Repository navigation
[PWGDQ] Update table with properties for prompt-non prompt J/psi separation - #18370
JonathanWitte wants to merge 1 commit into
Conversation
… add FT0C centrality value, and setup stable filling outside the cut loop
|
O2 linter results: ❌ 0 errors, |
| dqanalysisflags::MassDileptonCandidate, dqanalysisflags::Ptpair, dqanalysisflags::Etapair, dqanalysisflags::Ptassoc, dqanalysisflags::Etaassoc, dqanalysisflags::Phiassoc, | ||
| dqanalysisflags::Ptleg1, dqanalysisflags::Etaleg1, dqanalysisflags::Phileg1, dqanalysisflags::Ptleg2, dqanalysisflags::Etaleg2, dqanalysisflags::Phileg2); | ||
| DECLARE_SOA_TABLE(JPsieeCandidates, "AOD", "DQPSEUDOPROPER", | ||
| // bookkeeping / labels |
There was a problem hiding this comment.
Remove comments, they just blow up the code, in general, if the namings are unique enough, no extra comments are needed
Same for the lines below
| (fConfigML.applyBDT && fOutputMlPsi2ee.size() > 2) ? VarManager::fgValues[VarManager::kBdtNonprompt] : -999.f); | ||
| } | ||
| // flat table for prompt / non-prompt BDT training: one row per pair (not per cut; select cuts offline via PairFilter) | ||
| // Only meaningful with track covariances, the two-prong fitter and an event vertex covariance: |
There was a problem hiding this comment.
Such calculations shouldnt be done directly here, but rather in the VarManager. So introduce new Variables kDCASigxy1, kDCASigxy2, kDCASigz1, kDCASigz1 and fill values in VarManager::FillPair and VarManager::FillPairVertexing, similar lxyOverErr and lxyzOverErr
|
|
||
| // decay-length significances: kVertexingLxyOverErr / kVertexingLxyzOverErr are only filled in the | ||
| // KFParticle branch of FillPairVertexing, not with the DCAFitter used here -> compute them explicitly | ||
| auto ratio = [](float num, float den) -> float { |
There was a problem hiding this comment.
This check here too is not necessary. You divide by an error which can not be negative
| if (t1.sign() * t2.sign() < 0 && vertexingOK) { | ||
|
|
||
| // track DCA significances (DCA w.r.t. the collision the track was stored with, as kDCAxy1/kDCAz1) | ||
| auto dcaSig = [](float dca, float cov) -> float { |
There was a problem hiding this comment.
Remove this function, track.cYY (etc.) shouldnt be negative so the check is not necessary
| @@ -2293,10 +2375,6 @@ struct AnalysisSameEventPairing { | |||
| } | |||
| } | |||
| if (sign1 * sign2 < 0) { | |||
There was a problem hiding this comment.
Why introduce the check above again if it is already here?
| const bool ambiOutOfBunch = (twoTrackFilter & (static_cast<uint32_t>(1) << 30)) || (twoTrackFilter & (static_cast<uint32_t>(1) << 31)); | ||
| // kVertexingProcCode == 0: the fit failed; FillPairVertexing then returns early without updating the | ||
| // projected / pseudo-proper / pointing-angle variables, so they would be stale -> skip the pair | ||
| const bool vertexingOK = VarManager::fgValues[VarManager::kVertexingProcCode] > 0; |
There was a problem hiding this comment.
Check here is also unneccessary, it is done already in the VarManager
2 participants
Thorough review and upgrade of the table used for P/NP separation
PromptNonPromptSepTable / o2::aod::JPsieeCandidates / "DQPSEUDOPROPER"
Recalculating missing properties
Removing TPC track quality properties due to MC mismatch
Add FT0C centrality in table
Move table fill outside of cut loop and guard missing options with condition