Etude et impact du bruit de fond corrélé pour la mesure de l'angle ...

5.6 Estimation of correlated background for ✓ 13 measurement 153
Analysis Fast Neutron Stopping Muon Total
[cpd (%)] [cpd (%)] [cpd (%)]
High energy analysis – – 1.17 ± 0.15 (16)
OV Tag 0.44 ± 0.14 (32) 0.48 ± 0.11 (23) 0.92 ± 0.18 (19)
This work 0.33 ± 0.16 (48) 0.60 ± 0.20 (37) 0.93 ± 0.26 (28)
Table 5.17: Summary of the correlated background rate obtained from three
di↵erent and independent analysis.
5.6.1 Reducing the correlated background with the OV veto
tel-00821629, version 1 - 11 May 2013
Since the OV is observed to tag the correlated background with high purity,
the ¯⌫ e selection is performed by rejecting the events found in coincidence
with an OV hit, in order to reduce the background contamination.
The background spectral shapes obtained from the OVT analysis are linear,
with small slopes compatible with flat shapes, then the OVV is expected to
reduce the correlated background of about the same factor the FN and the
SM are tagged.
The OV information is available for ⇠ 2/3 of the total detector live time
of 240.17 d, then the averaged background reduction over the full data set
is expected to be of (25 ± 5) % for FN and (49 ± 8) % for SM. Weighting
the FN and SM reductions by the respective rate obtained from the OVT
analysis, a total background reduction of ⇠ 37 % is expected.
The analysis of the high energy tail of the prompt spectrum, provides total
correlated background rate of (0.75±0.12) cpd and (1.17±0.15) cpd respectively
with and without OVV. An averaged correlated background reduction
of (36 ± 7) % is obtained and is consistent with the expected reduction.
The FN and the SM background have been re-evaluated as described in
Sec. 5.4 and Sec. 5.5, adding the OVV. The purities of the FN and the SM
separation, defined as the fraction of FN (SM) contaminating the sample of
SM (FN), change of 5 % and 10 % respectively since the two components
are scaled by the OVV with di↵erent rejection e ciencies. The analysis efficiencies
do not scale by applying the OVV e ciency.
The total correlated background spectral shape upon OVV is shown in
Fig. 5.32. The linear fit provides a slope of ( 0.2 ± 0.9)/(3 MeV) 2 with
2 /dof =8.7/8. The spectral shape is consistent with the one obtained
without OVV, shown in Fig. 5.31, but a small decrease of the slope is observed.
Such change suggest the OV tagging e ciency, obtained by comparing
the ratio between the tagged and the untagged number of events at
high energy, is not constant but slightly decrease at low energy. For this