Triggers, Fitters and VErtex LOcators Manchester Christmas Meeting 2010 Adam Webber 21/12/10 VELO
Hlt2Mu2Tracks 1 The trigger: L0 (hardware) Hlt1 – Inclusive (software) Hlt2 – Inclusive/Exclusive (software) The Idea: Extend on Hlt2MuTrack… B μ D/K/… π/k ν The problem: Not very good efficiency with current triggers Efficiency is split between ~3 Hlt1 and ~5 Hlt2 lines Signal Efficiency (± 0.1 %) Efficiency w.r.t. previous trigger level (%) L072.9 Hlt Hlt Rectangular cuts optimisation on several million real MinBias events and several thousand signal MC (from several different channels) L0 (40->1MHz) Hlt1 (1k->30kHz) Hlt2 (30->2kHz)
Status 2 An optimistic outlook (loose L0): ChannelEfficiency/Rate (%) (each w.r.t previous trigger level) L0Hlt1Hlt2Topo (“or” of all) Hlt2 Mu2Tracks Bd->DsMuNu Bs->DsMuNu Bd->K*MuMu Bs->JpsiPhi MB (mu= A more realistic outlook (tighter L0): ChannelEfficiency/Rate (%) (each w.r.t previous trigger level) L0Hlt1Hlt2Topo (“or” of all) Hlt2 Mu2Tracks Bd->DsMuNu Bs->DsMuNu Bd->K*MuMu Bs->JpsiPhi MB (mu=
B,B lifetime ratio 70s0 The Plan: Take the ratio of Bs and B yields in bins of lifetime Signal trigger acceptance functions cancel when we take their ratio Separate B from Bs by fitting the mass spectrum (make these plots nicer…)
2 Highs and Lows Observed: Time Expected: Happiness My
2 Highs and Lows Observed: An exponential decay followed by a rise at larger proper lifetime Time Expected: An exponential decay Happiness My
2 Highs and Lows Expected: An exponential drop in #Bs/#B followed by a rise due to cosh(ΔΓ s t) term. Observed: An exponential drop followed by a rise at larger proper lifetime More than we bargained for. From a simple fit we can measure ΔΓ s as well as the ratio of B lifetimes. “Fantastic” – we said. Time Happiness My
2 Highs and Lows Expected: An exponential drop in #Bs/#B with increasing proper lifetime. Observed: An rapid rise followed by a drop at larger proper lifetime The reality: I plotted #B/#Bs instead of #Bs/#B. Correcting this flips our plot… Time Happiness My
2 Highs and Lows Expected: An exponential drop in #Bs/#B with increasing proper lifetime. The reality: used exp(-Γ)=exp(-τ) instead of exp(-Γ)=exp(-1/τ) by accident… Time Happiness My Observed: An rapid rise followed by a drop at larger proper lifetime Conclusion: Unexpected bias in selection. Requires investigation.
VELO Resolution The Problem: Vertex resolution in data is not as good as with MC. Idea 1: Lower sensor operational voltage: V Resolution = pitch/√12
VELO Resolution The Problem: Vertex resolution in data is not as good as with MC. Idea 1: Lower sensor operational voltage: smaller V Resolution = weighted mean of several strips
VELO Resolution The Problem: Vertex resolution in data is not as good as with MC. Idea 2: Lower strip threshold. Make more multi-strip clusters. Proof of principle: Would be here if I’d gone home after dinner last night…
Merry Christmas!!! I also grew a pretty good moustache. But it had to go after some classic quotes from my girlfriend: “I think I love you a little bit less” “It makes me feel sick” and