Hits distribution for high energy electrons, geom=UPGR05, used detectors: FST,IGT,TPC+vertex
All analysis is based on the Geant information, all events have zVertex=0.

  1. Display of geometry=UPGR05. Response to 40 GeV ET electrons is shown, zVert=0cm, electron eta range [1,2].

  2. A different electron sample was used in the analysis below,total 100 events. 
    * gkine #part. partID  ptrange   yrange   phirange  vertexrange
* gkine  1     3      5.0 40.1  0.8 2.5  1.5 1.8    0. 0.1

  3. Algo:
    * find the prim electron track from geant (from StMcEvent). For this track get all geant hits form: FST, IGT, and TPC. For FST & IGT sort hits vs Z-value.
    * prune multiple hits from the same 'pixel'
    FST, IGT : ignore subsequent hits if delZ <5 mm
    TPC: ignore subsequent hits from the same sector& padrow, clip hits at |Z|>198 cm, drop padrow=1 and 13.

  4. FST response: Left : Z of fired disk vs. track eta. Right : # of fired disks vs. track eta.

  5. IGT response: Left : Z of fired disk vs. track eta. Right : # of fired disks vs. track eta.
    Note, the 'blue' hits on the left plot and the blob above the triangle at eta=1.5 on the right plot are unphysical. Not sure what is wrong - a minor problem.

  6. TPC response: Left : Ry of the fired padrow vs. track eta. Right : # of fired padrows vs. track eta.
    Note, padrow 1 & 13 are masked out.

  7. Summary: # of hits from FST,IGT, and TPC (the same plots). The last plot is the sum, includes the vertex.
    The last plot can be compared to Ross earlier result of totHit vs. eta. There are minor difference in used phi distribution of electrons, I have masked 2 padrows and added the vertex.
    The main reason for discrepancy at eta above 1.6 is due to the double counting of FST & IGT hits - there was no pruning in Ross code.

  8. How important is the vertex constrain for pT reco?
    Relevant quantities (assuming all tracks are straight lines):
    L1 = track length (firstHit -lastHit), using any of the trackers (FST,IGT, and/or TPC )
    L2 = distance between vertex and middle of the L1.
    The 'vertex factor' defined as Vf=(L2-L1/2)/L1, tells us how much adding the vertex will improve relative length of the track, hence accuracy of determination of its saggita.
    I think if one multiplies Vf * ratio (vertex resolution/first hit resolution) we will get quantity proportional to the pT improvement if vertex is added to the helix fit. (perhaps one need to square it)
    Structures on the plots below:
    * at eta=[1.1,1.2] -FST disks come to play,
    *at eta 1.8 we loose the last TPC point and later only FST+IGT are tracking (I masked out TPC padrow #1)
    * at eta > 2 the 4th IGT disk enters the game

  9. Conclusion:
    The largest relative improvement of pT due to the vertex constrain will be for track at eta below 1.2 (if zVertex=0).
    Code: StGeantEleFinderMaker.cxx