1. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 1 Andrea Dainese (INFN LNL) Hadronic Charm: Introduction & D 0  K 

2. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 2 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation Physics (& comp. with theory)

3. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 3 Heavy-quark production at the LHC Baseline predictions: NLO (MNR code) in pp + binary scaling (shadowing included for PDFs in the Pb) ALICE baseline (PPR) for charm / beauty: system :  s NN : Pb-Pb (0-5% centr.) 5.5 TeV p-Pb (min. bias) 8.8 TeV pp 14 TeV 4.3 / 0.27.2 / 0.311.2 / 0.5 115 / 4.60.8 / 0.030.16 / 0.007 0.65 / 0.800.84 / 0.90-- Theoretical uncertainty of a factor 2-3 MNR code: Mangano, Nason, Ridolfi, NPB373 (1992) 295.

4. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 4 MC Sample for Tools Development 5×10 6 pp events at 14 TeV (PYTHIA “PhysicsRun” cocktail): 85% pp min. bias w/o heavy quarks ~14% pp (MSEL=1) with charm, p t hard -binned ~1% pp (MSEL=1) with beauty, p t hard -binned Fast production: full ITS, TPC param, no TOF, no TRD Generated via LCG in Legnaro, Torino, CNAF in April06

5. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 5 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation Physics (& comp. with theory)

6. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 6 Event reconstruction (RAW  ESD  AOD) Reconstruction in the barrel  tracks Standard AOD (Analysis Object Data), extracted from ESD (Event Summary Data):

7. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 7 Primary Vertex B e X d0d0 rec. track The key issue for HF reco: track impact parameter Distance of closest approach to the primary vertex At “high” p t : d 0  c   HF decay tracks displaced by ~100  m  ALICE ITS with SPD < 60  m (r  ) for p T > 1 GeV/c BA, CT, LNL, PD

8. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 8 Track impact parameter measurement in pp collisions Track impact parameter resolution:  d0 =  vtx   track Vertex reconstructed from tracks (LNL, To) Bias (underestimate of d 0 ) if the considered track is used for vertex fit:  of d 0 distribution for primaries  m  bias!  of d 0 /error(d 0 ) distr. for primaries 90% of tracks from D decay used to find primary vertex  primary vertex has to be reconstructed for each D candidate excluding its daughters D-decay

9. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 9 Effect of ITS misalignment on d 0 resolution Impact parameter resol:  track = a scatter /p t  b meas  c misalign Effect studied with full simulation of exptected initial (full+) and residual (after realignment) misalignments See talk tomorrow Effect on D 0 measurement studied (see later) null residual full full+

10. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 10 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation Physics (& comp. with theory)

11. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 11 Analysis scheme (ESD/AOD ➔ AOD) Standard AOD + HF vertices (15-20% in size) 2-prong HF 3-prong HF... ESD or

12. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 12 AliAnalysisVertexingHF single-track cuts on quality, p t (d 0 ?) [common cuts for all analyses] build all (+,-) pairs and compute secondary vtx create AliAODRecoDecayHF2Prong apply reco cuts store D 0 loop on all tracks (+ & -): build triplets, create AliAODRecoDecayHF3Prong, apply reco cuts (common for the 3 particles?) loop on all tracks (+ & -) … create AliAODRecoDecayHF … store D +, D s +,  c store D 0 for D +,D s +,  c +,D *+ for D 0  K  for D 0  K , J/  (from B)  ee store J/  tight D 0 mass cut loop on all tracks for D* + candidates store D *+ Bari, Legnaro, Torino

13. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 13 x pull x resolution [  m] Secondary vertex reconstruction Two different vertexers available classical vertexer (used also for primary vertex reco): finder + fast fitting algorithm (LNL-To) Kalman-filter vertexer, with possibility to apply topological & mass constraints (GSI), under test Possibility to remove D 0 decay tracks from primary vertex  unbiased (or less biased) estimate of impact parameters

14. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 14 Set of classes for storage and analysis of candidates AliVParticle AliAODRecoDecay AliAODRecoDecayHF AliAODv0 AliAODRecoDecayHF2ProngAliAODRecoDecayHF3Prong D +,D s,  c,...D 0,J/  from B,... AliAODRecoDecayHF4Prong D0D0 PWG3 PWG2

15. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 15 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation

16. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 16 Selection Variables & PID Variables: DCA: distance of closest approach between the two tracks cos  *: cosine of decay angle d 0 xd 0 : product of the two tracks r  impact parameters cos  pointing : cosine of pointing angle 3 PID “modes” considered (TOF): keep all pairs (no PID) keep only pairs with a K reject only ( ,  ), (K,K)

17. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 17 Extraction of signal raw yield: Invariant mass analysis Here, 5 million pp events (1/200 of 1-year’s statistics) p t > 0 “PPR cuts”

18. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 18 Extraction of signal raw yield: Invariant mass analysis Beware of reflections: Reflections Combinatorial no PID RS/TS~0.30 RS/TS~0.25 RS/TS~0.20RS/TS~0.15 PID: ( ,  ), (K,K) rejected

19. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 19 Extraction of signal raw yield: Invariant mass analysis Two possible approaches: direct fit w/o bkg subtraction (only if large S/B, > 50%?) background subtraction with event mixing + fit Direct fit tried: exponential + gaussian fit side-bands with expo: initial expo parameters fit everything with expo + gauss with constraint on total integral ------------ From Fit ---------------------- Total S integral (+- 3  ) = 39.7 +- 9.3 S/B (+- 3  ) = 85% SGNC as S/  S (+- 3  ) = 4.3 (58 for 10 9 evts) SGNC as S/  S+B (+- 3  ) = 4.3 ---------- From Simulation ------------ Total S integral (+- 3  ) = 39.5 S/B (+- 3  ) = 90% SGNC as S/  S+B (+- 3  ) = 4.3 (59 for 10 9 evts)

20. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 20 0<p t <1 GeV/c Significance maximization Optimize cut values in p t bins: maximize significance in cut- variables space (e.g. 4D space) Example: 2<p t <3 GeV/c3<p t <5 GeV/c5<p t <8 GeV/c A.Rossi

21. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 21 Results Checked performance w/o PID: not much worsening, but have to tighten cuts (larger systematic errors?) pp (10 9 evts) PPR pp (10 9 evts)

22. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 22 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation Physics (& comp. with theory)

23. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 23 Corrections & Errors: Beauty feed-down (~10%) Methods: 1.Monte Carlo with state-of-the-art pQCD input 2.use measurements from ALICE (from single , then e) Systematic error: uncertainty on b-bbar cross section from theory (1) or data (2). For (1), the syst. error estimated to be ~8-10% according NLO pQCD

24. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 24 Corrections & Errors: Acceptance, Reco & Selection Efficiency Method: Embedding of MC signal in real events, and calculate all corrections in one go (selected  in-acceptance). Average correction in {p t, y} grid  use ALICE common correction framework (  A.Mastroserio) Needed tuned MC (good descr. of tracking effs and resols). d 0 resol. is crucial, must be evaluated from data vs p t, , PID, N ITSclusters Systematic error: - compare weights in different runs, and with the two field orientations (+z and -z) - check stability of extracted yield VS variation of cuts - repeat weights calculations (MC) with different sets for alignment corrections

25. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 25 Effect of misalignment on S/B and significance (S/  S+B) Residual misalignment: negligible effect “Full” misalignment: 15-20% worsening of statistical errors

26. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 26 Evaluation of d 0 res. from data Evaluation of d 0 resolution: d 0 distribution is dominated by primary particles for |d 0 |<d 0 MAX Fit in this range provides the resolution, to be then compared with that in AliRoot

27. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 27 p t  0.4 GeV/c primaries all Evaluation of d 0 res. from data p t  1.1 GeV/c p t  4.5 GeV/c A.Rossi

28. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 28 Evaluation of d 0 res. from data Fit of all tracks’ d 0 distr. in the range ~ (-3 , +3  ) allows to extract the d 0 resolution Resolution is a convolution of track position resolution, primary vertex resolution, misalignment effects  devise method to separate track and vertex contributions Study other (complementary) methods e.g. using cosmic muons crossing all ITS  d 0 resolution at high p t  d0 =12  m: OK! no misal. ITS

29. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 29 Invariant mass analysis and significance maximization (in bins of p t, y,   RP …) Analysis Scheme Event reconstruction (RAW  ESD  AOD) Charm “production” (ESD/AOD  AOD for D’s) D 0 signal selection (using PID and geom/kinem cuts) Correction for efficiencies, acceptance, BR Cross section normalisation Physics (& comp. with theory)

30. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 30 Calibrating the probe (pp,  s = 14 TeV) D 0  K  measured at 14 TeV, then scaled by pQCD to 5.5 TeV Expected sensitivity in comparison to pQCD: 1 year at nominal luminosity (10 9 pp events) Detector configuration: ITS + TPC + TOF ~1 month stat. err. vs # events

31. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 31 Charm E Loss : R AA 1 year at nominal luminosity (10 7 central Pb-Pb events, 10 9 pp events) E loss calc.: Armesto, Dainese, Salgado, Wiedemann

32. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 32 Heavy-to-light ratios 1 year at nominal luminosity (10 7 central Pb-Pb events, 10 9 pp events) probes colour-charge dep. of QCD energy loss probes mass dep. of QCD energy loss E loss calc.: Armesto, Dainese, Salgado, Wiedemann

33. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 33 Summary Prepared software tools for reconstruction of (most) hadronic charm decays (+J/  ) in one go (Ba - LNL/Pd - To) will (soon) run on ESD and on AOD will (soon) allow to choose between two vertexers to be tested on the Grid Track impact parameter measurement is crucial for hadronic charm (esp. in Pb-Pb) studied possible systematics on this quantity (primary vtx, misalignment...) D 0  K  : run a realistic analysis (sgn+bkg together) on 5M pp events tried a first version of invariant mass fit to extract raw yield  warning: signal reflection developed multi-dimensional SGNC maximization method results (signal SGNC) compatible with PPR ones set up procedure to apply corrections

34. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 34 EXTRA SLIDES

35. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 35 First pp data: machine scenario  * ~ 10 m (will be 0.5 m for Pb-Pb) Bunch spread: transverse  bunch (  *) ~ 70  m, long.  bunch ~ 7.5 cm Vertex spread: transverse  vertex =  bunch /  2~50  m, long.  vertex = 5.3 cm Luminosity: (  1/   vertex ) ~10 30 cm -2 s -1 (70 kHz for  pp = 70 mb) Revol’s scenario: 20x10h ~5-10x10 7 min. bias events Subsystems directly used for D 0 analysis: ITS, TPC, TOF Assumption: TPC, full ITS, half TOF (-0.9<  <0.9,  =  ) First pp data: detector scenario

36. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 36 Candidate cuts (1)

37. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 37 Corrections & Errors: Acceptance, Reco & Selection Efficiency Selection efficiency: from selected to reconstructed signal Reconstruction eff.: from reconstructed to in-acceptance Acceptance: from in-acceptance to dN/dy at y=0 central Pb-Pb

38. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 38 Beauty feed-down Feed-down can be up to 15% after selection Can use upper cut on |d 0 | to control it To subtract it: 1)use simulation with state- of-the-art pQCD c and b predictions for LHC 2)later, use b cross section measured at LHC 3)later, use d 0 of D 0 to estimate feed-down

39. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 39 Simulation & Reconstruction Detectors: full transport in ITS, track references at TPC R in Primary Vertex:  * = 10 m   x =  y = 50  m, as expected for the 2007 run Reconstruction: TPC tracking response parametrized (old parametrization from 2002  conservative efficiencies and resolutions) Slow Points in ITS z of vertex from SPD Kalman in ITS with AliITStrackerV2 full vertex reco with tracks NOT INCLUDED (to be done before analysis)

40. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 40 TPC parametrization full tracking TPC param full tracking TPC param full tracking TPC param TPC param / full tracking

41. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 41 Reco & Sele TO-DOs TO DO for single-track cuts (p t and d 0 ): - estimate CPU time, signal and background efficiency vs cut values - estimate cut values compatible with “final” candidate cuts TO DO for secondary vertex: implement vertex fitting with errors and  2 (as done for primary) TO DO for inv. mass analysis: prepare fit procedure/code; take into account reflections in fit; background subtraction IN PROGRESS IN PROGRESS, talk on Friday

42. Terzo Convegno sulla Fisica di ALICE, LNF, 13.11.2007 Andrea Dainese 42 Corrections & Errors TO-DOs TO DO for beauty feed-down: - define procedure to generate MC with B -> D 0, and reweight results according to a given pQCD d  b /dp t - define procedure to use ALICE single  measurement to infer B production in the barrel TO DO for systematic errors: - define procedure for embedding (how many {p t,y} bins? needed MC stats? which real events? vs dN ch /dy? vs z vertex ?) - define procedure to evaluate d 0 resol. from data IN PROGRESS