hep-ex0508053/data-analysis-spacut/doc.tex
1:  The method to reject spallation products shown in Section~\ref{sec:bg}
2: is described in this section. In order to identify spallation events,
3: likelihood functions are defined based on the following parameters: 
4: %
5: \begin{itemize}
6:  \item $\Delta L$ : Distance from a low energy event to the track of
7: the preceding muon.
8:  \item $\Delta T$ : Time difference from muon event to the low energy event.
9:  \item $Q_{res}$ : Residual charge of the muon event,
10: $Q_{total} - Q_{unit} \times L_{\mu}$, where $Q_{total}$ is the total charge,
11: $Q_{unit}$ is the total charge per track length and $L_{\mu}$ is reconstructed
12: track length of the muon track.
13: \end{itemize}
14: %
15: Some fraction of muons deposit very large amounts of energy in the detector
16: and in such cases vertex position reconstruction is not reliable.
17: Therefore, the spallation likelihood functions are defined for the following
18: two cases,\\
19: in the case of a successful muon track fit;
20: %
21: \begin{eqnarray}
22: \lefteqn{L_{spa}(\Delta L,\Delta T, Q_{res}) = }\nonumber\\
23: &&L_{spa}^{\Delta L}(\Delta L, Q_{res}) 
24: \times L_{spa}^{\Delta T}(\Delta T) 
25: \times  L_{spa}^{Q_{res}}(Q_{res}),
26: \end{eqnarray}
27: %
28: in the case of a failure fit the muon track;
29: %
30: \begin{equation}
31: L_{spa}(\Delta T, Q_{total}) =
32: L_{spa}^{\Delta T}(\Delta T) 
33: \times  L_{spa}^{Q_{total}}(Q_{total}).
34: \end{equation}
35: %
36: where $L_{spa}^{\Delta L}(\Delta L, Q_{res})$, 
37: $L_{spa}^{\Delta T}(\Delta T)$, $L_{spa}^{Q_{res}}(Q_{res})$ and 
38: $L_{spa}^{Q_{total}}(Q_{total})$ are likelihood functions 
39: for $\Delta L$, $\Delta T$ and $Q_{res}$.
40: 
41:  Figure~\ref{dl} shows the $\Delta L$ distributions 
42: from spallation candidates for six $Q_{res}$ ranges,
43: and the spallation-like function is made from these plots.
44: Here, the selection criteria is $\Delta T < 0.1~sec$ 
45: and $N_{eff} \ge 50$ (equivalent to 7.2~MeV). The peak around 0 $\sim$ 100~cm
46: is caused by spallation events and that around 1500~cm is due to chance
47: coincidence. The distribution of non-spallation events is calculated using
48: a sample which shuffles the event times time randomly; 
49: it is the dashed line in the figure, and the non-spallation-like 
50: function is also made by these plots.
51: After subtracting the non-spallation from the spallation function,
52: and taking a ratio with a random coincidence distribution, 
53: $ L_{spa}^{\Delta L}(\Delta L, Q_{res}) $
54: was obtained.
55: %
56: \begin{figure}[htbp]
57:   \begin{center}
58:   \includegraphics[scale=0.46]{./data-analysis-spacut/dl_lkh.eps}
59:   \caption{$\Delta L$ distribution for each $Q_{res}$ ranges,
60:            (1) $Q_{res}<2.4\times 10^4$~p.e.
61:            (2) $2.4\times 10^4 < Q_{res} < 4.8\times 10^4$~p.e.
62:            (3) $4.8\times 10^4 < Q_{res} < 9.7\times 10^4$~p.e.
63:            (4) $9.7\times 10^4 < Q_{res} < 4.8\times 10^5$~p.e.
64:            (5) $4.8\times 10^5 < Q_{res} < 9.7\times 10^5$~p.e.
65:            (6) $9.7\times 10^5$~p.e.$ < Q_{res}$.
66:            The solid line shows the data, and the dashed line shows the
67:            random sample.}
68:   \label{dl}
69:   \end{center}
70: \end{figure}
71: %
72: 
73:  Figure~\ref{dt} shows the $\Delta T$ distributions
74: from spallation candidates for each time range.
75: Here, the selection criteria is $\Delta L <$ 300~cm
76: and $N_{eff} \ge 50$ and $Q_{res} < 10^6$~p.e..
77: These distributions are fitted with the following function:
78: %
79: \begin{equation}
80: L_{spa}^{\Delta T}(\Delta T) = \sum_{i=1}^{7} A_i \left( \frac{1}{2} \right)^{-\frac{\Delta T}{\tau^i_{1/2}}}
81: \end{equation}
82: %
83: where $\tau^i_{1/2}$ are half-life times of typical radioactive elements 
84: produced by spallation. The evaluated half-life times and 
85: radioisotopes are summarized in Table~\ref{tab:dttable}.
86: %
87: \begin{figure}[htbp]
88:   \begin{center}
89:   \resizebox{8.0cm}{!}{
90:   \includegraphics{./data-analysis-spacut/dt.eps}
91:   \vspace{-1cm}
92:   }
93:  \caption{$\Delta T$ distribution for each time range,
94:           (1) $0.0 < \Delta T < 0.1$~sec
95:           (2) $0.1 < \Delta T < 0.8$~sec
96:           (3) $0.8 < \Delta T < 4.0$~sec
97:           (4) $4.0 < \Delta T < 15$~sec
98:           (5) $15 < \Delta T < 100$~sec
99:           (6) $0 < \Delta T < 100$~sec.
100:    Cross marks are data
101:   and lines are fitted likelihood function $L_{spa}^{\Delta T}(\Delta T)$}.
102:   \label{dt}
103:   \end{center}
104: \end{figure}
105: %
106: \begin{table}
107:  \begin{center}
108:   \begin{tabular}{|c|c|c|c|}
109:    \hline
110:    $i$ & radioactivity & $\tau^i_{1/2}$ & $A_i$ \\
111:    \hline
112:    1 & $^{12}_{5}$B & $2.02\times 10^{-2}$ & $1.20\times 10^5$ \\
113:    \hline
114:    2 & $^{12}_{7}$N & $1.10\times 10^{-2}$ & $3.39\times 10^4$ \\
115:    \hline
116:    3 & $^{9}_{3}$Li & $1.78\times 10^{-1}$ & $3.39\times 10^2$ \\
117:    \hline
118:    4 & $^{8}_{3}$Li & $8.40\times 10^{-1}$ & $1.25\times 10^3$ \\
119:    \hline
120:    5 & $^{15}_{6}$C & 2.45 & $1.35\times 10^2$ \\
121:    \hline
122:    6 & $^{16}_{7}$N & 7.13 & $6.76\times 10^2$ \\
123:    \hline
124:    7 & $^{11}_{4}$Be & 13.83 & 7.79 \\
125:    \hline
126:   \end{tabular}
127:   \caption[]{The parameters of the likelihood $L^{\Delta T}_{spa}(\Delta T)$}
128:   \label{tab:dttable}
129:  \end{center}
130: \end{table}
131: %
132:  In order to obtain the likelihood function for residual charge 
133: $L_{spa}^{Q_{res}}(Q_{res})$, time correlated events with 
134: low energy events ($\Delta T <$ 0.1sec., $N_{eff} \ge$ 50) and 
135: non-correlated events are selected.
136: Figure~\ref{qres}(a) shows the $Q_{res}$ distribution for 
137: spallation and non-spallation candidates.
138: After subtracting the non-spallation from the spallation function,
139: the resulting distribution's fit by a polynomial function 
140: is shown in Figure~\ref{qres}(b).
141: %
142: \begin{figure}[htbb]
143: % \includegraphics[scale=0.21]{./data-analysis-spacut/spaq1.eps}
144: % \includegraphics[scale=0.21]{./data-analysis-spacut/spaq2.eps}
145:  \includegraphics[scale=0.44]{./data-analysis-spacut/spaq.eps}
146:  \caption{(a) $Q_{res}$ distribution for the spallation candidate events 
147: 	  (solid line) and the non-candidate events (dashed line).
148:           (b) Cross marks show the result of subtracting 
149:           dashed from solid line in (a),
150:           and dotted line shows the likelihood function
151:           $L_{spa}^{Q_{res}}(Q_{res})$.}
152: 
153:   \label{qres}
154: \end{figure}
155: %
156: 
157:  To employ the spallation cut, the likelihood values are calculated for 
158: all muons in the previous 100 seconds before a low energy event, and 
159: a muon is selected which gives the maximum likelihood value($L_{max}$). 
160: Figure~\ref{likemax} shows $L_{max}$ distributions for both the 
161: successful muon track reconstruction case and the failed reconstruction case.
162: To be considered a spallation event the selection criteria 
163: are $L_{max} > 0.98$ (when fit succeeded) and $L_{max} > 0.92$ (when 
164: fit failed).
165: 
166:  The dead time for low energy events caused by the spallation cut is
167: estimated to be 21.1\%. This estimation is calculated using a low energy sample
168: which is shuffled in time and position. It has some position dependence because
169: of the SK tank geometry. Figure~\ref{dead} shows the position dependence
170: of the dead time as a function of the distance from the inner barrel 
171: wall and the top or bottom walls. The systematic error due to position 
172: dependence is estimated by comparing between MC and data, and 
173: it is estimated to be $\pm0.2\%$ for flux, and $\pm0.1\%$ for day-night flux
174: difference and other time variations.
175: %
176: \begin{figure}[htbp]
177:   \begin{center}
178:   \includegraphics[width=4.2cm,clip]{./data-analysis-spacut/likemax3.epsi}
179:   \includegraphics[width=4.0cm,clip]{./data-analysis-spacut/likemax2.epsi}
180:   \caption{The maximum likelihood value distribution for
181:       success (left) and failure (right) of muon track reconstruction.
182:       The blank histogram shows all data while the 
183:       hatched histogram shows the plot for the random sample; the latter
184:       is used to evaluate dead time.}
185:   \label{likemax}
186:   \end{center}
187: \end{figure}
188: %
189: \begin{figure}[htbp]
190:   \includegraphics[scale=0.44]{./data-analysis-spacut/deadtime.eps}
191:   \caption{Position dependence of the spallation cut's dead time. 
192:            The horizontal axis shows the distance from the barrel (left), 
193: 	   and the minimum distance from the top or bottom (right).}
194:   \label{dead}
195: \end{figure}
196: 
197: 
198: 
199: 
200: 
201: 
202: