|
We search for an anomalous boson which decays into dijets in top decays as hadronic decaying W does in the SM ttbar
lepton+jets decays. In the MSSM, this channel is predicted for charged Higgs in low tan β around unity. This tree level particles
end up with one lepton, missing Et, and 4 jets in detector. This 4 jets are called leading jet.
(Click here for eps file)
|
|
This plot shows how dijet mass distributes in exact 4 jets (left) and with additional jet (right) in ttbar events in simulation.
Each jet is traced back to the mother quark in generate level and identified. Good identification requires
ΔR(jet, quark) less than 0.4, so NoGood4jets means that at least one jet out of final 4 jets is not identified.
Even though we require two b-tagged jets and reconstruct dijet mass with un-tagged jets, there exist b-jet contaminations.
2hjets is both jets are decayed from 120 GeV/c2 Higgs boson. h+blep and h+bhad
means that one jet comes from Higgs but the other jet is actually b-jet. We can clearly see that the real Higgs mass (magenta)
has broader resolution and this is caused by losing energy by final state radiation (FSR) jet from Higgs. Nearly 50% of total events
have one or more extra jets in final state.
(Click here for eps file)
|
|
This plot shows how the Higgs mass in >4 jets ttbar events is recorvered by adding nearby FSR-jet (5th) to mother jet
(black: use two jets only, red: after adding extra jet to the closest leading jet). We consider the extra jet is a FSR-jet from the closest leading
jet if ΔR (extra jet, closest leading jet) is less than 1.0. Note that we use only most eneretic extra jet if there are more than one extra jets.
From this procedure, the histogram mean is close to the true mass (120 GeV), 103.3±21.8 GeV to 105.7±20.8 GeV, and the
entries in the signal mass window, 100 GeV < M(H+) < 140, increased by 7.4%.
(Click here for eps file)
|
|
A binned likelihood function is constructed using the probabilities, PiW, PiH,
and Pibkg, from mass templates. Fit parameters are Nttbar, Br(t→H+b),
and Nbkg. The number of backgrounds are well-known, so it is Gaussian constrained in the fitter.
The acceptances of W+ (Aw) and H+ (Ah) are calculated
from Pythia Monte Carlo simulation.
|
|
A template consists of H+ mass, W+ mass, and non-ttbar backgrounds shape.
(Click here for eps file)
|
|
Mass templates : H+ mass distribution (60 GeV/c2 to 150 GeV/c2) compared to the W+ mass.
(Click here for eps file)
|
|
The likelihood shape is plotted as a function of Br(t→H+b). Only for the positive side (physical) branching ratio,
the likelihood value is integrated up to 95% of total area. Then the projection onto branching ratio axis is set to the upper limit
on the Br(t→H+b) at 95% Confidence Level of this likelihood fit.
(Click here for eps file)
|
|
Upper limit branching ratio using null-Higgs pseudo experiments. No systematic uncertainty is considered in the limit.
(Click here for eps file)
|
|
Output branching ratio shift due to each systematic, Δ Br(t→H+b), is set to systematic uncertainty. That uncertainty is fitted as a
linear function of input Br(t→H+b). This plot shows the individual systematic uncertainty.
(Click here for eps file)
|
|
Overall systematic uncertainty is calculated by square-root-sum of individual systematic uncertainties above. This uncertainty is put as an width of Gaussian for likelihood smearing.
(Click here for eps file)
|
|
Likelihood marginalization. The systematic uncertainty (Δ(x')) in each
input branching ratio (x') is put as a width of Gaussian to smear the likelihood value at x'.
|
|
From the first likelihood fit, red likelihood shape is on hand. After smearing the likelihood shape with the systematic
uncertainties in the Gaussian (See the Gaussian formula above ↑)
the likelihood is smeared out to the black distribution. The upper limit on the Br(t→H+b) @ 95% C.L. is
increased (red→black) accordingly.
(Click here for eps file)
|
|
After likelihood smearing with systematic uncertainties, the uppre limit Br(t→H+b) @ 95% C.L. is increased.
(Click here for eps file)
|
We study that how the other decay channel would affect the upper limits
on B(t → H+(→ cs-bar) b). Considering any generic
charged bosons X, X → ud-bar is also possible decay channel. For
this study, we generated Pythia Monte Carlo with X → ud-bar in top
decays for 60, 100, 120 GeV X. The resolution of ud-bar invariant mass
is narrower than the cs-bar invariant mass as compaired below, so that
the using ud-bar mass template gives a better upper limits.
However, we take the conservative limits using
cs-bar invariant mass, which is shown top of this page.
(Click for eps page. To enlarge the gif plot, use "View Image")
