Vertex Finding in AliVertexerTracks

1
Vertex Finding in AliVertexerTracks

• E. Bruna (TO), E. Crescio (TO), A. Dainese (LNL),
  M. Masera (TO), F. Prino (TO)

2
Vertex Finding

• GOALS
• First estimation of vertex position to be passed
  to the fitter
• Calculation of dispersion s (AliVertexGetDispers
  ion() ) of tracks around the found vertex
• can be used to select good secondary vertices
• Five algorithms for vertex finding implemented in
  AliVertexerTracks
• Can be selected with AliVertexerTracksSetFinderA
  lgorithm
• fAlgo1 or 2? approximate tracks as straight
  lines and calculate the minimum-distance point
  among all the tracks at once
• fAlgo3 ? average among DCA points of all
  possible pairs of tracks treated as Helices
• fAlgo4 or 5 ? average among DCA points of all
  possible pairs of tracks approximated as straight
  lines
• Default fAlgo1 (which gives the best resolution,
  see next slides)

3
StrLinVertexFinder (fAlgo 4,5)

• Approximate tracks as straight lines (analytical
  method)
• See next slide
• Build all possible pairs of tracks
• Calculate the point of minimum distance (DCA) of
  the 2 lines
• Reject tracks with DCA gt fDCAcut
• Calculate the intersection point of the 2
  tracks on the DCA segment
• Possibility to use (fAlgo4) or not (fAlgo5)
  track parameter errors as wieghts
• Calculate the vertex position as the average of
  the intersections of all pairs of tracks
• Calculate the dispersion as the standard
  deviation of the intersection points around the
  found vertex

track 3
track 1
track 2
4
Straight Line Approximation

• Geometrical calculation of the error introduced
  by approximating the track (helix) with a
  straight line close to the primary vertex.
• Good approximation error is negligible w.r.t.
  tracks rf d0 resolution ( 100 mm for 0.5 GeV/c
  tracks)

5
HelixVertexFinder (fAlgo3)

• Same algorithm (based on DCA of track pairs) as
  StrLinVertexFinder
• Tracks treated as helices
• DCA calculation no more analytical, but based on
  minimization
• Sometimes does not converge (GetDCA stopped at
  not a minimum error)
• Reject tracks with DCA gt fDCAcut
• Tracks propagated to the DCA points
• Intersection points calculated from DCA track
  points after propagation
• Possible improvement on vertex precision and
  accuracy

track 3
Track DCA distribution from 10000 pp events
track 1
track 2
intersection points
6
StrLinVertexFinderMinDist (fAlgo1,2)

• Calculate the point of minimum distance from
  tracks
• Tracks approximated as straight lines (analytical
  method)
• Minimize the quantity D2d12d22d32 where
• All tracks at once, no pairing
• Errors sx, sy and sz used for fAlgo1 and not for
  fAlgo2
• The dispersion s is given by

track 3
track 1
d3
d1
d2
track 2
SecondaryVertex
7
Comparing the VertexFinders (I)
RMS x
RMS y

• Case of 3 charged body ( D ?Kpp ) decay vertex
  reconstruction
• The method based on the minimization of the
  distance from all tracks at once (fAlgo1)
  provides the best resolution

RMS z
8
Comparing the VertexFinders (II)

• Different finder algorithms maintained because of
  different features that can be exploited
• StrLinVertexFinder (fAlgo4,5) and
  HelixVertexFinder (fAlgo3)
• Possibility of track selection based on DCA
• Useful for rejection of displaced secondary
  tracks (mainly from strange particles) in primary
  vertex calculation when no information on the
  (x,y) beam position in the LHC fill is available
• StrLinVertexFinderMinDist (fAlgo1)
• Better resolution ? better vertex determination
• Better calculation of track dispersion, useful
  for secondary vertex selection (see next slides)

9
D mesons in ALICE central barrel

• No dedicated trigger in the central barrel ?
  extract the signal from Minimum Bias events
• Large combinatorial background (benchmark study
  with dNch/dy 6000 in central Pb-Pb )
• SELECTION STRATEGY invariant-mass analysis of
  fully-reconstructed topologies originating from
  displaced vertices
• build pairs/triplets/quadruplets of tracks with
  correct combination of charge signs and large
  impact parameters
• particle identification to tag the decay products
• calculate the vertex (DCA point) of the tracks
• good pointing of reconstructed D momentum to the
  primary vertex

10
D mesons hadronic decays

• Most promising channels for exclusive charmed
  meson reconstruction

Meson Final state charged bodies Branching Ratio Branching Ratio
D0 ?K-p 2 3.8 3.8
D0 ?K-ppp- 4 Total 7.48
D0 ?K-ppp- 4 Non resonant 1.74
D0 ?K-ppp- 4 D0 ?K-pr0 ? K-ppp- 6.2
D ?K-pp 3 Total 9.2
D ?K-pp 3 Non resonant 8.8
D ?K-pp 3 D ?Kbar0(892)p ? K-pp 1.29
D ?K-pp 3 D ?Kbar0(1430)p ? K-pp 2.33
Ds ?KK-p 3 Total 4.3
Ds ?KK-p 3 Ds ?KKbar0?KK-p 2.0
Ds ?KK-p 3 Ds ?fp?KK-p 1.8
11
Vertex finder D?Kpp
p
p
K-
bending plane
D
12
Vertex finder Ds?KKp
R. Silvestri, E. Bruna
Ds?KKp
D?Kpp

• Better resolution for D due to larger average
  momentum of daughter tracks

13
Vertex finder D0?Kppp
R. Romita, G. Bruno
xreco-xtrue cm
yreco-ytrue cm

• 4 body decay vertex
• For comparison pT integrated resolutions for 3
  body decays

RMSx RMSy RMSz
D?Kpp 120 120 127
Ds ? KKp 140 140 138
D0 ? Kppp 195 245 205
zreco-ztrue cm
14
Vertex selection D ?Kpp

• Vertex quality selection based on track
  dispersion (AliVertexGetDispersion()) around
  the found vertex
• Distribution of track dispersion for Kpp triplets
  from D decay (signal) and combinatorial triplets
  (background)
• Fraction of selected signal and background
  triplets as a function of the cut on track
  dispersion ( s ).

BLACK signal (Kpp from D) RED BKG (Kpp
combinatorics)
Accepted triplets with s lt sMAX
ZOOM
BLACK signal (Kpp from D) RED BKG (Kpp
combinatorics)
BLACK signal (Kpp from D) RED BKG (Kpp
combinatorics)
sMAX (cm)
sMAX (cm)
15
Decay vertices in AliVertexerTracks

• AliVertexerTracksVertexForSelectedTracks is the
  method for secondary vertex determination
• Arguments (NEW version presently under test)
• TObjArray (or TTree) of AliESDtracks
• Bool_t optUseFitter if kFALSE the fitting step
  is not performed and the vertex given by the
  finder is used
• Bool_t optPropagate if kTRUE after the fitter
  tracks are propagated to the found vertex.
• Track selection
• PrepareTracks just propagate tracks to the x, y
  of the primary vertex
• No rejection of tracks based on impact parameter
  (assume that user macro already did the
  selection)

16
Summary on heavy flavour vertices

• AliVertexerTracks used for D?Kpp reconstruction
  in PbPb and pp
• Vertex position is obtained with a resolution
  100 mm
• A quality parameter (the dispersion) is
  calculated and is used for vertex selection
• See Elena Brunas PhD thesis for more details
• Ongoing studies on
• D0?Kppp
• Ds?KKp
• B ?J/Y ?ee- (G.Bruno)
• Next step include kinematical constraint for the
  resonant decay chains (e.g. Ds? KK0? KKp or
  Ds? fp? KKp)
• Need (especially in pp) to remove the candidate
  secondary tracks from the primary vertex
  determination
• See Andreas talk about the fitter