class AliTPCtracker: public AliTracker

update track information using current cluster - track->fCurrentCluster

 decide according desired precision to accept given
 cluster for tracking

 dummy

 TPC tracker destructor

fill esds using updated tracks

 Use calibrated cluster error from OCDB

 Use calibrated cluster error from OCDB

rotate to track "local coordinata

 Initial approximation of the track curvature

 Initial approximation of the track curvature

 Initial approximation of the track curvature

 Initial approximation of the track curvature times center of curvature

 Initial approximation of the tangent of the track dip angle

 Initial approximation of the tangent of the track dip angle

 This function find proloncation of a track to a reference plane x=x2.

 load clusters

 load clusters to the memory

 load clusters to the memory from one
 TClonesArray

 load clusters to the memory

 Calculate crosstalk estimate

 filter outlier clusters


    1.)..... booking part
    nSectors=72;
    nTimeBins=fParam->Get....
    TH2F hisTime("","", sector,0,sector, nTimeBins,0,nTimeBins);
    TH2F hisPadRow("","", sector,0,sector, nPadRows,0,nPadRows);
    2.) .... filling part
    .... cluster loop { hisTime.Fill(cluster->GetDetector(),cluster->GetTimeBin()); }

    3.) ...filtering part
    sector loop { calculate median,mean80 and rms80 of the nclusters per time bin; calculate median,mean80 and rms80 of the nclusters per par row; .... export values to the debug streamers - to decide which threshold to be used... }

    sector loop
    { disable clusters in time bins > mean+ n rms80+ offsetTime disable clusters in padRow > mean+ n rms80+ offsetPadRow // how to dislable clusters? - new bit to introduce }
    //
    4. Disabling clusters

 unload clusters from the memory

 Filling cluster to the array - For visualization purposes

 transformation

 ApplyXtalk correction
 Loop over all clusters
      add to each cluster signal corresponding to common Xtalk mode for given time bin at given wire segment
 cluster loop

 Correct the cluster charge for the ion tail effect
 The TimeResponse function accessed via  AliTPCcalibDB (TPC/Calib/IonTail)

 This function fills outer TPC sectors with clusters.

 This function fills inner TPC sectors with clusters.

       Return pointer to a given cluster

 This function tries to find a track prolongation to next pad row

 Get track space point by index
 return false in case the cluster doesn't exist

 This function tries to find a track prolongation to next pad row

 This function tries to find a track prolongation to next pad row

 This function tries to find a track prolongation.

 This function tries to find a track prolongation.

 overlapping factor

 shared clusters

sort trackss according sectors

sort tracks in array according mode criteria

 Loop over all tracks and remove overlaped tracks (with lower quality)
 Algorithm:
    1. Unsign clusters
    2. Sort tracks according quality
       Quality is defined by the number of cluster between first and last points

    3. Loop over tracks - decreasing quality order
       a.) remove - If the fraction of shared cluster less than factor (1- n or 2)
       b.) remove - If the minimal number of clusters less than minimal and not ITS
       c.) if track accepted - sign clusters

Called in - AliTPCtracker::Clusters2Tracks()
          - AliTPCtracker::PropagateBack()
          - AliTPCtracker::RefitInward()

 Arguments:
   factor1 - factor for constrained
   factor2 -        for non constrained tracks
            if (Float_t(shared+1)/Float_t(found+1)>factor) - DELETE

 Dump clusters after reco
 signed and unsigned cluster can be visualized
 1. Unsign all cluster
 2. Sign all used clusters
 3. Dump clusters

 loop over all clusters and unsign them

sign clusters to be "used"

 snumber and sdensity sign number of sigmas - bellow mean value to be accepted
 loop over "primaries"

 back propagation of ESD tracks

return 0;

 back propagation of ESD tracks

 Post process events

read seeds from the event

 This function creates track seeds.
 SEEDING WITH VERTEX CONSTRAIN

 cuts[0]   - fP4 cut
 cuts[1]   - tan(phi)  cut
 cuts[2]   - zvertex cut
 cuts[3]   - fP3 cut

 This function creates track seeds.

 cuts[0]   - fP4 cut
 cuts[1]   - tan(phi)  cut
 cuts[2]   - zvertex cut
 cuts[3]   - fP3 cut

 This function creates track seeds - without vertex constraint

 cuts[0]   - fP4 cut        - not applied
 cuts[1]   - tan(phi)  cut
 cuts[2]   - zvertex cut    - not applied
 cuts[3]   - fP3 cut

reseed using track points

reseed using founded clusters

 Find the number of clusters

reseed using founded clusters

  find multi tracks - THIS FUNCTION IS ONLY FOR DEBUG PURPOSES
                      USES MC LABELS
  Use AliTPCReconstructor::StreamLevel()& kStreamFindMultiMC if you want to tune parameters - cuts

  Two reasons to have multiple find tracks
  1. Curling tracks can be find more than once
  2. Splitted tracks
     a.) Multiple seeding to increase tracking efficiency - (~ 100% reached)
     b.) Edge effect on the sector boundaries


  Algorithm done in 2 phases - because of CPU consumption
  it is n^2 algorithm - for lead-lead 20000x20000 combination are investigated

  Algorihm for curling tracks sign:
    1 phase -makes a very rough fast cuts to minimize combinatorics
                 a.) opposite sign
                 b.) one of the tracks - not pointing to the primary vertex -
                 c.) delta tan(theta)
                 d.) delta phi
    2 phase - calculates DCA between tracks  - time consument

 Find Splitted tracks and remove the one with worst quality
 Corresponding debug streamer to tune selections - "Splitted2"
 Algorithm:
 0. Sort tracks according quality
 1. Propagate the tracks to the reference radius
 2. Double_t loop to select close tracks (only to speed up process)
 3. Calculate cluster overlap ratio - and remove the track if bigger than a threshold
 4. Delete temporary parameters

  find Curling tracks
  Use AliTPCReconstructor::StreamLevel()&kStreamFindCurling if you want to tune parameters - cuts


  Algorithm done in 2 phases - because of CPU consumption
  it is n^2 algorithm - for lead-lead 20000x20000 combination are investigated
  see detal in MC part what can be used to cut

  find kinks


 RS something is wrong in this routine: not all seeds are assigned to daughters and mothers array, but they all are queried
 to check later

 refit kink towards to the vertex

 update Kink quality information for mother after back propagation

 update Kink quality information for daughter after refit

 check kink point for given track
 if return value=0 kink point not found
 otherwise seed0 correspond to mother particle
           seed1 correspond to daughter particle
           kink  parameter of kink point

 reseed - refit -  track

 This function reades track seeds.

 clusters to tracks

 clusters to tracks

 This is a track finder.

 tracking of the seeds

tracking routine

 tracking

 seeding adjusted for laser and cosmic tests - short tracks with big inclination angle
 no primary vertex seeding tried

sum tracks to common container
remove suspicious tracks
 RS: Attention: supplied tracks come in the static array, don't delete them

 try to track in parralel

 if we use TPC track itself we have to "update" covariance

 if we use TPC track itself we have to "update" covariance

 make back propagation

 make forward propagation

 propagate track forward
UnsignClusters();

 make back propagation, in between row0 and row1

 gets cluster shape

This function "cooks" a track label. If label<0, this track is fake.

This function "cooks" a track label. If label<0, this track is fake.

return pad row number for given x vector

 Fill the cluster and sharing bitmaps of the track

 return flag if there is findable cluster at given position

 Adding systematic error estimate to the covariance matrix
                !!!! the systematic error for element 4 is in 1/GeV
 03.03.2012     MI changed in respect to the previous versions

 Adding systematic error - as additive factor without correlation

                !!!! the systematic error for element 4 is in 1/GeV
 03.03.2012     MI changed in respect to the previous versions

 check events affected by TPC HV dip

 account that this seed is "deleted"

 return next free slot where the seed can be created

 mark all seeds in the pool as unused

 try to track in parralel

 tracking

fill the cluster occupancy info into the ESD friend

 check if the given sector row is active

  return (row>=fInnerSec->GetNRows()) ? fOuterSec->GetX(row-fInnerSec->GetNRows()):fInnerSec->GetX(row);

return (row>=fInnerSec->GetNRows()) ? fOuterSec->GetMaxY(row-fInnerSec->GetNRows()):fInnerSec->GetMaxY(row);