Abstract
We search for the neutral higgs production associated with the W boson using
high-pT isolated like-sign dilepton events in ppbar collisions at
sqrt(s) = 1.96 TeV.
The data were collected with the CDF-II detector at the
Fermilab Tevatron collider and correspond to an integrated luminosity of 7.6
fb-1.
The expected number of backgrounds is 124.4 ± 10.0
for the events with the
first lepton pT larger than 20 GeV/c and the 2nd lepton
pT larger than 6 GeV/c in the CDF central region, |η| < 1.1,
while we observe 134 events in the data to find no significant disagreements.
The expected numbers of Wh and Zh events are 2.3 and 0.23, respectively
for the fermiophobic higgs
of the mass 110 GeV/c2 assuming the Standard Model cross section.
The expected numbers of events for the Standard Model higgs of the mass 160
GeV/c2 are 0.72 and 0.084, respectively.
We apply the Boosted Decision Tree technique for separating the backgrounds
and signal events to improve the search sensitivity,
then calculate limits in the
Bayesian framework using the output distributions.
We obtain observed (expected) limits on the cross section ratio to be
4.7 (4.8+1.9-1.5) for the 110 GeV/c2
fermiophobic higgs and 10.5 (10.2+4.3-3.1) for the
160 GeV/c2 Standard Model higgs at the 95% confidence level.
|
Supporting documents
Public note (7307 v4.0)
|
|
General remarks
The results should be considered preliminary until
published in a refered journal. They can be used in conference
presentations with proper reference to the CDF Collaboration.
Please
notify us
if you want to include these results in your presentation.
|
Analysis
Introduction
Our physics objective is to search for the low-mass fermiophobic higgs ans
the high-mass Standard Model higgs boson using like-sign dilepton events
produced by
qq' -> Vh -> VW*W*
-> l±l± + X, V = W, Z.
The relevant regions of the higgs mass are above 110 GeV/c2 for
the fermiophobic higgs where the branching francion of
h->W*W* supersedes that of h-> gamma gamma, and above
135 GeV/c2 for the Standard Model higgs where the branching
fraction of h->bb is overtaken by this channel.
|
Background
The physics backgrounds can be classified into reducible and irreducible
backgrounds. The reducible backgrounds are Drell-Yan,
tt,
and W + (heavy-flavor hadrons), while irreducible
backgrounds are WZ and ZZ.
Since fale-leptons and residual photon-conversions are estimated separately
using the data, we reject them founc in the MC by looking at the
generator-level information to avoid double counting.
For irreducible backgrounds, those contributions are small due to their
small production cross sections.
The residual photon-conversion backgrounds arise from an election originating
from the photon conversion with an unobserved partner track due to its
low momentum.
We estimate the backgrounds by multiplying (lepton + conversion) events by
residual photon-conversion rate (Rres). We define the rate as
Rres = (1-ε)/ ε, where ε is the conversion
detection efficiency and the Rres is shown in
here.
The efficiency is measured by comparing the conversions found in the real data
with conversion MC samples that are tuned to match with the sub-sample of
real conversions in the high-pT region of partner-tracks
where the efficiency is well known.
The residual conversion rate is parametrized by the parent photon pT.
For fake lepton, we estimate by multiplying (lepton + isolated track) events
by the fake lepton rates derived from inclusive jet samples.
The fake lepton rate (Rfake) is defined as a rate of
leptons in the jet samples relative to isolated tracks with certain
energy depositions especially in the hadron calorimeters.
We reject W and Z events in the jet samples to avoid prompt real-leptons
from electroweak processes by using MC-based subractions.
And the fake rates are parametrized by five variables: pT,
isolation(ISOcal0.4),
pseudorapidity (η), impact parameter (d0), and φ within
the wedge in the calorimeter (cph).
The fake electron rate are corrected by subtracting residual conversions
in the jet samples because we estimate the amount of residual
photon-conversion events separately as mentioned above.
These distributions are shown here.
For cross-check, we introduce three regions:
- Side-Band: 2nd lepton failing the lepton identification and passing other selections,
- Zero-Silicon: 2nd lepton with no silicon higs and passing other selections,
- Opposite-Sign: charge combination is opposite-sign and all selections are
required.
Those plots are shown in Additional Plots.
|
Boosted Decision Tree
We employ a multivariate analysis based on the Boosted Decision Tree (BDT)
technique to get more search sensitivity.
We train the BDT using fake leptons and residual photon-conversions,
because they are dominant backgrounds for our LS dilepton analysis.
We perform the training in each mass of the higgs sample independently,
using nine variables as the BDT inputs:
1st lepton pT, 2nd lepton pT,
Dilepton system pT, Missing ET, Dilepton mass,
MetSpec, HT, Number of Jets with ET > 15 GeV,
and Sphericity.
These input variables are shown here and
the outputs are here.
|
Results
We see no significant discrepancies between background expectations and
the data in the BDT outputs.
Using these output distributions, we set the ratios of the 95% C.L. upper
limits on the cross section times brancing fraction to the
prediction for the fermiophobic higgs (assuming the Standard Model
production cross sections) and for the Standard Model higgs.
The tables and plots are shown in
plot page.
|
|
| | | |
