160 likes | 289 Vues
This presentation explores the significant effects of trigger systems on the performance of flavour tagging in B-meson decay measurements. We investigate the challenges posed by different tagging efficiencies across various decay channels due to trigger influences and present solutions such as masking signal decay to obtain unbiased tagging samples. Additionally, we compare tagging performance via full simulations and analyze systematic effects like K+/K- asymmetries and B production asymmetries. Understanding these dynamics is crucial for accurate asymmetry measurements in particle physics.
E N D
Trigger and Tagging Systematics • Introduction • What is the problem?? • Affect of trigger on tagging • Tagging performance • w/o trigger • Compare full simulation with PYTHIA • Two other systematic effects: • K+/K- asymmetry • B production asymmetry Bfys Meeting, N.Tuning
Introduction • Flavour tagging is necessary for asymmetry measurement: • Was it a B, or was it aB ?? • Tag with muons, electrons or kaons: • Not 100% efficient… • ε= N tag found /N all • Not 100% correct: • ω= N wrong tag found /N tag found B b c s b K+ W+ ν e+,μ+ Bfys Meeting, N.Tuning
Introduction • Need to know tagging performance in asymmetry measurement: • Use control channels: • B0dJ/(+-) K*(K+-) • B+uJ/(+-) K+ • B0sDs-+ • But… is the tagging the same for B0dJ/(+-) K* and B0d+_ ? • Example (TDR): (NB1: =(), = ()) (NB2: ()()) =0.25 =0.35 Rutger Hierck (3.4) (2.1) Bfys Meeting, N.Tuning
PYTHIA Dependence on trigger Small dependence on b-b correlation No dependence on b-b correlation =Acc Effect of trigger on tagging • Study at generator level: • Sel: all decay part. in acc. • L0: PT,e,h >threshold • L1: • 2d-cut on IP vs PT of 2 charged particles in acc. • Overrule if 2 in acc. • Tag: • ,e,K not from signal B, and in acc. • PT ,e,K >threshold • P,e,K >threshold • IP/IPSK > threshold L1 is more efficient for B0 -+ events with tagging part. than without Bfys Meeting, N.Tuning
Effect of trigger on tagging • Tagging in full simulation suffers from signal decay. • “clones” see Massi • PYTHIA and full simulation: same trends after trigger. PYTHIA DaVinci Coincidence Clones from J/ Bfys Meeting, N.Tuning
DaVinci Effect of trigger on tagging • Full simulation (DaVinci) • Muons and kaon from B0d J/(+-) K*(K+_) affect tagging • +-: add random tags • K+: add “correct” tag • Differences B0dJ/ K* andB0dJ/K* not understood. Bfys Meeting, N.Tuning
What triggers L1? Origin of tracks, used in L1: • The track from other B is used at L1 • Increase in tagging performance after L1 Bfys Meeting, N.Tuning
The problem… The trigger changes the tagging performance differently for different channels! Therefore it will be difficult to determine ε and ω experimentally… In addition (independent of the trigger), the tagging is different per channel Change (i.e. correct) the tagging algorithm Solution: following slides (maybe…) Bfys Meeting, N.Tuning
Compare tagging efficiency • Muon channels: • Enhanced muon tagging • KS channels: • Enhanced kaon tagging? Solution: masking signal: • Reconstruct decay • Mask all hits and clusters from decay • Perform tagging See also Hans’ talk 12 November: http://agenda.cern.ch/fullAgenda.php?ida=a021827 Bfys Meeting, N.Tuning
Masking Why masking? • Get unbiased tagging before trigger, unbiased by signal decay: • Mask signal decay, and redo tagging. • Get unbiased sample after trigger, unbiased by trigger: • Mask signal decay, and redo trigger. Split triggered data in separate samples: • After masking signal: triggered YES • Else: triggered purely on signal • Else: triggered on signal+other B Unbiased sample! But small… Bfys Meeting, N.Tuning
Tagging for different trigger samples DaVinci Compare tagging performance for different samples: • TOT: could have triggered without signal tracks • Unbiased but small… • TOS: triggered without tagging B • 2 signal tracks • 1 signal, 1 other track • TOB: triggered on both B’s Bfys Meeting, N.Tuning
Masking • Consider the unbiased sample: • Mask signal decay • Redo trigger • Take events that pass • This is only unbiased if the events pass the trigger in the first place • L1: the event passes with track 2+3, but not with 1+2 … • Need to be repaired, eg: • Cut only on PT, not the 2d-cut on IP vs PT • Take OR of tracks (1,2), (2,3), (3,4), … • L0: event might pass after masking, but is vetoed in the first place… • Keep this in mind for future trigger algorithms… Bfys Meeting, N.Tuning
Two other effects… • Difference in K+ and K- reconstruction efficiencies • Difference in B and Bproduction rates Bfys Meeting, N.Tuning
K+/K- asymmetry? PDG: • Different cross section with matter: DaVinci: • Different probability of reaching TT2: • 2-3% difference in reconstruction efficiency • directly affects kaon tag Tagging kaons Bfys Meeting, N.Tuning
B production asymmetry • Two effects: • leading parton: • enhance B0d at high pT • Valence quark forms b-hadron • Not present for B0s! • “drag effect”: • enhance B0d (orB0s) at high • Remnant pulls b-hadron PYTHIA • Small Bd0 asymmetry: • Amax = 8 3‰ @ high pT, high Bfys Meeting, N.Tuning
Conclusions • Tagging is biased by signal decay products: • Solution: Mask the signal decay products before doing tagging. • (But it remains difficult…) • Tagging is biased by trigger: • Solution: Split the data in separate samples. • See: • H.Dijkstra, N.T., N.Brook: LHCb note 2003-157 Bfys Meeting, N.Tuning