At the last HFTsoft meeting, it has been shown than increasing the precision of the LADR volume (PIXL geometry) removes the slope seen for the residuals (hit position - track projection) in the v direction ( v = longitudinal to Z, relative to the sensor).
Increasing the precision gives yLOCAL close to 0 (~1micron) whereas it was previously ~ 8 microns.

1) Simulation used : single pion, vertex ~0 with sigma = 0.1 cm

 detp geom y2013
 gexec $STAR_LIB/xgeometry.so
 gclose all
 user/vsig 0.01 .1
 gfile o test_default_precision.fz
 aguser/gkine 200 8 0.1 5.0 -1.0 +1.0 0 6.28
 trig 100

.x StRoot/macros/bfc.C(1,100,"y2013,AgML,ITTF,Sti,tpcI,StiRnD,PixelIT,TpcFastSim,pxlFastSim,-ssdfast,VFMCE,McEvent,
geant,IdTruth,fzin,NoSvtIt,NoSsdIt,-StiPulls,Bana,-tags,clearmem,evout,McEvOut,MiniMcMk,McAna,MakeEvent","")

2) Results
The results are below. Each plots show :

• left   : residuals in u direction (transverse to beam) vs Zglobal (of pixel hits)
• right : residuals in v direction (transverse to beam) vs Zglobal (of pixel hits)

3) The test here is by setting the local coordinate Y of hits to 0 (in StPxlFastMaker.cxx)

               smearedX = localPixHitPos[0]; 
               smearedZ = localPixHitPos[2]; 
               //test by setting yL=0 
               smearedY = 0; 
               localPixHitPos[0] = smearedX;localPixHitPos[2] = smearedZ;localPixHitPos[1] = smearedY; 
               gGeoManager->GetCurrentMatrix()->LocalToMaster(localPixHitPos, smearedGlobalPixHitPos); 
               StThreeVectorF gpixpos(smearedGlobalPixHitPos); 
               (...)   
               StPxlHit* tempHit = new StPxlHit(gpixpos, mRndHitError, hw, mcPix->dE() , 0); 

4) Results


5) comments - conclusions

1. setting yLoc = 0 when filling StPxlHit does not improve the residuals for the geometry modified with a precision to 1micron
2. setting yLoc = 0 when filling StPxlHit does improve the residuals for the default geometry.
   1. the slope/split of residuals at Z =0 is attenuated
   2. but the slope is still present