1: The systematic errors are summarized in Table~\ref{t:syst}.
2: We consider the following sources of systematic uncertainty:
3: \ben
4: \item Sample composition: the fraction of peaking \Bub\ events in the
5: sample, ${\cal S}_{\Bub}$, is free in the fit; therefore we do not assign any
6: systematic error on this fraction. As mentioned in Section~\ref{s:sample},
7: there is a $\pm 2.3 \%$ systematic uncertainty on the background rate
8: due to possibly peaking combinatorial background. We therefore vary the fraction of \BB\
9: events in the signal region by that amount, repeat the fit, and add the variation in the result
10: to the systematic error
11: (entry (a) of Table \ref{t:syst}). We neglect the statistical error on the sample composition because it
12: is significantly smaller than this systematic effect.
13: \item Analysis bias (entry b): we take 100$\%$ of the bias observed in the fit on the Monte Carlo sample.
14: \item Signal and background PDF description: most of the parameters in the PDF are free in the fit
15: and therefore do not contribute to the systematic error. We vary the parameters that are fixed in the
16: fit by their uncertainty, repeat the fit, and take the corresponding variation in $\tBz$ and \dmd\ as systematic errors.
17: We take the uncertainty on \tBu (entry (c)), and on $D_{\ctl}$ (entry (d)) from the PDG~\cite{ref:PDG}.
18: \item We consider effects due to the detector $z$ scale (entry (e)), the knowledge of the PEP-II boost
19: (entry (f)), the actual position of the beam spot (entry (g)), and SVT alignment (entry (h)).
20: Detailed studies of these effects have been
21: performed in \babar\ in other mixing and lifetime analyses
22: (dilepton \cite{ref:dilepton} and fully reconstructed
23: \BtoDs\ ~\cite{ref:xl} analyses) and provided consistent results.
24: As the methods for vertex reconstruction are
25: very similar we provisionally take
26: our systematic error due to these effects from Ref.~\cite{ref:xl}.
27: \item We vary the parameters describing the fraction of decay-side tags by their statistical errors (entry (i)).
28: \item Binned fitting: we vary the number of bins in \deltat\ from 100 to 250
29: and in \st\ from 20 to 50, and we repeat the fit. Alternatively, we use the average value of the
30: likelihood in the bin instead of the value corresponding to the center of the bin. We take the
31: systematic error to be the maximum variation with reference to the default result (entry (j)).
32: \item Outlier description: we vary the value of the offset of the outlier Gaussian
33: from $-5~{\rm ps}$ to $5~{\rm ps}$. Alternatively, we use a flat PDF for their description (entry (k)).
34: \item Fit range: we vary the \deltat\ fit range from $\pm$ 18~{\rm ps} to
35: $\pm$ 10~{\rm ps} and the \st\ range between 1.8~{\rm ps} and 4.2~{\rm ps} (entry (l)).
36:
37: \bt \bc
38: \caption{\label{t:syst} Systematic uncertainties.}
39: \begin{tabular}{|l|c|c|c|c|}
40: \hline
41: Source & Variation & $\delta\tBz$ (ps) & $\delta\dmd$ (ps$^{-1}$) \\ \hline \hline
42: & & & \\
43: (a) \BB\ fraction & $\pm 2.3\%$ & $\pm0.001$ & $\pm0.001$ \\
44: (b) Analysis bias & - & $\pm0.027$ & $\pm0.003$ \\
45: (c) \tBu\ & 1.671$\pm$0.018 ps& $\pm0.002$ & $\pm0.001$ \\
46: (d) ${\cal D_{\ctl}}$ & 0.65$\pm$0.08 & $\pm0.005$ & $\pm0.001$ \\
47: (e) $z$ scale & - & $\pm 0.006$ & $\pm0.002$ \\
48: (f) PEP-II boost & - & $\pm 0.002$ & $\pm0.001$ \\
49: (g) Alignment & - & $\pm 0.006$ & $\pm0.003$ \\
50: (h) Beam spot position & - & $\pm 0.005$ & $\pm0.001$ \\
51: (i) Decay-side tags & - & $\pm 0.002 $ & $\pm0.001$ \\
52: (j) Binning & - & $\pm 0.002 $ & $\pm0.002$ \\
53: (k) Outlier & - & $\pm 0.001 $ & $\pm0.002$ \\
54: (l) \deltat\ and \st\ cut & - & $\pm 0.008 $ & $\pm0.003$ \\
55: \hline
56: Total & & $\pm 0.030$ & $\pm0.007$ \\
57: \hline \hline
58:
59: \end{tabular}
60: \ec \et
61: \een
62:
63: \section{CONSISTENCY CHECKS}
64: We rely on the assumption that the parameters of the background PDF do not depend on $\mnusq$.
65: We verify this assumption for the continuum background with the fit to the off-peak events.
66: To check this assumption for the \BB\ combinatorial PDF, we perform several cross checks on the data and the Monte Carlo.
67: We compare the simulated \BB \deltat\ distribution in several independent regions of $\mnusq$
68: with Kolmogorov-Smirnov tests and always obtain a reasonable probability for agreement. We fit \BB\ events
69: separately in the signal and background $\mnusq$ region and compare the parameters of the PDF.
70: We fit the signal plus background Monte Carlo events in the signal region only,
71: fixing all the parameters of the \BB\ sample to the values obtained in a fit in the background region, and do not
72: see any appreciable deviation from the result of the full fit.
73: Finally, we repeat
74: the fit both on the data and the Monte Carlo using different $\mnusq$ ranges for the background region.
75: Once again, we do not observe any appreciable difference in $\tBz$ and \dmd\ relative to the default result.
76: