130 likes | 248 Views
CLIC Higgs talks at ECFA LC 2013. Booked:. Higgs decay to b, c, and gluons @ 1.4 TeV (update). Tomáš Laštovička IoP Prague LCD-WG Analysis Meeting, April 9 th , 2013. Summary of changes.
E N D
CLIC Higgs talks at ECFA LC 2013 • Booked:
Higgs decay to b, c, and gluons @ 1.4 TeV(update) Tomáš Laštovička IoP Prague LCD-WG Analysis Meeting, April 9th, 2013
Summary of changes • Dominant background sample ee → qqvv (2096) with 126 GeV Higgs was replaced by 2201 sample with mH = 12 TeV. • ee → qqlv(2159) sample added. • 4Q background samples (qqqqνν, qqqq, qqqqll) from HHνν analysis added. • Neural net input variables added: ymin, ymax , pT of jets
Data samples • 126 GeV Higgs, 1.4 TeV, full sim/rec CLIC_SiD, γγ overlay, 1500 fb–1
Results Update UPDATED (BR: 56.1 %) (BR: 2.83 %) (BR: 8.38 %)
Results Update UPDATED
Next steps • Isolated lepton finder, requires to re-LCFI all samples. • Data samples • Increase statistics of • ee→ qqvv, ee → qq, ee → qqlv • No events available for • ee → qqll • Address e-γ and γ-γ backgrounds. • Simultaneous determination of cc and gg. • Whole NN classifier rather than cut and count. • Investigate limits of the classification. • Repeat the 3 TeV Higgs decay analysis with the 126 GeV Higgs? • Re-simulate 2025 in the SiD framework
Analysis Chain • Adapted from 3 TeV Higgs branching ratio analysis + experience from hhνν and some minor improvements SelectedPandoraPFOCollection with FastJetkT, R = 1.0 and exclusive 2 jets • Channel separation and jet true flavour tag • From quarks originating from initial Higgs • LCFI Flavour Tagging • Neural nets retrained on a fraction of qqνν sample • No significant improvement in analysis results (b-tag) when compared to 3 TeV flavour tag nets.
Neural Net Inputs • Current inputs Invariant mass and total energy of jets Distance of jets in η – ϕ plane Number of leptons and number of photons Acoplanarity Sums of jet flavour tags (b-, c-, c(b)- and b(l)) max(|ηi|) • Additional (implemented but not used at the moment) • Event invariant mass, ymin, ymax, pTmax, pTmin, ETmiss, Σ|ηi|, isolated leptons, etc… Highest NN input significance
Control Plots (for the b-channel signal) Disclaimer: does not include full set of SM backgrounds.
Neural Net Classification h → bb h → cc h → gg • b-channel (BR: 56.1 %) : is nice and easy • Note the linear histogram scale and good hνν inter-channel separation due to flavour tag. • c-channel (BR: 2.83 %) : is the most difficult one • gluons (BR: 8.38 %) : has sizeable background from hνν Disclaimer: does not include full set of SM backgrounds.