This PR is a further response following BTV investigations and aims at recovering as many pixel hits as possible compared to CKF, in the attempt to further improve impact parameter resolution. In summary:
(1) it enables backward search for iterations seeded by pixel quadruplets;
(2) it tightens the seed duplicate cleaning/merging.
For (1), first two BPIX layers in mkFit layer plan are inverted, only for iterations seeded by pixel quadruplets and only in transition regions and endcaps. In addition, hit selection windows are set also in pixel layers, to allow for backward search.
(2) allows for a further reduction in fakes+duplicates, with no cost in efficiency and with an increase of merged pixel hits in endcaps.
PR description:
This PR is a further response following BTV investigations and aims at recovering as many pixel hits as possible compared to CKF, in the attempt to further improve impact parameter resolution. In summary: (1) it enables backward search for iterations seeded by pixel quadruplets; (2) it tightens the seed duplicate cleaning/merging.
For (1), first two BPIX layers in mkFit layer plan are inverted, only for iterations seeded by pixel quadruplets and only in transition regions and endcaps. In addition, hit selection windows are set also in pixel layers, to allow for backward search. (2) allows for a further reduction in fakes+duplicates, with no cost in efficiency and with an increase of merged pixel hits in endcaps.
It supersedes PR #384.
PR validation (MTV results):
TTbar, PU=50 (vs. CKF and vs. PR386): https://mmasciovecchio.web.cern.ch/BTV_Nov2021/MTV_PR386_plusBkwSearchPixelQuad_TTbarPU50_122X/
10muPt0p2to10 (vs. CKF and vs. PR386): https://mmasciovecchio.web.cern.ch/BTV_Nov2021/MTV_PR386_plusBkwSearchPixelQuad_pixelLessSeedCleaning_10muPt0p2to10_122X/
High-pT QCD (cross-check vs. CKF and vs. mkFit as in 12_2_0_pre2): https://mmasciovecchio.web.cern.ch/BTV_Nov2021/MTV_highPtQCD_mkFit_for1220pre3/