In the scope of my qualification task (JIRA: ATLIDTRKCP-538), I have applied an additional cut for LRTs in the production of 3 SP seeds before adding top and bottom SP candidates to the relevant data objects. The mentioned pT cut uses conformal mapping before the top-center and bottom-center space points are constructed. Although the main pT cut is applied in L.2480, in L.2395 the "coordinates" function requires a lot of CPU time. Applying a pT cut early on, in our case before the space point triplets are constructed, reduces the "coordinates" function's run time. With this additional cut, the CPU time ratio of the ITkSiSpTrackFinderLargeD0 to the ITkSiSpTrackFinder has decreased from 8.3 to 2.8, as tested using a ttbar sample. The efficiencies versus eta, pT, and prodR remain unchanged for the signal sample and the relevant plots can be found in the attachment.
