Following last week's meeting, I have rerun the BFC chain and compared the dmetric and StiPulls for :

1. default geometry (as in DEV)
2. geom.1 : geometry with 2 huge overlaps removed (see link above)
3. single SSD barrel

1) Dmetric vs. Z


2) Dmetric vs. #phi


3) StiPulls for TPC only (using mHitsG branch)
a) pulls in Y and Z


b) pulls resolution in Y and Z


4) StiPulls for TPC+SSD-geom. 1(using mHitsR branch)
a) pulls in Y and Z


b) pulls in Y and Z


5) StiPulls for TPC+SSD-single barrel (using mHitsR branch)
a) pulls in Y and Z


b) pulls resolution in Y and Z

Comments :

1. pulls for TPC only are better than pulls for TPC+SSD-geom1 or TPC+SSD-single
2. pulls resolution for TPC are closed to 1 (sigma of gaus)
3. pulls for TPC+SSD-geom 1 are worst than TPC alone
4. pulls resolution for TPC+SSD-geom1 are bad !
5. pulls for TPC+SSD-single are worst than TPC alone but better than TPC+SSD-geom1
6. pulls resolution for TPC+SSD-single are closed to 1