Oct 24/24 PionLT/KaonLT Analysis Meeting Notes ---------------------------------------------- (Notes by GH) Today: Mark Jones will give a special presentation on hcana tracking improvements resulting from the VCS anaysis Present ------- Regina - Garth Huber, Nathan Heinrich, Muhammad Junaid, Nacer Hamdi, Alicia Postuma FIU - Pete Markowitz CUA - Tanja Horn JLab - Dave Gaskell, Mark Jones Mark ---- - VCS experiment saw a build up of events between photon and pi0 MM^2 peaks - noticed more than expected Chi-square>100 events in Heep elastics - Heep plot of Chi-Square vs W^2 clearly shows the excess at large Chi-Square - tried various hcana algorithm improvements to find what was causing these large Chi-Square events 1) cut on DriftTime after StartTime correction is applied (removes 10-20ns jitter) - reduces # junk hits, but did not help large Chi-Square problem 2) modified NewLinkStubs algorithm - loops through Ch1,2 SpacePoints - if Stubs in Ch1,2 pass: xt_track_criterion, yt_track_criterion, xpt_track_criterion, then a track is formed - if after above loops, no track is found, makes tracks from all Ch1,2 SpacePoint combos and picks the lowest Chi-Square combo 3) made a NewFindSpacePoints algorithm - old algorithm was ok, but since SHMS/HMS DC have similar layout, clustering algorithm can be simplified - make clusters of like UU',XX',VV' planes instead of pairs of unlike planes - then make combos of UU',XX' into UX clusters, VV',XX' into VX clusters - have X,Y info from UX,VX clusters 4) modify TrackFit algorithm to eliminate hit w/largest residual and do a new fit - plot of SHMSxp stubs difference (StubXpdiff) of golden track - set xp_track_criterion=0.4 to exclude tracks >0.4, <-0.4 - #possible tracks got a lot bigger than old code, since only one golden track is selected per event, total #tracks should remain the same - Result: new code has 10X fewer events with Chi-Square>100, Heep W peak is narrower - *NB* These algorithm changes are OFF by default, need to set flag ON and set spacepoint criterion to implement - *NB* most effective one is: TrackLargeResidCut - only does track fit from stubs if residual was large - *NB* Mark has also updated the chamber alignment parameters, these are needed for tighter residual cuts applied in the above algorithms - they are available from: https://github.com/JeffersonLab/hallc_replay/blob/master/PARAM/SHMS/DC/pdc_newalign_geom.param and https://github.com/JeffersonLab/hallc_replay/blob/master/PARAM/HMS/DC/hdc_newalign_geom.param - Nacer and Junaid will investigate if this makes an improvement to their Heep replays Junaid ------ PionLT SHMS pi+ PID studies - fixed integral calculation, now #events do not drop as much when cuts are applied - used: Aero>2.5pe and 1.2