A Study of SVT efficiency

Purpose: To find out out SVT tracking performance as a function of various variables (phi, pt, eta, and z etc) and also to debug the SVT system.

Efficiency = Matched SVT tracks / Good offline tracks

Good offline tracks (=COT+SVX tracks)

• Good COT track
  • Pt>2.5 GeV, |d0| < 1 cm
  • |z0| < 47 cm, |eta| < 1.0
  • track quality cuts (Axial: N_hits>=20, N_segments>=3, Stero: N_hits>=20, N_segments>=2)

• Good SVX track
  • SVX track linked with good COT track
  • SVX hits at Layer 0, 1, 2, and 3 (current SVT tracking requires this configuration)

• Matched XFT track for above COT track (currently, NOT implemented, XFT efficiency is about 90-95% according to XFT people)

Matched SVT track w.r.t. good offline tracks

Matching in phi ( |del (phi)|<0.03 )

Matching in curvature ( |del (curv)| < 0.001, |del (curv)/curv | <0.3)

Additional matching in barrel Z_id and phi wedge between SVT and SVX.

In case we find more than one matched track, the track with smallest chisq is selected.

SVT efficiency plots (preliminary!)

• Effficiency as a function of phi, eta, and pt for the case of passing centeral barrels (2,3) with passing SAME barrel
  • Overall efficiency, 55.4% for the case of NOT allowing offline tracks with crossing different phi wedges, 54.6% for the case of ALLOWing (difference is shown around wedge boundary region).
  • A dip in efficiency around phi=4.6?
    • There is NO reconstructed SVT track around phi=4.6 for barrel = 2 (see the SVT most inner Z barrel vs phi plot ), why?
    • This is not due to some missing strips (around phi=4.6), I see continuous distributions for both SVX and SVT hits at layer 0,1,2, and 3. (also see the SVT hit positions vs phi plot ) for layer 0,1,2,and 3 at wedge=9)
• Effficiency as a function of phi for the case of passing centeral barrels (2,3) with crossing DIFFERENT barrels (overall eff = 26.8%)
  • Why is the efficiency lower? compared to the case of passing SAME barrel.

What can cause lower SVT efficiency? under study

• Possible source in requirement for the good offline tracks
  • Linked SVX hits along the offline tracks are fake? due to problem in the offline tracking? (need to look at the residual of SVX hits)
  • Lack of XFT information? (need to require this information for a good offline tracking, currently not done due to memory leakage problem in Stntuple). Probably, this will account for 5-10% inefficiency.

• Possible source in requirement for matched SVT tracks (most of below cases can be understood by using SVT emulation tool)
  • NO reconstructed SVT hits though there are good SVX hits?
  • NO corresponding XFT track to SVT system? due to failure of XFT track reconstruction or async. problem
  • Missing pattern though thre are good SVT hits and XFT track?
  • Wrong SVT fitting results though thre are good SVT hits, XFT track, and matched pattern?