This blog entry is incrementally summarize the progress of using the HFT to study TPC alignement. 
I. TPC hits pulls of cosmic rays (update May/23rd). 
II. ZPulls vs. PadRow (update May/27th). 
III. ZPulls vs. PadRow (update May/28th).
Conclusion and summary by Gene:
http://www.star.bnl.gov/HyperNews-star/protected/get/startpc/1352/1/1/1/1/1/1.html

IV. Prompt hits location (update May/29th). 

 I. TPC hits pulls of cosmic rays (May/23rd). 

Data:

1. Cosmics RFF events. Runs: days 50-103 with PXL and IST.
2. Require matching of two cosmic tracks with (a la ~fisyak/macros/Cosmics.C). 
3. p_{T}>1 GeV/c and nTpc>=15 for each track.
4. DCA to (0,0,0): |impact|<5cm. and |z|<20. cm. 
5. Chain: DbV20140412,P2014a,mtd,btof,BEmcChkStat,Corr4,OSpaceZ2,OGridLeak3D,-hitfilt,pxlRaw,pxlCluster,pxlHit,istRaw,istCluster,istHit,sstHit,StiPulls
6. IST hit errors are from the DB which were 17 microns instead of 170 microns at the time we ran this procution. 

Observations:
Bellow are the pulls of TPC hits with (left) and wihtout (right) requiring track to go through HFT hits (local Sti coordinates, i.e. Y is in the rPhi direction). 

1. ZPulls are not centered around zero. Inner sector hits have a shift by ~0.9mm while outer has a -0.25mm shift (opposite signs for the east side). 
2. Inner TPC ZPulls are ameliorated if HFT is required (HFT aligned with inner sector?). 
3. YPulls mean is just ok. The east side is less satisfactory than the right side. 
4. YPulls width seem to have a repeating structure (lowest left plot). 

all	nPxl>1 && nIst>0

II. ZPulls vs. Pad Row (May/27th):

Below is Zpulls vs. Pad Row asked for by Gene during TPC meeting. It seems to confirm the difference in T0 between inner and outter TPC as hypothesized by Gene during the meeting. 


III. ZPulls vs. Pad Row  (update May/28th):

Below is Zpulls vs. Pad Row from Au+Au200GeV Run14, run 15080038, |Vz|<20cm && nTpcHits>30. Default tpcPadrowT0 from db (timestamp: 20131220.000100). 

Below is Zpulls vs. Pad Row from U+U, |Vz|<20cm && nTpcHits>30. Default tpcPadrowT0 from db (timestamp: 20120202.000000). 

Below is Zpulls vs. Pad Row from Au+Au200GeV Run14, run 15080038, |Vz|<20cm && nTpcHits>30. tpcPadrowT0 same as U+U (timestamp 20120202.000000). 

Below is a projection of the plot above. It seems that timestamp 20120202.000000 tpcPadrowT0 works very well for AuAu200GeV Run14 data. 

IV. Prompt hits location (update May/29th):
I took a look at the location of the prompt hits after Gene's suggestion.

The following is from Run14 AuAu200GeV run 15080038 with tpcPadRowT0 timestamp 20120202.000000. 
First figure: timeBucket of prompt hits vs. padrow.
 
Second figure: global Z - GG_Z0 vs. padrow (tpcPadRowT0 timestamp 20120202.000000).

The following is from Run14 AuAu200GeV run 15080038 with wrong tpcPadRowT0 (timestamp 20131220.000100):

From UU with tpcPadRowT0 timestamp 20120202.00000:

-- Mustafa