1: \section{Data analysis and selection criteria}
2:
3: The search for $\nudecay$ described in this paper uses the full data
4: sample collected with the $\overline{V} T_1 T_2$ trigger \cite{nomad}
5: during the years 1996-1998.\ The data correspond to a total
6: number of protons on target ({\em pots}) of $4.1\times 10^{19}$.\
7: The strategy of the analysis was to identify $\nudecay$ candidates
8: by reconstructing in the DC isolated low invariant mass $\pair$ pairs
9: that are accompanied
10: by no other activity in the detector.\ The measured rate
11: of $\pair$ pairs was then compared to that expected from known sources.\
12:
13: %Candidate events are expected to consist of
14: %an $\pair$ pair unaccompanied by any other particles.\
15: The following selection criteria were applied:
16:
17: \begin{itemize}
18: \item two and only two tracks forming a vertex within the DC fiducial volume
19: of $2.4\times2.4\times3.5 ~\rm m^3$ equivalent to a mass of 1.97 tons;
20:
21: \item at least one of the two tracks identified as an electron
22: by the TRD (pion contamination probability $<10^{-3}$ \cite{trd});
23:
24: \item any additional track or converted photon \cite{profile}
25: in the event were allowed only if
26: their energies were less than 0.4 GeV or if they could be identified as
27: due to bremsstrahlung photons from one of the two electron candidates;
28:
29: \item no $\gamma$'s in the
30: ECAL with energy $E_{\gamma}>0.4 ~\rm GeV$ (or $E_{\gamma}>0.3 ~\rm GeV$ for
31: $\gamma$'s converted in the PRS), which are incompatible
32: with bremsstrahlung photons from the initial tracks;
33:
34: \item total HCAL energy $<0.4~\rm GeV$.\ This cut serves as an HCAL
35: veto and is confirmed by random trigger events and Monte Carlo (MC) studies
36: \cite{trigger};
37:
38: \item the total energy of the pair must be greater than 4 GeV and its
39: invariant mass $m_{\pair} < 95~\rm MeV$ to remove background from
40: pairs of particles other than $\pair$.
41:
42: %\item MUON VETO and no FCAL activity, $E_{FCAL}<1~GeV$.
43: \end{itemize}
44:
45: Only 207 events passed these criteria.\ At the next step we used a
46: collinearity variable ${\mathcal C} \equiv 1- cos \Theta_{\nu \pair}$, where
47: $\Theta_{\nu \pair}$ is the angle between the average neutrino beam direction
48: and the
49: total momentum of the reconstructed $\pair$ pair.\ A cut on this variable
50: allowed a more effective background suppression.\
51: A MC simulation of heavy neutrino decays shows (see Figure 5) that
52: the $\nudecay$ events have $\C < 2\times10^{-5}$.\
53: This was true for a $\nus$ mass up to $\simeq 190~\rm MeV$.\
54:
55: In order to avoid biases in the determination of selection criteria, a
56: blind analysis was performed.\ Events in a signal box
57: defined by $\C < 2\times10^{-5}$ were excluded from the analysis of the data
58: until the validity of the
59: background estimate in this region was established.\ This was done by
60: verifying that the MC simulation of standard processes reproduced the data
61: outside the box.\
62:
63:
64: The accuracy of the collinearity
65: determination obtained with MC simulations was checked using
66: a $\nu_{\mu}CC$ data sample with an $\ee$ pair from a photon
67: converted in the DC target at a large ( $\gtrsim 100~\rm cm$)
68: distance from the primary vertex.\
69: Figure 6 shows the $(1-cos\Theta_{\pair})$ distribution of such events
70: in the data and simulation, where $\Theta_{\pair}$ is the
71: angle between the $\ee$ pair momentum and the line joining the primary
72: vertex to the conversion point.\ The small difference between
73: the MC and data distributions in Figure 6 would result in an
74: overall efficiency correction of less than 6\%.\ However, in order to
75: conservatively account for possible instrumental effects not present in the MC,
76: the MC efficiency was multiplied by the efficiency of reconstructing
77: $\pair$ pairs with a collinearity variable $\C < 2\times10^{-5}$ in the data
78: sample of Figure 6 ($\simeq 75\%$).\ Nonetheless, the two distributions are
79: in reasonable agreement at all energies studied.\
80: This validates the resolution in the variable $\C$ (a few mrad in
81: $\Theta_{\nu \pair}$) predicted by the MC program.\
82:
83:
84: The reconstruction efficiency for the $\nudecay$ decay in the NOMAD fiducial
85: volume was calculated from the MC simulation
86: as a function of $\pair$ energy in the
87: range 4 GeV to 50 GeV.\ The MC simulation was used to
88: correct the data for acceptance losses, experimental resolution and
89: reconstruction efficiencies.\ Two checks using both experimental data
90: and the MC simulation have been performed in order to verify
91: the reliability of the simulation and to estimate the systematic uncertainties
92: in the $\pair$ pair efficiency reconstruction
93: in the energy range predicted by the simulation.\\
94:
95: %Two methods were used in order to check the efficiency of $\ee$ pairs
96: %reconstruction in NOMAD as a function of energy.\ Reconstructed
97: %electron-positron pairs were used to compute the efficiency for signal
98: %pairs in the energy range predicted by the simulations,
99:
100: The first method is to select two samples of reconstructed $\pi^0$'s,
101: one in which the two decay photons reach the ECAL, $N^{\pi^0}_{2\gamma}$,
102: and another in which one of the photons converts in the drift chambers,
103: $N^{\pi^0}_{\gamma\ee}$.\ For data and MC events the ratio
104:
105: \begin{equation}
106: R_{Data, MC} =\big( \frac{N^{\pi^0}_{\gamma\ee}}
107: {N^{\pi^0}_{2\gamma}}\big)_{Data, MC}
108: \end{equation}
109:
110: was then formed.\ The value of the double ratio $RR=R_{Data}/R_{MC}$
111: is then a measure of any differences in $\pair$ reconstruction efficiency
112: between the data and MC.\ The use of $\pair$ pairs from
113: $\pi^0$ decay enhances the purity of the $\pair$ sample.\
114:
115: The method works well mostly for the low
116: energy region, $E_{\ee}\lesssim 10~\rm GeV$,
117: when the $2\gamma$ opening angle is relatively large
118: and the distance between the photons in the ECAL is larger than the
119: ECAL cell size.\ At higher energies the precision
120: of this method is affected by the statistical uncertainties in the number of
121: $\pi^0$'s reconstructed
122: in the $2\gamma$ mode, because the resolution on
123: the $2\gamma$ opening angle becomes worse
124: and the $\pi^0$ peak is not well identified anymore.\
125:
126: A similar method allowing a more accurate evaluation of the
127: $\ee$ efficiency correction factor at higher energies
128: is based on the inclusive $\ee/\gamma$ double ratio
129: $RR$ defined again as $RR=R_{Data}/R_{MC}$ with
130: \begin{equation}
131: R_{Data, MC}=\big(\frac{N^{\gamma}_{\ee}}{N^{\gamma}_{\gamma}}\big)_{Data, MC}
132: \end{equation}
133: where $ N^{\gamma}_{\gamma}, ~N^{\gamma}_{\ee}$ are the numbers of
134: single isolated photons and $\ee$ pairs in the same data sample of
135: $\nu_{\mu}CC$ events used for the collinearity check.\
136:
137: It was found that the two methods agree quite well in the low energy region
138: and yield a correction factor close to 1.\ However,
139: in the high energy region the $\ee$ efficiency correction factor
140: varied from $0.7\pm0.04$ to $0.4\pm0.03$
141: depending on the $\pair$ energy.\
142:
143:
144:
145:
146:
147:
148:
149:
150:
151:
152:
153: