Skip to content

Various Scifi tracking improvements and code cleanup

Dorothea Vom Bruch requested to merge dovombru_SciFi_studies_Flavios_new_parametrization into master
  • Able to define triplet seeds from various layers
  • Allow for the option of selecting more than one triplet per h1 candidate, currently commented out because not throughput performant enough yet
  • Clean up unused constants in LookingForwardConstants.cuh, grouped by LookingForward and LookinForward_sbt
  • Fix in extend missing x layers: include last x layer of T3 now
  • dzdy correction applied now when propagating from Velo/UT to SciFi (implemented by @flpisani )
  • dratio term used for propagation within SciFi, new parametrization for going from Velo/UT to SciFi (implemented by @flpisani )
  • Create functions for propagation within SciFi -> easier to switch between different versions
  • Inline small functions in LookingForwardTools.cuh -> faster, especially those for extrap (propagation within SciFi)

With this MR, we obtain the following efficiencies on 1k BsPhiPhi events, staying at about the same throughput:

TrackChecker output                               :      2307/    32630   7.07% ghosts
for P>3GeV,Pt>0.5GeV                              :        16/      235   6.81% ghosts
Long                                              :     28705/    61553  46.63% ( 47.53%),       204 (  0.71%) clones, pur  97.02%, hit eff  92.95%
Long, p > 5 GeV                                   :     25460/    40062  63.55% ( 64.52%),       168 (  0.66%) clones, pur  97.22%, hit eff  94.13%
Long, pt > 20 GeV                                 :         1/        1 100.00% (100.00%),         0 (  0.00%) clones, pur  95.83%, hit eff 100.00%
Long strange                                      :       896/     2987  30.00% ( 31.08%),         7 (  0.78%) clones, pur  96.43%, hit eff  91.93%
Long strange, p > 5 GeV                           :       726/     1397  51.97% ( 52.42%),         6 (  0.82%) clones, pur  96.89%, hit eff  94.11%
Long from B                                       :      2767/     3967  69.75% ( 70.40%),        16 (  0.57%) clones, pur  97.46%, hit eff  94.60%
Long from B, p > 5 GeV                            :      2647/     3282  80.65% ( 81.05%),        15 (  0.56%) clones, pur  97.54%, hit eff  95.07%
Long electrons                                    :       622/     4782  13.01% ( 13.58%),        16 (  2.51%) clones, pur  95.61%, hit eff  91.48%
Long electrons from B                             :        71/      246  28.86% ( 31.10%),         3 (  4.05%) clones, pur  95.77%, hit eff  91.87%
Long electrons from B, p > 5 GeV                  :        66/      158  41.77% ( 45.17%),         3 (  4.35%) clones, pur  95.78%, hit eff  92.62%
Long from D electrons, p > 3 GeV, pt > 0.5 GeV    :        34/       62  54.84% ( 55.66%),         0 (  0.00%) clones, pur  97.53%, hit eff  93.83%
Long from D, p > 3 GeV, pt > 0.5 GeV              :       750/      948  79.11% ( 77.63%),         2 (  0.27%) clones, pur  97.16%, hit eff  93.88%
Long from B electrons, p > 3 GeV, pt > 0.5 GeV    :        60/      102  58.82% ( 60.67%),         3 (  4.76%) clones, pur  96.42%, hit eff  92.85%
Long from B, p > 3 GeV, pt > 0.5 GeV              :      2530/     3051  82.92% ( 81.90%),        15 (  0.59%) clones, pur  97.57%, hit eff  94.99%
Long from B, p > 3 GeV, pt > 0.5 GeV, eta < 2.5, (phi-pi/2)<0.8:       137/      268  51.12% ( 49.18%),         1 (  0.72%) clones, pur  96.15%, hit eff  91.80%
Long from B, p > 3 GeV, pt > 0.5 GeV, eta > 2.5, (phi-pi/2)<0.8:      1215/     1377  88.24% ( 88.07%),         6 (  0.49%) clones, pur  97.50%, hit eff  95.10%
Long from B, p > 3 GeV, pt > 0.5 GeV, eta < 2.5, (phi-pi/2)>0.8:       192/      277  69.31% ( 66.53%),         1 (  0.52%) clones, pur  97.13%, hit eff  93.63%
Long from B, p > 3 GeV, pt > 0.5 GeV, eta > 2.5, (phi-pi/2)>0.8:       986/     1129  87.33% ( 86.64%),         7 (  0.70%) clones, pur  97.92%, hit eff  95.55%

FYI @flpisani @dcampora

Merge request reports