D ?K-p p :reconstruction and perspectives for the measurement of the elliptic flow

1
D?K-ppreconstruction and perspectives for the
measurement of the elliptic flow

• Elena Bruna
• University of Torino

Terzo Convegno Nazionale sulla Fisica di Alice,
12-14 Nov 2007
2
Outline

• Exclusive reconstruction of D ? K-p p in
  Pb-Pb and pp collisions
• Selection strategy
• Results
• pp Results with more statistics w.r.t. previous
  analysis
• pp A look at the systematic errors due to the
  selection procedure
• cut variables considered distance
  primary-secondary vertices, cosqpoint
• Perspectives for the measurement of D azimuthal
  distribution in Pb-Pb

3
Selection strategy (1/4)

• Cuts on single tracks ( pT, d0rf )
• In Pb-Pb reduce the combinatorial background from
  109 to 106 triplets per event keeping 7 of the
  signal
• pT cut different for K and p (identified by the
  charge sign D?K-pp)
• Cuts on Kp candidate pairs
• K and p have opposite charge sign
• Cut on the distance d between the vertex of the 2
  tracks and the primary vertex

4
Selection strategy (2/4)

• Cuts on Kpp candidate triplets of tracks
• selection based on the products of impact
  parameters of the two Kp pairs ? 25 of BKG
  triplets rejected, all signal kept
• Cuts on the secondary vertex quality

BLACK signal (Kpp from D) RED BKG (Kpp
combinatorics)
BLACK signal (Kpp from D) RED BKG (Kpp
combinatorics)
s (cm)
5
Selection strategy (3/4)

• Four selection variables
• Distance between primary and secondary vertex
  (dPS)
• cosqpoint
• Sum of squared impact parameters s
  d012d022d032
• Max. pT among the 3 tracks pMMaxpT1,pT2,pT3

BKG
cosqpoint
SIG
d PS(mm)
6
Selection strategy (4/4)

• n-dimensional matrix procedure (n4), i.e. one
  dimension for each cut variable (dPS, pM,
  cosqpoint, s)
• Fill matrices Sijkl (signal) and Bijkl
  (background)
• Each cell ijkl contains the number of triplets
  passing the set of cuts
• dPS gt cuti AND pM gt cutj AND cosqpoint
  gtcutk AND s gt cutl

cosqpointCUT
Significance S/?(SB) for 2ltpTlt3
GeV/c (normalized to 107 PbPb events)

• Advantages
• Global maximization
• Easy to add new selection variables

d CUT(mm)
7
Results D?K-pp in Pb-Pb (1)

• Significance S/vSB (normalized to 107 ev.) vs pT
• Relative statistical error (sS/S 1/vS) on D vs
  pT (in 107 ev.)
• S/ev10-3 , B/ev 10-4

8
Results D?K-pp in Pb-Pb (2)

• Selection efficiency and pT spectra
• pT integrated e (D) 1.5 (Ideal PID), 0.6
  (Real PID), 1 (no PID)

9
Results D?K-pp in pp (1)

• Signal statistics improved by a factor 2 w.r.t.
  previous analysis
• Significance and relative statistical error vs.
  D pT
• S/ev5 10-6 , B/ev 5 10-6
• Significance and relative statistical error
  (1/?S) normalized to 109 pp Minimum Bias events

10
Results D?K-pp in pp (2)

• Selection efficiency and pT spectra
• Selection efficiency (pT integrated) e 4 (NO
  PID)

11
Results Pb-Pb vs pp
(PID)
(no PID)

• sS/S the same for Pb-Pb and pp for 2ltpTlt8 GeV/c

(PID)
(no PID)
S/vSB in one year
12
Studies on D Invariant Mass

• Signal Inv. Mass distribution fitted with a
  Gaussian function, in different D pT ranges
• Signal normalized to 109 pp randomly generated
  according to a Gaussian function
• The same for the Bkg triplets, but a polinomial
  function was used

13
Invariant mass distributions
Statistics enhanced in order to match 1 year of
data taking
2ltpTlt3 GeV/c
0ltpTlt2 GeV/c
pTgt5 GeV/c
3ltpTlt5 GeV/c
14
Systematic errors on pp events

• Two samples of events
• 1) pp MB (85) pp with charm (14) and beauty
  (1) events with c?D?Kpp (divided in 4
  subsamples of pThard)
• extract the D number with Gaussian fit to the
  invariant mass distribution Nsel
• 2) only events with c?D?Kpp (divided in 4
  subsamples of pThard)
• extract the selection efficiency e
• D reconstructed NrecNsel/e
• Nrec calculated for different values of the cut
  variables

15
Cut on dprim-sec
Nsel vs d (norm to 109)
COUNT From FIT
2ltpTlt3 GeV/c
Nrec vs d (norm to 109)
COUNT From FIT
e vs d
16
Cut on cosqpoint
Nsel vs cosqpoint (norm to 109)
COUNT From FIT
2ltpTlt3 GeV/c
Nrec vs cosqpoint (norm to 109)
COUNT From FIT
e vs cosqpoint
17
Perspectives for D elliptic flow

• GOAL Evaluate the statistical error bars for
  measurements of v2 for D mesons reconstructed
  from their Kpp decay in Pb-Pb collisions
• v2 vs. centrality (bcollision impact parameter)
• v2 vs. pT in different centrality bins
• HOW fast simulation to generate
• ND (DpT,Db) with an angular distribution
• dND/dj v2D,in cos 2(j-YRP)
• For each D an event made of Ntracks is
  superimposed
• dN/dj v2ev,in cos 2(j-YRP)
• Inputs of the simulation ND (DpT,Db) , v2D,in,
  Ntracks ,v2ev,in
• Outputs v2D measured.

18
v2 vs. pT
6ltblt9 fm

• Large stat. errors on v2 of
• D ? Kpp in 2107 MB events
• How to increase the statistics?
• Sum D0?Kp and D?Kpp
• Sufficient for v2 vs. centrality
• Semi-peripheral trigger
• v2 vs. pT that would be obtained
• from 2107 semi-peripheral
• events (e.g. 6ltblt9 fm)

MB trigger
Semi-peripheral trigger
19
Summary

• D in PbPb and in pp
• The results show that the D ? K-pp analysis in
  Pb-Pb and pp is feasible with a pretty good
  Significance.
• The selection strategy based on n-dimensional
  matrices has been developed.
• Recent updates
• Significance in pp improved with enhanced
  statistics
• Invariant mass studies give the expected spectra
  in 1 year of data taking
• No evident systematic error on the selection
  procedure from a first look more statistics
  needed
• D v2
• Large stat. errors on v2 of D ? Kpp in 2107 MB
  events
• How to increase the statistics?
• Sum D0?Kp and D?Kpp
• Semi-peripheral trigger
• Under study the methods for background subtraction

20
Backup slides
21
Initial Nsel statistics (vs dprim-sec)
2ltpTlt3 GeV/c
0ltpTlt2 GeV/c
3ltpTlt5 GeV/c
pTgt5 GeV/c
22
Initial Nsel statistics (vs cosqpoint)
0ltpTlt2 GeV/c
2ltpTlt3 GeV/c
3ltpTlt5 GeV/c
pTgt5 GeV/c
23
Initial Bkg statistics with M-Minvlt1s (vs d)
2ltpTlt3 GeV/c
0ltpTlt2 GeV/c
3ltpTlt5 GeV/c
pTgt5 GeV/c
24
Results from RHIC

• The Non-Photonic electrons RAA shows interesting
  results
• Suppression is very large when compared to the
  expectation from radiative energy loss
• Other possible mechanisms?
• Collisional EL, resonances, in medium
  fragmentation
• First hint of strong charm v2 for pTlt2 GeV/c
• Very important, need to disentangle charm from
  bottom

PHENIX nucl-ex/0611018 STAR nucl-ex/0607012
25
Results scaled to a lower multiplicity scenario
for Pb-Pb

• Results presented so far for Pb-Pb based on
  dNch/dy6000
• Ntracks7000
• BKG/evN(N-1)(N-2)/3! 6x1010
• Extrapolations from RHIC results seem to favour a
  lower multiplicity scenario, dNch/dy2000
• Ntracks7000/3 2300
• BKG/ev 2x109
• Lets consider the D pT interval 0ltpTlt2 GeV/c
• The results in the highest multiplicity scenario
  (dNch/dy6000) are not satisfactory
• BKG/ev is downscaled by a factor 30
• SIG/ev not rescaled
• S/vSB (normalized to 107 ev.) 10 in case of
  Real PID it is possible the study of the low-pT
  spectra

26
Systematic uncertainities on D measurements

• Acceptance, reconstruction and particle
  identification efficiencies (10)
• Centrality selection (7)
• Parameters of the Woods-Saxon profile and nuclear
  density (5)
• Nucleon-nucleon inelastic cross section (5)
• Branching ratio D?K-pp (3.6)
• Feed-down from beauty
• Nc?D ND - Nb?B ? D
• Contamination
• K Nb?B ? D/ Nc?D 4

27
Feed-down from beauty

• D from B are more displaced
• The cut on distance between primary to secondary
  vertex increases the fraction of selected D
  coming from B decay

Contamination K vs d
d1000mm ? K 10
Histograms normalized to the same area