Transcript Investigation on Higgs physics Group

Investigation on Higgs physics
Group
Ye Li
Graduate Student
UW - Madison
Search for a SM Higgs Boson
decaying to W pairs
• Matrix Element and Likelihood Ratio
Method
• pp-bar Collisions at 1.96 TeV
• Integrated Luminosity: 1.1 fb-1
• Standard Model Higgs Boson
• 9 Higgs Mass Hypotheses
– ranging from 110GeV/c2 to 200GeV/c2
• Channel: two leptons + missing energy
• Matrix Element Method:
– Make maximum use of the data
– Probabilities used as discriminants
– Calculated using the LO theoretical prediction
for the fully-differential cross-section
– 8 variables: two lepton momentums (6
components), two neutrino missing energy (2
quantities)
– Compared to the neural network method:
discriminant more physics
• Likelihood Ratio Discriminator
– Directly calculate the event probability:
P(xobs;α)
– Use PDF, detector acceptance and efficiency
function and detector resolution function
– Missing information of two neutrinos
• Accurate calculation requires the missing info
• Smearing due to it is larger than the detector level
smearing
– Discriminator:
• Result (take MH=160 Gev/c2 as the
example):
– 323 candidates: 286.1 +- 23.3 bg and 3.9 +0.3 signal at NNLL calculation
– The observed 95% C.L. limit is 1.3pb which is
3.4 times the Standard Model prediction while
the median of expected limit is 4.8+2.2-1.4
(systematics included)
• See plots at:
– http://wwwcdf.fnal.gov/physics/new/hdg/HWW_ME/plots.
htm
Search for the SM Higgs boson in
the ZH->llbb Channel
• Small background of mostly real Z + jets
events due to the requirement of two
leptons and a Z mass constraint
• Loosen the tight and loose lepton cuts
from standard high PT (top) analyses to
improve statistics
• 2 Artificial Neural Networks:
– One to correct two candidate Higgs jets
– The other to maintain signal efficiency and
improve signal discrimination
• Background:
– Bg w/ 2 genuine leptons and 1 b- or c-jet
(generator modeling shape)
– Bg w/ a fake lepton (use lepton fake rate)
– Bg w/ a mistaged light jet (use mistag matrix)
• Event Selection:
• Result:
– The data agrees well with the background
– Limit: