110 likes | 129 Views
Analyzing LCT efficiency using hardware and software triggers, with results above 99%. Comparison with Yuriy Pakhotin's findings and strategies to improve efficiency with correlated LCTs. Investigating triggers quality levels for optimal results.
E N D
Trying to get >99% LCT Efficiency • wrote simple analyzer module to try out Yuriy Pakhotin’s results showing >99% LCT efficiency using • Hardware trigger: • ME1 and ME3 • Software trigger: • Hits in ME1/2 xor ME1/3 and ME3/2 • Only one ALCT with 6 layers • At least one CLCT with 6 layers • Y-coordinate (wire-group) in ME1 less than Y-coordinate in ME3: select muons with IP-like angle • «Found» LCT in ME2/2: • At least 1 ALCT • At least 1 CLCT • Hit is considered as ALCT0 and CLCT0
CoincidenceEff.cc • For each event, scan over ME3/2 chambers • If found 1 valid ALCT with 6 layers and at least 1 valid CLCT with 6 layers in ME3/2/chamber, loop over ME1/2/chamber and ME1/3/chamber • If 1 valid ALCT with 6 layers and at least on valid CLCT found with 6 layers, add to denominator • Look in ME2/2/cham • If at least 1 ALCT and at least 1 CLCT found add to numerator Defines the efficiency
note • Local run 525 • Looped over 300k events (UF study loops over all ~600k events) • Bug fixed from last week
Comparisons of theta distributions Angle distributions (normalized to 1) Muons without hit in ME2/2 Muons with hit in ME2/2 nonIP-like muons 63 0.5 Result of CoincidenceEff.cc 812 Pakhotin’s results 0.9
Theta vs. global y of ALCT in ME3/2 Gap ?? ME 1/2 coincidence ME 1/3 coincidence
Results from ALCTs and CLCTs • Same as Pakhotin’s talk before theta cut • 92.3% (585/634) • Cutting on 0.4 < theta < 0.8 • 96.4% (350/363) • Cutting on 0.5 < theta < 0.8 • 99.5% (204/205) • Get ~5x less statistics than Pakhotin study (when looking at same number of total events) and need to make stronger cut to get similar results. Why? This is the point where Pakhotin’s results see ~97% efficiency This is the point where Pakhotin’s results see >99% efficiency
Same with Correlated LCTs? • Now, redo with Correlated LCTs where these LCTs in ME1 and ME3 are required to have a TMB quality of • 9 or 15: 6 layers di-strips or half-strip patterns…..see next slide • 15: 6 layers, half-strips…see slide 9
Results from correlated LCTs with TMB Quality = 10 or 15 • Same as Pakhotin’s talk before theta cut • 92.0% (625/679) • Cutting on 0.4 < theta < 0.8 • 96.2% (354/368) • Cutting on 0.5 < theta < 0.8 • 99.5% (207/208) • Same results as seen with ALCTs and CLCTs This is the point where Pakhotin’s results see ~97% efficiency This is the point where Pakhotin’s results see >99% efficiency
Results from correlated LCTs with TMB Quality = 15 • Same as Pakhotin’s talk before theta cut • 91.6% (403/440) • Cutting on 0.4 < theta < 0.8 • 96.1% (221/230) • Cutting on 0.5 < theta < 0.8 • 99.2% (117/118) • Same results as seen with ALCTs and CLCTs, but here no Di-strip patterns This is the point where Pakhotin’s results see ~97% efficiency This is the point where Pakhotin’s results see >99% efficiency
LCT Trigger Efficiency using Phi Overlap of Chambers • If you require the LCTs quality to be 15 in the efficiencies studies using chamber overlaps, you get similar results (>94%)., but we need to get this number >99%. How?? Increasing Trigger Quality Increasing Trigger Quality
Questions • Why do we get 5x less statistics than UF study? • Why do we need tighter cuts on theta to get >99% efficiency? • What can we do to get efficiency using chamber overlap >99%? Is it an edge effect? • Other questions?