40 likes | 152 Views
vertex charge purity vs efficiency. Dnorm approach, all tracks used. L/D approach. L/D > 0.3. detector dependence clearly seen for Dnorm approach, but not for L/D – WHY?. detector dependence for L/D seen in the BRAHMS framework (Nicolo’s study)
E N D
vertex charge purity vs efficiency Dnorm approach, all tracks used L/D approach L/D > 0.3 detector dependence clearly seen for Dnorm approach, but not for L/D – WHY?
detector dependence for L/D seen in the BRAHMS framework (Nicolo’s study) • its absence in SGV needs to be understood (might point to problem with SGV) • Strategy: • difference in Qvtx purity corresponds to difference in number of missed tracks • (= tracks from B decay according to MC information, but not assigned • by ZVTOP and track attachment) • find out, where this difference come from in BRAHMS, and compare with SGV • suggestion: compare the L/D distributions of missed tracks • would expect L/D value of a track to depend on detector resolution • then the slope of that distribution should change with detector (migration)
L/D distribution for missed tracks (SGV) L/D not evaluated (0 entered) for tracks failing track selection cuts (see next page) improved detector: efficiency 69.6% 4329 tracks missed (15.3% vertices affected if exactly 1 track/vtx missed) degraded detector: efficiency 67%, 4281 tracks missed (15.7%) 2 detectors compared cut value change in slope?
effect of track selection cuts (SGV) compare same detectors as before, i.e. an improved detector to a degraded one • for each cut one entry per track, • which it discards • ( > 1 entry / track possible ! ) • cut, which removes largest • number of tracks: • transverse momentum > 0.1 GeV • cut, which shows • largest detector dependence: • Rf impact parameter resolution