Effect of tracking parameters on Ks reco eff

Effect of tracking parameters on Ks reco eff. to see whether we gain significantly by adjusting p. Tmin and DCA cuts in gtr_find Kin Yip (presented by Vivek Jain) Brookhaven National Lab. Stating the result first KS reco efficiency vs decay distance (plot) Followed by some plots of the Ks info. in Monte Carlo

With all the tools provided by the GTR experts, I have found that extending DCA cuts has very small effect whereas lowering p. Tmin would improve the Ks reconstruction efficiency with the price of more CPU time. In fact, extending DCA cuts do not increase even the no. of tracks. According to Slava: …. if you relax cuts you start picking up bad tracks (fakes). I keep comparing tracks against each other as a track moves towards the inner layer, and there is a probability fake track kills a good track and then fake gets delete at later stages after failing some of the cuts. I have tried changing both the DCA cuts in CFT and SMT in quite a few releases and the fact that DCA global cuts provided in GTR has very small effect is consistent.

• Sample: J/ KS MC sample with 1. 1 min. bias (PT(B) > 3 Ge. V) – There are many Ks in the MC coming from hard interaction, min. biased events … – But this time, I have paid particular attention to reco-ed Ks which match with the KS in MC – I consider those Ks (MC/reco) within the kinematic region | | 2, |z| 60 cm, p. T > 1 Ge. V for efficiency calculation • In this sample, for KS from Bd, 73. 4% with p. T >1 Ge. V, 85. 8% with | | 2 ; • For KS from Min. Biased, 10. 7% with p. T >1 Ge. V, 38. 7% with | | 2. • My Ks reco : – in principle just looping over GTrack pairs which come from the same vertex, using vkalman_xxx packages – applying my impact parameter (IP) significance cut IP/ >2 – releases such as p 11. 06. 00, t 02. 06. 00, p 11. 07. 00 etc. have been used …

Efficiencies of reco-ed Ks (matched with Monte Carlo Ks from B) • matched by the vertex distance in (x, y, z) and KS mass window • Efficiencies have some small dependency on the matching criteria.

GTrack CPU time per event bphytool CPU time per event Original 11 seconds 5 seconds p. T = 0. 2 Ge. V 17 seconds 11 seconds cpu time is release/version dependent. Signal/Noise, for example, for a 2 (from KS fitting) cut, changes from 2. 1 to 1. 6. For comparison, very few Min. bias. KS satisfy the selection criteria. Those which are selected don’t get reconstructed (using the same matching criteria), mainly due to low p. T.

DCA (or Impact Parameter) for Monte Carlo Ks from B

Decay distance vs DCA for Monte Carlo Ks from B FYI: The GTR default cut “dcamax” is 1 cm.

p. T vs for Monte Carlo Ks from B