150 likes | 331 Views
Primary vertex with TPC-only tracks. Motivation: TPC stand-alone analyses (multiplicity, spectra). Andrea Dainese INFN Legnaro. (twice: ITS+TPC, TPC only; both stored in ESD). (twice: ITS+TPC, TPC only). (twice: ITS+TPC, TPC only; both stored in ESD). Barrel Tracking & Vertex Recons.
E N D
Primary vertexwith TPC-only tracks Motivation: TPC stand-alone analyses (multiplicity, spectra) Andrea Dainese INFN Legnaro ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
(twice: ITS+TPC, TPC only; both stored in ESD) (twice: ITS+TPC, TPC only) (twice: ITS+TPC, TPC only; both stored in ESD) Barrel Tracking & Vertex Recons. • ITS local reconstruction (➔ RecPoints) • Vertex (x,y,z) from SPD tracklets • Barrel Tracking • Vertex with tracks • Compute diamond transverse profile, as mean and RMS of vertices in high-multiplicity events of each “LHC fill” • Repeat zvtx with SPD starting from mean (x,y) • important, to have zvtx for all events (even with one tracklet), w/o bias • Barrel Tracking • Vertex with tracks, using info on diamond profile • final (x,y) resolution better than diamond RMS for ALL events 1st pass on subsample 2nd pass full statistics ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Vertex reconstruction with tracksAliVertexerTracks • Three main “tasks” implemented: • TRACKS SELECTION • try reject secondary tracks • VERTEX FINDING • give first estimate of vtx position • VERTEX FITTING: • give optimal estimate of the position of the vertex • give vertex covariance matrix • give a c2 (vertex quality) ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Primary vertex reco with tracksAliVertexerTracks::FindPrimaryVertex() • Primary vertex reconstruction: iterative procedure • Select tracks • track points to fiducual cylinder (R<3 cm, |z|<30 cm) • for ITS+TPC: kITSrefit and >4 ITS clusters • for TPC: kTPCin, >50 cls and |tgl|<1 • 0th iteration: • VERTEX FINDING, using track pairs vtx0 • 1st iteration: • TRACKS SELECTION: |d03D(vtx0)| < 0.5 cm for ITS+TPC < 2.5 cm for TPC • VERTEX FINDING • VERTEX FITTING vtx1 • 2nd iteration: • TRACKS SELECTION: |d0rf(vtx1)| < 3 sd0 • VERTEX FINDING • VERTEX FITTING vtx2 (final estimate) ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Impact parameter distr. of TPC tracks z rf reasonable pulls ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Diamond constraint • Why diamond constraint? • vertex also for events with 1 “primary” track • better rejection of secondaries • (x,y) resolution always better than diamond size • Exercise on CAF (as it could be done on the data) • use 10k triggered events, pp 900 GeV • select vertices with > 10 tracks • weighted average of (x,y,z) • diamond: • centre: weighted average (tried also with plain average: compatible results) • sigma: error of weighted average (20-30 mm) diamond rms from machine (200 mm @ 900 GeV) • use this diamond as constraint and re-evaluate vertices (2nd reconstruction pass)
TPC Vertex efficiency • Efficiency: events with vertex / triggered (MB1=SPD||V0A||V0C) • Different normalization: events with vertex / events triggered and with >0 TPC tracks pointing to the beam pipe
TPC Vertex efficiency PHOJET 900 0T PHOJET 900 0T PHOJET 900 0.5T PHOJET 900 0.5T
Resolutions and pulls (PHOJET 900 0T) closed markers: RMS; open markers: s fit no constraint with constraint Ideal simulation and reconstruction First Physics test production (TPC misaligned, miscalibrated) with constraint no constraint
Code • AliVertexerTracks can take an array (TObjArray) of AliExternalTrackParam tracks • Separate “MeanVertex” objects in OCDB for ITS+TPC and TPC-only • Selection criteria for AliVertexerTracks kept in new class AliGRPRecoParam (from rec.C or OCDB) • separate criteria (incl. constraint flag) for ITS and TPC “mode” • performance optimisation without touching compiled code • In AliReconstruction: AliVertexerTracks tVertexer; // TPC + ITS primary vertex tVertexer.SetVtxStart(fDiamondProfile); // from OCDB tVertexer.SetITSMode(grpRecoParam->GetITSCuts()); ... find vertex ESD // TPC-only primary vertex tVertexer.SetVtxStart(fDiamondProfileTPC); // from the OCDB tVertexer.SetTPCMode(grpRecoParam->GetTPCCuts()); ... find vertex ESD ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Next • The extraction of the mean vertex and storage in the OCDB is run manually on CAF • OK for first data at 900 GeV • check stability of diamond position vs time, run#, ... • Later the procedure will have to become automatic and run on ESDs accessed on grid • how to identify the run/event/time at which the diamond moves?
Comparison with MC vertex: shift Ideal simulation and reconstruction First Physics test production (TPC misaligned, miscalibrated)
y track x vertex need an estimate of the vertex and its cov. matrix Tracks Selection • Idea: use “primary”tracks for vertex reconstruction • Only tracks with 5 or 6 points in ITS • Select primary tracks on the basis of their d0 • Cut |d0| < nsd0(pt) sd0(pt) = svtxstrack(pt) from track’s covariance matrix ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Vertex Fitter • “Fast Vertex Fitting”, CMS Note 1997/051 • Tracks are propagated to the point given by the vtx finder • Each track gives an independent measurement of the vertex position: a c2 is written as the sum of the single track c2s w.r.t. a generic vertex position rvtx: where ri is the (x,y,z) position of the ith track and Vi = Wi-1 is the cov matrix of ri, extracted from the track’s cov matrix • The solution that minimizes this c2 is analytic: vertex covariance matrix vertex ALICE Offline Week, CERN, 22.10.08 Andrea Dainese
Easy to add “beam constraint” • Same procedure as before + constraint: vertex vertex covariance matrix where: ALICE Offline Week, CERN, 22.10.08 Andrea Dainese