Sep, 04 (tulika) v00-00-92 ======================================================================== (1) Some structural changes were necessitated for running with the new tmb_tree in p16.04. References to tmb_analyze were replaced by tmb_tree. (2) In GenericDMu_Finder.cpp the MBPSelector picks up the p14 MB probability file still which is supposed to be p16 compatible (there is no MB probability file for p16). (3) Turned back ON the mass-constraint of the D track in methods getD0Info, getDsInfo, getDInfo. The mass-constraint is still turned off in getDstarInfo (oversight on my part - will remedy in the next version). NOTE: The Lxy > 0 cut on the D vertex is still commented out a la v00-00-91. For behaviour similar to v00-00-89, please put back in this cut ( in methods getD0Info, getDstarInfo, getDsInfo, getDInfo ). (4) Increased the K* mass window range in DsKstarK_Finder.cpp. Sep01, 04 (christos) v00-00-91 ======================================================================== Changes for all D+mu modes: (1) Added to root-tuple: - dca, sigma(dca) for tagging track of all SST algorithms wrt B vertex and PV (needed for D**0 pion contribution to same-side tagging) - collinearity cut for D track (for background suppression) (2) Removed Lxy > 0 cut for D vertex (cut may bias B lifetime and/or B decay vertex resolution; may decrease statistics): PLEASE STUDY THIS (3) D+mu decay vertices are now calculated w/o a mass-constraint of the D track (mass-constraint may bias the D spectrum for candidates far from the nominal D mass; have not studied effect on B vertex resolution): PLEASE STUDY THIS (methods getD0Info, getDstarInfo, getDsInfo, getDInfo) (4) Added option for applying impact parameter cuts on kaon, pion for all D+mu modes (should speed up code) Changes for D*+mu mode: (5) Calculating D* - D mass difference before vertexing D*+mu (6) Removed option for applying impact parameter cut on muon Jun06, 04 (christos) v00-00-89 ======================================================================== *** D0->K+pi mode only: *** Added access methods for setting Abs[dca/sigma(dca)] cuts for muon, pion and kaon. Expect very small (<1%) increase in efficiency for "saturated" events (ie. when # of B candidates exceeds maximum N_B_MAX) Setting cuts to 1.0 for muon, pion, kaon increases code speed by a factor of ~7 (single-muon data) because of the decrease of the output file *** D0->K+pi mode only: *** Reverted action of Jan31, 04 (so that I can cut on impact parameters earlier in the code) - I hope I have not broken anything! ;-) Jun05, 04 (christos) v00-00-88 ======================================================================== (1) Upgraded d0root tags; From: d0root_analysis v00-09-37 d0root_tmbtree v00-09-13 d0root_btag v00-09-50 To: d0root_analysis v00-09-52 d0root_tmbtree v00-09-26 d0root_btag v00-09-61 They fix a serioug bug in neutral tracking (relevant for the D0+mu reconstruction affecting phi<0 neutral tracks); The efficiency gain is ~25%. (2) New AATrack tag: v02-10-07 (includes new run database with more beam spots) (3) Miscellaneous: (a) internal: - renaming of jet- and soft-jet-tracks variables - minor reorganization of good-track definition (b) installation: swapped order of d0root_analysis/read_event and d0root_analysis/algorithms compilation Apr21, 04 (christos) v00-00-86 ======================================================================== Removed forgotten histograms with distributions of phi for tracks in the event Apr21, 04 (christos) v00-00-85 ======================================================================== (1) PV determination in MC: now using beam-spot (constant width of 30 microns) as initial seed (just like when running on real data) (2) Now using bflavor_tag v00-00-80; added entries for all SST tagging algorithms to output root-tuple (3) other root-tuple changes: (a) muon: added # of SMT, CFT hits (b) other charged tracks: added # of CFT hits, renamed # of SMT hits variable (nhits --> nsmt) (c) removed delta-R(muon, charged track), cos(muon, charged track) (d) added eta, phi of D, D+mu (e) PV multiplicity: changed from Float_t to Int_t (f) Now resetting event-scope information at beginning of event (4) Changes in setup (do not affect analysis): (a) p14.05.01 --> p14.06.00 (b) d0root_analysis v00-09-36 --> v00-09-37 (c) d0root_btag v00-09-45 --> v00-09-50 Feb08, 04 (christos) v00-00-80 ======================================================================== (1) bug fixes: (a) Ds->K* K mode: proper resetting of pointer to non-K* kaon track in event loop (b) Ds->Phi Pi mode: memory leak fixed; Fitter was not Clear()-ed at beginning of getDsVertex method (credit goes to vj for spotting the problem) (2) added helicity, kinematic angles to root-tuple (requested by vj) (a) Ds->Phi Pi mode: angles between Phi->KK kaons and Ds at Phi rest frame; angle between Pi at Ds rest frame & Ds direction in lab (b) Ds->K* K mode: angles between kaon, pion (from K*) and Ds at K* rest frame; angle between (non-K*) kaon at Ds rest frame & Ds direction in lab (c) D0->K Pi mode: angles between Pi (and K) at D0 rest frame & D0 direction in lab (3) Misc. (a) D+ -> K Pi Pi mode: removed redundant Fitter->Clear() calls at end of endEvent method (b) Renamed KPiStarCandidate to MuKPiStarCandidate for consistency Jan31, 04 (christos) v00-00-78 ======================================================================== Moved calculation of impact parameters for tracks in final state at the end of event reconstruction; should speed things up a bit since calculation is now done only if event is to be saved in root-tuple Jan30, 04 (christos) v00-00-77 ======================================================================== Changes in PV determination: (1) chi^2 on tracks: reduced to 10 (from 100; value "100" recommended only with 1-pass PV finder w/o beam constraint) (2) beam-spot: now using AA database (removed Leah's macro and supporting code) (3) setting PV's beam-spot only upon new run (used to be on every event) Jan28, 04 (christos) v00-00-75 ======================================================================== (1) Now using d0root tags from v9.1: addpkg d0root_analysis v00-09-36 addpkg d0root_tmbtree v00-09-13 addpkg d0root_btag v00-09-45 addpkg AATrack v02-10-05 This set of tags fixes the bug in D0Vertex::RemoveTrack (GetEntriesFast --> GetEntries) introduced in d0root_analysis v00-09-20 (through v00-09-24); This bug affected versions v00-00-71 through 73 of bmixing_reco (2) Commented out the delete's of VertexGlobalFinder and MassConstraintVertex in the destructors: were causing ROOT to hang at the end of the session; (3) Removed #include TMBTreeClassess_linkdef.h from code (4) Now using bflavor_tag v00-00-70 Dec11, 03 (christos) v00-00-73 ======================================================================== removed PV-finder destructors from GenericDMu_Finder destructor; there seems to be a conflict between d0root classes and ROOT about who owns D0ParticleCollection that results in segmentation faults at the end of the job d0root classes now call Clear() method at beginning of event, only if they are to be used in event (used to be at the end of event, whether they had been used or not) Dec08-09, 03 (christos) v00-00-71 ======================================================================== (1) Moved from using muon's nseg as a quality criterion to using the muon-id group's quality definitions; for B->D+mu reconstruction this does not make a difference (since we only use muons matched to tracks: equivalent to nseg>=0), however we now can cut on the muon quality (extra entry stored in the output root-tuples, in addition to nseg) (2) Best-PV selection: based on loose-or-better matched-to-track muons (used to be nseg>=1 matched-to-track muons); so, a little tighter (3) Code now uses latest b-flavor tagging code (new addition: SST algorithm) (4) Removed muo_cand dependencies from code (5) Removed (unnecessary) TMBTree-pointer initializing Nov28-Dec01, 03 (christos) v00-00-65 ======================================================================== (1) Moved to d0root v9.0 (better PV resolution, memory leaks fixed, code speedup) (2) Code now can use muons with nseg>=0 (bugfix in d0root); those guys used to be discarded; (3) New default PV-finder: 1-pass w/ beam constraint (switching to 2-pass finder if run not in database); faster & more reliable (4) Best-PV selection: based on muons w/ nseg>=1 in event (used to be min.bias probabilistic function) (5) User now has the option of defining jet-cone size (dR). Default value: 6.0 - I've seen a drop in the event yield for smaller cone sizes, but for dR = 2.0, my efficiency is up at the same levels again, and the code runs twice as fast w.r.t. dR = 6.0. Feel free to study. (6) Instantiation of vertex-mass-constraint classes outside event loop [code speedup, memory leaks(?) fixed] (7) Vertex fitters are now Clear()-ed every time they are used (used to be at end of events; was this a bug???) (8) Replaced GetEntries() by GetEntriesFast() for d0root-related arrays, collections, lists (code speedup) (9) Bug-fix: Clearing pv-finder before storing event-info to root-tuple causes occasional crashes Nov19-20, 03 (christos) v00-00-50 ======================================================================== Increased jet cone size to: float jetcone_cuts[8] = { 7.0, // r-phi cone size 2.0, // Z cone size 0.15, // pT cut 0.5, // min track-jet pT 1.5, // pT track seed 1, // hits 0.40,// dca 0.75 // zdca}; MC Efficiency w.r.t v23 is now > 99.5% Swapped findPrimaryVertex() and findMuons(); code does not look for PV in events where there are no muons (much faster to determine than PV) *** Code runs four times faster on data w.r.t. v23! *** I think there is still room for improvement. I want to tune the jet parameters on data (hoping to make the jet cone smaller and limit the D+mu search in only one hemisphere; with delta-R as large as 7, the whole "jet" concept comes down to a simple "track clustering" around the z-axis - we should be able to do better than that). Added Njet_trks & Njet_trks_soft to root-tuple Added timing measurements to Make.C Nov18, 03 (christos) v00-00-30 ======================================================================== Added jet-finder algorithm; D+Mu finders now loop over tracks clustered in jets (instead of all event's tracks); initial cone choice defined with float jetcone_cuts[8] = { 3.0, // r-phi cone size 2.0, // Z cone size 0.15, // pT cut 0.5, // min track-jet pT 1.0, // pT track seed 1, // hits 0.25,// dca 0.40 // zdca }; Code speed-up by 30% [tested on MC, mode: Ds(Phi+Pi)+mu] With mass cuts tuned on p14 data, v30 has about 15% less events than v23 for MC sample. Need to investigate. Nov18, 03 (christos) v00-00-23 ======================================================================== fixed bug in d0root_btag/algorithms/PrimaryVertexGlobalFinder2Pass.cpp (v8.0); updated script to replace file after installation Nov17, 03 (christos) v00-00-22 ======================================================================== bugfix: now calling MBPSelector::Clear at the end of event Nov17, 03 (christos) v00-00-21 ======================================================================== bugfix: added protection against null "most probable PV" pointer Nov13-16, 03 (christos) v00-00-20 ======================================================================== Moving from d0root v7.2 to v8.0: a. Now using new probabilistic class for best PV determination b. new 2-pass PV finder class (replaced custom version of two-pass PV finder) c. updated interface to d0root methods (new names, etc) increased pT cut for tracks in B reconstruction from 0.5 GeV to 0.6 GeV; removed all pT cuts from soft pion in D* --> D + pi reconstruction (there used to be a 0.5 GeV cut on slow pion); look for D* if D-mass < 1.95 GeV (used to be: between [1.75, 1.95] GeV) added beamspot for sets 22-24 added hitmasks, track errors & chi2 for all B tracks to roottuple updated "get-tag" piece of code for new bflavor_tag interface (ie. list of muons v00-00-32 --> list of charged tracks v00-00-40) bug fix: wrong array length for KPiStar branch (relevant only for D* --> D + pi mode) removed what seemed irrelevant include's in Makefile Sep16-18, 03 (christos) v00-00-15 ======================================================================== code now gives the option of applying extra cuts on tracks used for PV determination (abbott); implemented in new method getPVtracks (use switch "SELECT_PV_TRACKS" - turned off by default) added to root-tuple: PV's chi^2, delta-R(muon, charged track), hitmasks for charged tracks, beam spot information (using beam-spot database - provided by lwelty@fnal.gov) added doc-maker scripts code now adds package version to title of output root-tuple TTree Sep11, 03 (christos) v00-00-11 ======================================================================== updated documentation Sep10, 03 (christos) v00-00-10 ======================================================================== first implementation of code as independent CVS package August 03 (christos) (not in CVS) ======================================================================== decoupled primary vertex finder, event reader, muon finder from K+Pi, Phi+Pi finders