paragraph{Potential fit bias} Uncertainties in the potential biases of the $\mathcal{A}_{\CP}$ fits are determined by performing the fits on large samples
1190: of MC simulation of the signal decay modes and of \BB and continuum background decays.  All results are consistent with zero bias, and the 
1191: uncertainties of the fitted asymmetries on the simulated data samples are conservatively assigned as systematic uncertainties from biases of the fits.
1192: 
1193: 
1194: 
1195: \section{Implications for {\boldmath $\gamma$}}\label{sec:gamma}
1196: 
1197: Information on the weak phase $\gamma$ may be obtained by
1198: combining information from $B \to D^{(*)} \Dbar^{(*)}$ and $B \to D^{(*)}_s \Dbar^{(*)}$ branching
1199: fractions, along with \CP asymmetry measurements in $B \to D^{(*)} \Dbar^{(*)}$,
1200: and using an SU(3) relation between the $D^{(*)} \Dbar^{(*)}$ and $D^{(*)}_s \Dbar^{(*)}$ decays~\cite{DL,ADL}.
1201: For this analysis, we assume that the breaking of SU(3) can be parametrized via the ratios of decay
1202: constants $f_{D_s^{(*)}}/f_{D^{(*)}}$, which are quantities that can be
1203: determined either with lattice QCD or from experimental measurements~\cite{lattice}.
1204: 
1205: In this model, one obtains the relation (for $\Bz \to \Dp\Dm$ and 
1206: individual helicity states of $\Bz \to \Dstarp\Dstarm$):
1207: \begin{equation}
1208: \mathcal{A}_{ct}^2 = \frac{a_R \cos(2\beta + 2\gamma) - a_{\rm indir}\sin(2\beta + 2\gamma) - \mathcal{B}}{\cos 2\gamma - 1} 
1209: \label{eqn:maineq}
1210: \end{equation}
1211: where
1212: \begin{eqnarray}
1213: \mathcal{B} & \equiv & \frac{1}{2}(|A^D|^2 + |\bar{A}^D|^2) = \\
1214:   &        & \mathcal{A}_{ct}^2 + \mathcal{A}_{ut}^2 + 2\mathcal{A}_{ct}\mathcal{A}_{ut}\cos\delta\cos\gamma , \nonumber\\
1215: a_{\rm dir} & \equiv & \frac{1}{2}(|A^D|^2 - |\bar{A}^D|^2) = \\
1216:             &        & -2\mathcal{A}_{ct}\mathcal{A}_{ut}\sin\delta\sin\gamma ,\nonumber\\
1217: a_{\rm indir} & \equiv & \Im[e^{-2i\beta}(A^{D})^{*}\bar{A}^D] = -\mathcal{A}_{ct}^{2}\sin 2\beta\nonumber\\
1218:         & &     - 2\mathcal{A}_{ct}\mathcal{A}_{ut}\cos\delta\sin(2\beta + \gamma)\\
1219:         & &     - \mathcal{A}_{ut}^2\sin(2\beta + 2\gamma) ,\nonumber
1220: \end{eqnarray}
1221: and
1222: \begin{equation}
1223: a_R^2 \equiv \mathcal{B}^2 - a_{\rm dir}^2 - a_{\rm indir}^2.
1224: \end{equation}
1225: $A^D$ and $\bar{A}^D$ represent amplitudes of a given $\Bz$ and $\Bzb \to D^{(*)+}D^{(*)-}$ decay respectively,
1226: $\mathcal{B}$ represents the corresponding average branching fraction, and $a_{\rm dir}$ and $a_{\rm indir}$
1227: represent the corresponding direct and indirect \CP asymmetries respectively.
1228: The phases $\beta$ and $\gamma$ are the CKM phases and $\delta$ is a strong phase difference.
1229: $\mathcal{A}_{ct} \equiv |(T + E + P_c - P_t - P_{EW}^C)V_{cb}^*V_{cd}|$ and
1230: $\mathcal{A}_{ut} \equiv |(P_u - P_t -P_{EW}^C)V_{ub}^*V_{ud}|$ are the magnitudes of the combined $B \to D^{(*)} \Dbar^{(*)}$ decay amplitudes containing
1231: $V_{cb}^*V_{cd}$ and $V_{ub}^*V_{ud}$ terms respectively, and the $T$, $P$, and $E$ terms are the
1232: tree, penguin, and the sum of exchange and annihilation amplitudes respectively~\cite{DL}.  
1233: One can directly measure the parameters $\mathcal{B}$, $a_{\rm dir}$, and $a_{\rm indir}$ using information from
1234: $B \to D^{(*)} \Dbar^{(*)}$ decays; the parameter $\mathcal{A}_{ct}$ using information from $B \to D^{(*)}_s \Dbar^{(*)}$ decays;
1235: and the weak phase $\beta$ can be obtained from the measurements of $\sin 2\beta$ based on $\Bz \to c\bar{c}\KS$
1236: decays~\cite{HFAG} thus allowing for
1237: solution of $\gamma$ (up to two discrete ambiguities) via Eq.~\ref{eqn:maineq}.
1238: As the vector-pseudoscalar modes $\Bz \to \Dstarpm\Dmp$ are not \CP eigenstates,
1239: a slightly more complicated analogue to Eq.~\ref{eqn:maineq} is needed for these modes~\cite{ADL}.  Measurement of $\mathcal{A}_{\CP}$ for $\Dstarpm\Dmp$
1240: is also necessary to obtain information on $\gamma$ from the vector-pseudoscalar modes.
1241: 
1242: Using these relations, there are four variables besides $\beta$ for each $B \to D^{(*)} \Dbar^{(*)}$ decay for which to solve: $\mathcal{A}_{ct}$,
1243: $\mathcal{A}_{ut}$, $\delta$, and $\gamma$.  The branching fraction and the direct and indirect
1244: \CP asymmetries of the $B \to D^{(*)} \Dbar^{(*)}$ decay provide three measured quantities.  
1245: The other measurement that can be used is the branching fraction of the corresponding
1246: $B \to D^{(*)}_s \Dbar^{(*)}$ decay, by using the relation expressed in Eq.~\ref{eq:DsDrel}.
1247: 
1248: The values $a_{\rm indir}$ can, of course, only be measured in the neutral $B \to D^{(*)} \Dbar^{(*)}$ decays.  However, the charged
1249: $B \to D^{(*)} \Dbar^{(*)}$ decays can supplement the neutral decays by adding information on $\mathcal{B}$ and $a_{\rm dir}$, assuming only
1250: isospin symmetry between the charged and neutral modes.  Thus, information from the charged $B$ decay modes can assist the $\gamma$
1251: determination.
1252: 
1253: SU(3)-breaking effects can distort the relation between $D^{(*)} \Dbar^{(*)}$ and $D^{(*)}_s \Dbar^{(*)}$ decays as expressed in 
1254: Eq.~\ref{eqn:maineq}.
1255: However, the SU(3)-breaking can be parametrized by the ratio of decay constants $f_{D_s^{(*)}}/f_{D^{(*)}}$, such that
1256: the amplitude for $B \to D^{(*)}_s \Dbar^{(*)}$ decays
1257: \begin{equation}
1258: \mathcal{A}_{ct}^{\prime} = f_{D_s^{(*)}}/f_{D^{(*)}} \times \mathcal{A}_{ct}/\sin\theta_c
1259: \label{eq:DsDrel}
1260: \end{equation}
1261: where $\theta_c$ is the Cabibbo angle~\cite{PDG2004} and the parentheses
1262: around the asterisks correspond to the $B \to D^{(*)} \Dbar^{(*)}$ and $B \to D^{(*)}_s \Dbar^{(*)}$ decays that are used.
1263: The theoretical uncertainty of this relation is determined to be 10\%~\cite{DL}.
1264: 
1265: 
1266: We thus use the information from the vector-vector (VV) decays \Bz $\to D^{*+}D^{*-}$
1267: and \Bp $\to D^{*+}\Dstarzb$ and pseudoscalar-pseudoscalar (PP) decays \Bz $\to D^{+}D^{-}$
1268: and \Bp $\to D^{+}\Dzb$, as well as the vector-pseudoscalar (VP) decays
1269: $\Bz \to D^{*\pm}D^{\mp}$, $\Bp \to \Dstarp\Dzb$, and $\Bp \to \Dp\Dstarzb$,
1270: to form constraints on $\gamma$ using the method of Refs.~\cite{DL,ADL}.
1271: 
1272: To use the VV decays, we must make the assumption that the strong phases for
1273: the $0$ and $\|$ helicity amplitudes are equal.  
1274: The constraints from the PP decays require no such assumption.
1275: The assumption of equal $0$ and $\|$ helicity amplitudes is theoretically supported by a QCD factorization argument described in~\cite{ADL}.  
1276: Then, using Eq.~\ref{eqn:maineq},
1277: we combine the \Bz $\to D^{*+}D^{*-}$ and \Bp $\to D^{*+}\Dstarzb$ branching
1278: fractions and $\mathcal{A}_{\CP}$ information given above with measurements of the \Bz $\to D_s^{*-}D^{*+}$ and \Bp $\to D_s^{*+}\Dstarzb$ branching 
1279: fractions~\cite{PDG2004},
1280: measurements of the \Bz $\to D^{*+}D^{*-}$ time-dependent \CP asymmetries~\cite{babarDstDstCP,Belle2}, and the world-average values of $\sin
1281: 2\beta$~\cite{HFAG} and $\sin \theta_c$~\cite{PDG2004}.
1282: 
1283: We use a fast parametrized
1284: MC method, described in Ref.~\cite{ADL}, to determine the confidence intervals for
1285: $\gamma$.
1286: We consider 500 values for $\gamma$, evenly spaced between 0 and
1287: $2\pi$.  For each value of $\gamma$ considered, we generate 25000 
1288: MC experiments, with inputs that are generated according to Gaussian distributions with widths
1289: equal to the experimental
1290: errors of each quantity.  
1291: For each experiment, we generate random
1292: values of each of the experimental inputs according to Gaussian
1293: distributions, with means and sigmas according to the measured central
1294: value and total errors on each experimental quantity.  We make the
1295: assumption that the ratio $f_{D_s^*}/f_{D^*}$ is equal to $f_{D_s}/f_D
1296: = 1.20 \pm 0.06 \pm 0.06$ \cite{lattice}, allowing for the additional 10\% theoretical uncertainty~\cite{DL}.
1297: We then calculate the resulting values of $\mathcal{A}_{ct}$,
1298: $a_{\rm dir}$, $a_{\rm indir}$, and $B$, given the generated random
1299: values (based on the experimental values). When the quantities
1300: $a_{\rm dir}$, $a_{\rm indir}$, and $B$, along with $\beta$ and the
1301: value of $\gamma$ that is being considered, are input into
1302: Eq.~(\ref{eqn:maineq}), we obtain a residual value for each
1303: experiment, equal to the difference of the left- and right-hand sides
1304: of the equation. 
1305: Thus, using Eq.~\ref{eqn:maineq}, the 25000 trials per value of  $\gamma$ provide
1306: an ensemble of residual values that are used to create a likelihood for $\gamma$ to be at that value, given the experimental inputs.
1307: The likelihood, as a function of $\gamma$, can be
1308: obtained from $\chi^2 (\gamma)$, where $\chi^2 \equiv (\mu/\sigma)^2$,
1309: $\mu$ is the mean of the above ensemble of residual values, and
1310: $\sigma$ is the usual square root of the variance.  The value of
1311: $\chi^{2}(\gamma)$ is then considered to represent a likelihood which is
1312: equal to that of a value $\chi$ standard devations of a Gaussian
1313: distribution from the most likely value(s) of $\gamma$.
1314: We define the ``exclusion level,'' as a function of the value of $\gamma$, as
1315: follows: the value of $\gamma$ is excluded from a range at a given
1316: C.L. if the exclusion level in that range of $\gamma$ values is 
1317: greater than the given C.L.
1318: 
1319: We now turn to the VP decays. The method using VP decays shares the advantage
1320: with PP decays that no assumptions on strong phases are required.
1321: The disadvantage is that, as we will see, the constraints from the VP modes are weak.
1322: 
1323: We combine the information given above on the \Bz $\to D^{*\pm}D^{\mp}$, $\Bp \to \Dstarp\Dzb$, and $\Bp \to \Dp\Dstarzb$ branching fractions and $\mathcal{A}_{\CP}$
1324: information
1325: with measurements of the \Bz $\to D_s^{*-}D^{+}$, \Bz $\to D_s^{-}D^{*+}$, \Bp $\to D_s^{*+}\Dzb$, and \Bp $\to D_s^{+}\Dstarzb$ branching fractions~\cite{PDG2004},
1326: measurements of the \Bz $\to D^{*\pm}D^{\mp}$ time-dependent \CP asymmetries~\cite{Babar2,belleDstDCP}, and the world-average values of $\sin
1327: 2\beta$~\cite{HFAG} and $\sin^2 \theta_c$~\cite{PDG2004}.
1328: Similar to the MC $\gamma$ determination for the VV and PP modes,
1329: we generate random values of each of the experimental inputs according
1330: to Gaussian distributions, with means and sigmas according to the
1331: measured central value and total errors on each experimental quantity.
1332: We again obtain
1333: a confidence level distribution as a function of $\gamma$. 
1334: 
1335: 
1336: \begin{figure*}[!t]
1337: \begin{center}
1338: \includegraphics[width=0.9\textwidth]{gamma.eps}
1339: \vspace*{-0.35cm}
1340: \caption{ The measured exclusion level, as a function of $\gamma$,
1341: from the combined information from vector-vector,
1342: vector-pseudoscalar, and pseudoscalar-pseudoscalar modes.  The combined information implies that
1343: $\gamma$ is favored to lie in the range $[0.07-2.77]$ radians (with a +0
1344: or $+\pi$ radians ambiguity) radians at 68\% confidence level.}
1345: \label{fig:chi2combined}
1346: \end{center}
1347: \end{figure*}
1348: 
1349: Finally, we can combine information from the VV, PP, and VP modes. The
1350: resulting measured exclusion level as a
1351: function of $\gamma$ from each of the three sets of modes, as well as from their combination,
1352: is shown in Fig.~\ref{fig:chi2combined}. From the combined fit,
1353: we see that $\gamma$ is favored to lie in the range
1354: $[0.07-2.77]$ radians (with a +0 or $+\pi$ radians ambiguity) at 68\%
1355: confidence level. This corresponds to $[4.1^{\circ}-158.6^{\circ}] (+0^{\circ} \mbox{ or } 180^{\circ})$.
1356: 
1357: 
1358: These constraints are generally weaker than those found in Ref.~\cite{ADL} due to the fact
1359: that the measured \CP asymmetry in \Bztodstdst has moved closer to the world-average \stwob,
1360: with the newer \Bztodstdst measurements in Ref.~\cite{dstdst05prl}.  The closer this \CP
1361: asymmetry is to \stwob, the weaker the resulting constraints are on $\gamma$, due to the fact
1362: that the closeness of the \CP asymmetry to \stwob favors the dominance of the tree amplitude, rather than
1363: the penguin amplitude whose phase provides the sensitivity to $\gamma$.
1364: Although the constraints are not strong,
1365: they contribute to the growing amount of information available on $\gamma$
1366: from various sources.
1367: 
1368: \section{Conclusions}
1369: 
1370: In summary, we have measured branching fractions, upper limits, and charge 
1371: asymmetries for all $B$ meson decays to $D^{(*)}\Dbar^{(*)}$.  The results are shown in Table~\ref{tab:bfsacps}.
1372: This includes observation of the decay modes
1373: \Bz $\to D^{+}D^{-}$ and \Bp $\to D^{*+}\Dstarzb$, evidence
1374: for the decay modes \Bp $\to D^{+}\Dstarzb$ and \Bp $\to D^{+}\Dzb$
1375: at $3.8 \sigma$ and $4.9 \sigma$ levels respectively,
1376: constraints on \CP-violating charge asymmetries in the four decay modes \Bp $\to D^{(*)+}\Dbar^{(*)0}$,
1377: measurements of (and upper limits for) the decay modes \Bz $\to D^{*0}\Dzb$
1378: and \Bz $\to D^{0}\Dzb$, and improved branching fractions, upper limits, and charge asymmetries in all other $B \to D^{(*)}\Dbar^{(*)}$ 
1379: modes. 
1380: The results are consistent with theoretical expectation and (when available) previous measurements.
1381: When we combine information from time-dependent \CP asymmetries in 
1382: $\Bz \to D^{(*)+}D^{(*)-}$ decays~\cite{dstdst05prl,dstd05prl} and 
1383: world-averaged branching fractions of $B$ decays to $D_s^{(*)}\Dbar^{(*)}$ modes~\cite{PDG2004} 
1384: using the technique proposed in Ref.~\cite{DL} and implemented in Ref.~\cite{ADL}, we find the 
1385: CKM phase $\gamma$ is favored to lie in the range $[0.07-2.77]$ radians (with a +0
1386: or $+\pi$ radians ambiguity) at 68\% confidence level.
1387: 
1388: 
1389: 
1390: \section{Acknowledgments}\label{sec:Acknowledgments}
1391: 
1392: \input pubboard/acknowledgements
1393: 
1394: 
1395: 
1396: \begin{thebibliography}{99}
1397: 
1398: \bibitem{CKM}
1399: \hyphenation{Ko-ba-ya-shi}
1400: N.~Cabibbo, \jprl{10}, 531 (1963); M.~Kobayashi and T.~Maskawa, \progtp {\bf 49}, 652 (1973).
1401: 
1402: \bibitem{gronauetc}
1403: M.~Gronau, \jprl{63}, 1451 (1989); \plb{233}, 479 (1989);
1404: R.~Aleksan \textit{et al.}, Nucl.\ Phys.\ B {\bf 361}, 141 (1991); \plb{317}, 173 (1993).
1405: 
1406: \bibitem{rosner}
1407: J.~Rosner, \jprd{42}, 3732 (1990).
1408: 
1409: \bibitem{sandaxing}
1410: A.~I.~Sanda and Z.~Z.~Xing, \jprd{56}, 341 (1997).
1411: 
1412: \bibitem{phamxing}
1413: X.~Y.~Pham and Z.~Z.~Xing, \plb{458}, 375 (1999).
1414: 
1415: \bibitem{xing2}
1416: Z.~Z.~Xing, \jprd{61}, 014010 (2000).
1417: 
1418: \bibitem{DL}
1419: A.~Datta and D.~London, \plb{584}, 81 (2004).
1420: 
1421: \bibitem{ADL}
1422: J.~Albert, A.~Datta, and D.~London, \plb{605}, 335 (2005).
1423: 
1424: \bibitem{lattice}
1425: J.~Simone, Nucl.\ Phys.\ Proc.\ Suppl.\ {\bf 140}, 443 (2005).
1426: A value $f_{D_s}/f_D = 1.24 \pm 0.07$ is found in the later
1427: reference C.~Aubin {\em et al.}, \jprl{95}, 122002 (2005); this value is 
1428: consistent within uncertainty to the prior value which we use.
1429: One can also obtain information about the ratio of decay constants from
1430: experiment, \textit{e.g.} M.~Artuso {\em et al.} (CLEO Collaboration),
1431: \jprl{95}, 251801 (2005).
1432: 
1433: \bibitem{CC}
1434: Except as noted otherwise, throughout the article we use a particle name to denote either member of 
1435: a charge-conjugate pair.
1436: 
1437: \bibitem{BABARNIM}
1438: B.\ Aubert {\em et al.} (\babar\ Collaboration), \nima{479}, 1 (2002).
1439: 
1440: \bibitem{pep} 
1441: P.~Oddone, eConf-C870126, 423 (1987);
1442: PEP-II Conceptual Design Report, SLAC-R-418 (1993).
1443: 
1444: \bibitem{GEANT4}
1445: S.~Agostinelli {\em et al.} ({\tt GEANT4} Collaboration), \nima{506}, 250 (2003).
1446: 
1447: \bibitem{PDG2004}
1448: Particle Data Group, S.~Eidelman {\em et al.}, \plb{592}, 1 (2004).
1449: 
1450: \bibitem{FoxW}
1451: G.~C.~Fox and S.~Wolfram, \jprl{41}, 1581 (1978).
1452: 
1453: \bibitem{Dalitz}
1454: R.~H.~Dalitz, Phil.~Mag.~{\bf 44}, 1068 (1953).
1455: 
1456: \bibitem{cleofisher} 
1457: D.~Asner {\it et al.} (CLEO Collaboration), \jprd {\bf 53}, 1039 (1996).
1458: 
1459: \bibitem{argus}
1460: H. Albrecht {\em et al.} (ARGUS Collaboration), \plb{241}, 278 (1990); 
1461: {\it ibid} {\bf 254}, 288 (1991).
1462: 
1463: \bibitem{FeldmanCousins}
1464: G.~J.~Feldman and R.~D.~Cousins, \jprd{57}, 3873 (1998).
1465: 
1466: 
1467: \bibitem{matinverr}
1468: M.~Lefebvre {\em et al.}, \nima{451}, 520 (2000).
1469: 
1470: \bibitem{Belle2}
1471: H.~Miyake {\em et al.} (Belle Collaboration), \plb{618}, 34 (2005).
1472: 
1473: \bibitem{Babar1}
1474: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{89}, 061801 (2002).
1475: 
1476: \bibitem{CLEO1}
1477: E.~Lipeles {\em et al.} (CLEO Collaboration), \jprd{62}, 032005 (2000).
1478: 
1479: \bibitem{Babar2}
1480: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{90}, 221801 (2003).
1481: 
1482: \bibitem{Belle1}
1483: K.~Abe {\em et al.} (Belle Collaboration), \jprl{89}, 122001 (2002).
1484: 
1485: \bibitem{justinsThesis}
1486: J.~Albert, Ph.D.~dissertation, Princeton Univ.~(2002), SLAC-R-614.
1487: 
1488: \bibitem{ACPprev}
1489: Previous measurements for $\mathcal{A}_{\CP}$ in \Bz $\to \Dstarpm\Dmp$ are 
1490: $-0.03 \pm 0.11(stat) \pm 0.05(syst)$, from B.\ Aubert {\em et al.} (\babar\ Collaboration), 
1491: \jprl{90}, 221801 (2003); and $+0.07 \pm 0.08 \pm 0.04$, from 
1492: T.~Aushev {\em et al.} (Belle Collaboration), \jprl{93}, 201802 (2004).
1493: 
1494: \bibitem{Belle3}
1495: G.~Majumder {\em et al.} (Belle Collaboration), \jprl{95}, 041803 (2003).
1496: 
1497: \bibitem{Aleph1}
1498: R.~Barate {\em et al.} (Aleph Collaboration), \epjc{4}, 387 (1998).
1499: 
1500: \bibitem{Aleph1N}
1501: The upper limit of $13 \times 10^{-3}$ is for the sum of the
1502: $\Bp \to \Dstarp\Dzb$ and $\Bp \to \Dp\Dstarzb$
1503: branching fractions.
1504: 
1505: \bibitem{Dunietz}
1506: I.~Dunietz \textit{et al.}, \jprd{43}, 2193 (1991).
1507: 
1508: \bibitem{exclS2bg} 
1509: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{92}, 251801 (2004).
1510: 
1511: \bibitem{partialS2bg}
1512: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprd{71}, 112003 (2005).
1513: 
1514: \bibitem{phikstar}
1515: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{91}, 171802 (2003).
1516: 
1517: \bibitem{HFAG}
1518: Heavy Flavor Averaging Group, world charmonium
1519: average, hep-ex/0412073 (a more recent average has now 
1520: become available in hep-ex/0603003).
1521: 
1522: \bibitem{babarDstDstCP}
1523: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{91}, 131801 (2003).
1524: 
1525: \bibitem{belleDstDCP}
1526: T.\ Aushev {\em et al.} (Belle Collaboration), \jprl{93}, 201802 (2004).
1527: 
1528: \bibitem{dstdst05prl}
1529: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{95}, 151804 (2005).
1530: 
1531: \bibitem{dstd05prl}
1532: B.\ Aubert {\em et al.} (\babar\ Collaboration), \jprl{95}, 131802 (2005).
1533: 
1534: 
1535: \end{thebibliography}
1536: 
1537: \end{document}
1538: