B Physics with the initial detector in CMS
B Physics with the initial detector in CMS N. Magini University & INFN Firenze on behalf of the CMS collaboration PHYSICS AT LHC (LHC-Praha-2003) Prague, July 6 - July 12, 2003
B Physics at CMS B-decays program Rare decays CP violation B 0 s Mixing b production at LHC Starting luminosity: 2 x 1033 cm-2 s-1 s = 0. 5 mb about 106 b pairs/sec only 100 ev/sec on tape for ALL interesting physics channels The main challenge is the trigger strategy Some benchmark channels analysed – BS J/ KK – BS DS ( KK) Results from DAQ-TDR (CERN/LHCC 2002 -26) Praha 2003 B Physics with CMS 2
Trigger & DAQ architecture Two level trigger: Lvl-1 and HLT 40 MHZ 100 KHz 100 Hz Several staging scenarios possible. Each slice allows 12. 5 KHz Praha 2003 B Physics with CMS 3
Trigger & DAQ architecture Two level trigger: Lvl-1 and HLT 40 MHZ 50 KHz 100 Hz Several staging scenarios possible. Each slice allows 12. 5 KHz Praha 2003 4 DAQ slices at start-up => 50 KHZ B Physics with CMS 4
Triggering B events at Level 1 Integrated muon rates above threshold < 2. 1 Output of Level 1 Muon Trigger is dominated by b - and c-quark decays up to 25 Ge. V Praha 2003 B Physics with CMS 5
Level 1 Muon Trigger rates Low Lumi 16 k. Hz DAQ 3. 6 k. Hz for , < 2. 1 Total output rate: 50 k. Hz x 1/3 “safety” => 16 k. Hz allocated Lvl-1 thresholds optimized for discovery physics -> high-pt processes are selected 14 ; 3, 3 2. 7+0. 9=3. 6 k. Hz e. W =90% e. Z =99% e. Bs mm =15% b-jets are selected mainly by 1 and 2 trigger Praha 2003 B Physics with CMS 6
High Level Trigger 40 MHZ • • • • • 100 KHz Implemented on a PC farm: 300 msec/ev on 1 GHz PIII (± 50% uncertainty) Same software as for offline 100 Hz Low lumi A total of 30 Hz are allocated to muons 25 Hz single 5 Hz di-muons 30 Hz In addition non-isolated muons are rejected Praha 2003 B Physics with CMS 7
HLT single muon stream content W 25 Hz b/c Only ~5 Hz are b/c events: about 5 · 107 ev/year at low lumi Not enough for B phyiscs studies involving rare decays (typical Branching Ratios <10 -4 ) Praha 2003 B Physics with CMS 8
Using the Tracker at HLT • Previous results can be improved by using the Silicon Tracker during the High Level Trigger • HLT reconstruction has to be fast but does not need to be as global and precise as the offline one • HLT Tracking can therefore be: – Regional • Restricted to some phase-space region defined from external Lvl-1 information (e. g. a cone around muon/jet direction or the set of tracks coming from a vertex or above a PT threshold ) – Partial • Stopped after a number of hits have been assigned to the track – Conditional • Stopped when enough resolution is reached or on other condition Praha 2003 B Physics with CMS 9
Primary vertex reconstruction y y 10 -30 s = 26 m 7 cm 4 cm x z 0 z • Hit pairing with a straight line in rz – IP 1 mm – PT > 1 Ge. V • Matching with 3 rd layer track candidate • PV candidate if 3 track cross z-axis • Signal vertex from PT and Ntracks PV efficiency >95% in high luminosity events Average time: 50 msec / 1 GHz Praha 2003 B Physics with CMS 10
Partial reconstruction Sufficient track parameter resolutions with 4 - 5 hits half the time for full reconstruction needed IP Time vs hits in 100 Ge. V b-jets Pt Full tracker performance Praha 2003 B Physics with CMS 11
Benchmark channels Decays into muons (di-muon Lvl-1 Trigger) : • BS J/ KK Fully hadronic channel (Single muon Lvl-1 Trigger from other B +X) : • BS DS ( KK) Praha 2003 B Physics with CMS 12
Bs FCNC b->s or b->d only at loop-level in SM => BR(Bs mm)=(3. 5± 1. 0)x 10 -9 • Lvl-1: 2 PT>3 Ge. V, =15. 2% • HLT strategy: – Select pixel seeds with PT > 4 Ge. V in - region around trigger ’s – Conditional tracking: - stop if pt<4 Ge. V/c @ 5σ or Nhit=6 or σ(pt)/pt<0. 02 1. Bs reconstruction if only 2 track candidates with opposite charge in 150 Me. V window 2. Vertex: c 2 < 20 and dr > 150 m Average CPU Time = 240 msec / 1 GHz CPU Praha 2003 B Physics with CMS 13
Bs Mass resolution Full Tracker HLT s = 46 Me. V s = 74 Me. V Lvl-1 HLT Global Events/ 10 fb-1 Trigger Rate 15. 2% 33. 5% 5. 1% 47 <1. 7 Hz Offline analysis results (hep-ph/9907256), using SM BR=3. 5 x 10 -9 (Lvl-1 trigger in | |<2. 4 instead of | |< 2. 1) 10 fb-1 => 7 signal events with <1 background 5 s observation with 30 fb-1 and analysis could be perfomed at high lumi too Praha 2003 B Physics with CMS 14
Bs J/ BR(Bs J/ )=(9. 3± 3. 3)x 10 -4 J/ ℓ+ℓ- (BR≈6%) K+K- (BR≈49%) CP violation weak phase s= 2 dg= 2 2 SM predicts s ~O(0. 03) HLT strategy: • Lvl-2: reconstruct muons as for Bs q 100 Me. V window around J/ mass, c 2 < 20 and dr > 200 m • Rate = 15 Hz, <t>~260 ms 90% of Lvl-2 J/ are from b’s Lvl-3: reconstruct and Bs • Regional tracking around J/ direction • <t>~800 ms Praha 2003 B Physics with CMS 15
Bs J/ Lvl-1 Lvl-2 Rate Lvl-3 Rate Events/ 10 fb-1 16. 5% 13. 7% 14. 5 Hz 8. 7% <1. 7 Hz 83800 HLT mass resolutions s = 46. 5 Me. V Praha 2003 J/ s = 2. 2 Me. V B Physics with CMS s = 22. 4 Me. V Bs 16
Bs J/ ℓ+ CERN-2000 -04 K+ ℓ- K- Angular analysis of Bs->J/yf->mm. KK on a sample of 600 K yields: ΔΓs s(xs=20) Value 0. 15 xΓs 0. 04 Error 8. 0% 0. 014 0. 03 Error on ΔΓs vs stats s(xs=40) 30 fb-1 ΔΓs /Γs ~ 15 % d s(xs=20) ~ d s(xs=40) ~ Praha 2003 B Physics with CMS 0. 025 rad 0. 05 rad 17
Bs Ds ( KK) BS-BS mixing: Dm. S 14. 4 ps-1 @95%CL P(b Bs) x Br(Bs DS K K ) ~ 5 x 10 -5 • • L 1 1 trigger: PT>14 Ge. V, R=3. 2 KHz HLT strategy: - - pixel seeds in full acceptance and Primary Vertex - Partial tracking: 2 pixel and 1 strip hits - Topological cuts: DR(KK)<0. 3, DR( )<1. 2, DR(DS )<2. 0, D (BS, )>0. 6 - Kinematical cuts: PT( )>2 Ge. V, PT(DS)>4 Ge. V, PT(BS)>5 Ge. V - Mass reconstruction: DM <15 Me. V, DMDs<75 Me. V, DMBs<270 Me. V HLT efficiency 9% - <t> = 640 msec Praha 2003 B Physics with CMS 18
Bs Ds ( KK) Mass resolutions (only 3 hits are used) Ds s = 5 Me. V s = 25 Me. V Bs s = 95 Me. V 1 year low luminosity (20 fb-1): Lvl-1 1 KHz HLT 5 Hz 300 -400 signal events => Dm. S up to 20 ps-1 1000 events are needed to test allowed SM range: Dm. S 26 ps-1 @99%CL Can not be run on full Lvl-1 2. 7 KHz because of large output rate: still room to improve Praha 2003 B Physics with CMS 19
High Level Trigger table for low lumi Trigger Threshold ( =90 -95%) (Ge. V) Indiv. Rate (Hz) Cumul rate (Hz) 1 e, 2 e 29, 17 34 34 1 g, 2 g 80, (40*25) 9 43 1 , 2 19, 7 29 72 1 t, 2 t 86, 59 4 76 Jet * Miss-ET 180 * 123 5 81 1 -jet, 3 -jet, 4 -jet 657, 247, 113 9 89 19 * 52 1 90 237 5 95 10 105 e * jet Inclusive bjets Calibration/ other Praha 2003 Where is B – physics ? Bandwith for B-physics at LHC start-up will depend on: • Luminosity – Lower luminosity at startup => higher B-bandwith • Background conditions – If we are lucky, the safety factor of 3 may have been overestimated (of course, we might also be unlucky …) ü In addition a possible strategy is to introduce B-triggers as the luminosity drops during the fill B Physics with CMS 22
Conclusions ü CMS is a competitive detector for B-physics, even if it is not designed specifically for it ü Use of Fast Tracking is fundamental for B-decay selection in the High Level Trigger ü A proof of concept has been given on few benchmark channels in the DAQ TDR and on other interesting channels ü We should profit from LHC starting conditions : Low luminosity for a while lots of B physics ü In the next years : the Physics TDR will address in deeper detail the CMS B-physics potential, turning the attention from HLT selection to a detailed study of offline analysis Praha 2003 B Physics with CMS 24
- Slides: 21