I processed ~45k st_centralpro Run 12 UU193 events (using runs scattered over the UU operations duration) with and without the new TPC alignment, using the following two chains in SL14h:

Old alignment: DbV20120921,P2012b,AgML,btof,BEmcChkStat,Corr4,OSpaceZ2,OGridLeak3D,-hitfilt
New alignment: DbV20140919,P2012b,AgML,btof,BEmcChkStat,CorrX,Corr4,OSpaceZ2,OGridLeak3D,-hitfilt

(note: I had to use a local version of Run 12 UU TPC T0 with new alignment as this is not yet available from the database)

I used the same track-wise cuts as Rihan did in his h^-/h⁺ study of primary tracks for Run 12 UU using the official P12id production, however I implemented no event-wise cuts. The results are shown below. Some observations with the new TPC alignment compared to the old:

• Less primary tracks are reconstructed in all quadrants of -/+ charge and pseudorapidity
• h^-/h⁺ is closer between east and west TPC, by a factor of ~x3 better as determined by the vaue of B from a fit to (A + B * p_T²)
  • TPC distortions usually introduce effects that scale as p_T²
  • The overall scale value of the east/west double ratio (i.e. A, the value as p_T → 0) may be offset from 1.0 due to different east vs. west acceptances of different charge signs because of dead channels int the TPC
  • There is reason to believe a few specific sectors have incorrect GridLeak distortion corrections, which could contibute to the remaining east vs. west discrepancy; a sector-by-sector analysis (requiring at least x12 statistics) would elucidate
• h^-/h⁺ is closer to 1.0 at high p_T for both east and west

-Gene