AOD Discussion for PWG2

Recent AOD Developments
Markus Oldenburg
19th of September 2007, CERN
M. Oldenburg
September 19, 2007, CERN
1
Overview
•
•
•
•
Common base classes for AODs and ESDs
Current content of the standard AODs
Recent developments for stored vertices
Outlook
M. Oldenburg
September 19, 2007, CERN
2
Common base classes
AliVEvent
AliESDEvent
AliAODEvent
• access to
containers got
standardized
• common getters
and setters
AliVHeader
AliESDHeader
AliAODHeader
AliVParticle
AliExternalTrackParam
AliAODTrack
…
AliESDtrack
M. Oldenburg
September 19, 2007, CERN
3
Common base classes for the analysis
• analysis based on the task framework possible for ESDs
and AODs
– AliTagAnalysis takes “ESD” or “AOD” as a parameter
• analysis based on TSelectors/AliSelectors possible
(Gabriel Barros)
– AliAODSelector, AliESDSelector with a common base class
AliVSelector
– final tests ongoing
– to be committed to CVS soon
M. Oldenburg
September 19, 2007, CERN
4
Current content of the standard AOD
AliAODEvent
• contains an (extendable) TList
TClonesArray of tracks
AliAODTrack
TClones array of vertices
AliAODVertex
AliAODCluster
TClonesArray of clusters
(EMCAL hits, PHOS hits, …)
AliAODJet
TClonesArray of jets
AliAODTracklets
M. Oldenburg
TClonesArray of SPD tracklets
September 19, 2007, CERN
5
Current status of stored AliAODVertex’s
• vertices stored in AliAODVertex are good for navigational
purposes
– they contain a space point which connects to
• a parent particle
• an array of daughter particles
• so far each track is associated to one and only one vertex
– primary, V0, kink, cascade
• This is problematic in case of ambiguous situations / signals
above background (e.g. V0s).
• Solution: store ‘all’ possible vertex candidates (after some
reasonable filtering)
M. Oldenburg
September 19, 2007, CERN
6
Secondary vertices
• PWG2 and PWG3 need access to the ESD to fill their
branches of vertices correctly.
• They’d like to point back to the vertices stored in the
AliAODVertex array.
– duplication of information will be avoided
• This means:
• AliAODv0 will be filled/created at the same time as the
standard AOD.
• Will it be part of the standard AOD?
M. Oldenburg
September 19, 2007, CERN
7
Other news about the AODs
• track propagation and relation to vertices has to be
implemented
– as it is done for the ESDs
– ‘difficult’, because no direct access to the magnetic field and
the material
– possible solutions: use average material budget
• stored momentum has to be revisited
– Which momentum to store?
• at primary vertex?
• at inner TPC radius?
• …
– Where to get it from?
M. Oldenburg
September 19, 2007, CERN
8
Backup
M. Oldenburg
September 19, 2007, CERN
9
Reminder: Vertex Content of the AODs
• The standard AOD contains a TClonesArray of AliAODVertex
– vertex position
– chi2/NDF of vertex fit
– ID: points back to the array index of ESD vertex
(recently added)
–
–
–
–
type: kPrimary, kV0, kCascade, kKink
cov. matrix (compressed)
reference to parent particle
array of references to daughter particles
• 112 bytes per AliAODVertex
M. Oldenburg
September 19, 2007, CERN
10
V0 experts encountered problems
• DCA between (two) daughter tracks missing
– without this information the stored AliAODVertex is not usable
for V0 analysis
• DCA can not be recalculated on the fly from the daughter
tracks
– missing detector material and beam pipe information
• development of AliAODv0
– contains this ‘missing’ DCA information AND
– inherits from AliAODRecoDecay
• for intercompatibility reasons between PWG2 and PWG3
• AliAODRecoDecay does not contain a reference to AliAODVertex,
but a copy (= 112+64 = 176 bytes)
• new size: 184 bytes (instead of 112 + DCA)
M. Oldenburg
September 19, 2007, CERN
11