Unformatted text preview: ining one of the two highest ET jets satisﬁes a lower
purity (loose) identiﬁcation requirement. Events without an
additional tag are denoted as single tagged. For the tight
requirement, the per jet identiﬁcation efﬁciency is 50%
with a misidentiﬁcation probability of 0.5%. The loose requirement has an efﬁciency of 70% with a misidentiﬁcation
probability of 6.5%. Table XXXVII shows the data yield and
prediction for the initial preselection and for the single- and
double-tagged samples and the expected Higgs-boson signal
for a Higgs boson of mass 115 GeV=c2 .
The backgrounds from tt, W þ jets, Z þ jets, and diboson
processes are estimated using simulated events. These results
are validated using the EW enriched control sample. The
shape of the multijet background is taken from the multijet
modeling sample, and the normalization is determined by
forcing the number of multijet events plus the number of SM
predicted background events to equal the data yield in the
preselection sample. This procedure is validated by comparing the prediction with the multijet enriched sample.
The signal-to-background separation is then further improved using BDTs. For each mH considered, a multijet
boosted decision tree (MJ BDT) with 23 input variables is
Rev. Mod. Phys., Vol. 84, No. 4, October–December 2012 trained on Higgs signal and multijet backgrounds. Events
which have a MJ BDT output greater than 0.6 are retained.
These events are then input to a second BDT (SM BDT)
trained on the remaining backgrounds and Higgs signal
events using the same 23 variables input to the MJ BDT.
Figure 60 shows the BDT outputs for the data, predicted
background, and signal. The agreement between data and
prediction is good. Limits are extracted by ﬁtting signal and
background SM BDT outputs to the data distribution using
the same modiﬁed frequentist algorithm as was used for the
previously described D0 results. Figure 61 shows the Higgs
cross-section limits from the D0 analysis.
The systematic uncertainties for the D0 result are dominated by similar sources as for the CDF analysis. The D0 jet
energy calibration systematic is <10%, and the background
normalization systematic varies between 6% and 20%. D0
also reports additional systematic uncertainties from luminosity, trigger, and identiﬁcation efﬁciencies and b-tagging FIG. 60 (color online). The outputs for the (a) MJ BDT, (b) the
SM BDT for the single-tagged channel, and (c) the SM BDT for the
double-tagged channel for the D0 VH ! ET bb analysis.
6 Hobbs, Neubauer, and Willenbrock: Tests of the standard electroweak model at . . . 30
25 Observed Limit
Expected Limit DØ, 5.2 fb-1 TABLE XXXVIII. Expected and observed yields for the D0
VH ! bb and VH ! jj searches. Only statistical uncertainties are shown.
Search channel 20 Source 15
100 105 110 115 120 125 130 135 140 145 150 mH (GeV/c2) FIG. 61 (color online). The expected and observed 90% C.L.
upper bounds on the Higgs-boson production cross s...
View Full Document
This document was uploaded on 09/28/2013.
- Fall '13