Major KFParticle alteration#4332
Conversation
|
Warning Review limit reached
Next review available in: 35 minutes Enable usage-based reviews in Billing to review now. Otherwise, wait until the next included review is available. How can I continue?After more reviews become available, a review can be triggered using the To avoid repeated limits, reduce automatic review volume by pausing incremental auto-reviews earlier, using label-based review opt-in, excluding WIP or generated PR titles, or requesting reviews manually when the PR is ready. If your team needs uninterrupted high-volume reviews, an organization admin can enable usage-based reviews. How do review limits work?CodeRabbit enforces per-developer PR review limits for each organization. Most developers receive the normal plan review availability. For paid Pro and Pro+ PR reviews, CodeRabbit uses adaptive limits for sustained high-volume activity. When a developer's recent PR review activity reaches the 95th percentile or higher among CodeRabbit users, additional reviews become available more gradually as earlier reviews age out of the rolling window. Please refer docs for additional details. Review details⚙️ Run configurationConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (3)
📝 WalkthroughWalkthroughHFTrackEfficiency migrates truth-to-reco track matching from PHG4ParticleSvtxMap to SvtxEvalStack/SvtxTrackEval. KFParticle_sPHENIX replaces IP/IPchi2-based track, mother, and intermediate-state selection, constraint setters, and ntuple output with primary-vertex DCA (PV_DCA) based quantities throughout Tools, eventReconstruction, nTuple, sPHENIX, and truthAndDetTools files. ChangesHFTrackEfficiency SvtxEval Migration
PV-DCA Based Selection Replacing IP-Based Selection
Sequence Diagram(s)sequenceDiagram
participant process_event
participant SvtxEvalStack
participant findTracks
participant SvtxTrackEval
process_event->>SvtxEvalStack: next_event(topNode)
process_event->>findTracks: process daughters
findTracks->>findTracks: search PHG4TruthInfoContainer for daughterG4
findTracks->>SvtxTrackEval: best_track_from(daughterG4)
SvtxTrackEval-->>findTracks: matched reco track
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 8
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: cb632a3a-81b5-45b4-8525-5a3c8f4e005c
📒 Files selected for processing (11)
offline/packages/HFTrackEfficiency/HFTrackEfficiency.ccoffline/packages/HFTrackEfficiency/HFTrackEfficiency.hoffline/packages/HFTrackEfficiency/Makefile.amoffline/packages/KFParticle_sPHENIX/KFParticle_Tools.ccoffline/packages/KFParticle_sPHENIX/KFParticle_Tools.hoffline/packages/KFParticle_sPHENIX/KFParticle_eventReconstruction.ccoffline/packages/KFParticle_sPHENIX/KFParticle_nTuple.ccoffline/packages/KFParticle_sPHENIX/KFParticle_nTuple.hoffline/packages/KFParticle_sPHENIX/KFParticle_sPHENIX.hoffline/packages/KFParticle_sPHENIX/KFParticle_truthAndDetTools.ccoffline/packages/KFParticle_sPHENIX/KFParticle_truthAndDetTools.h
| PHG4Particle *daughterG4 {nullptr}; | ||
|
|
||
| for (unsigned int i = 1; i < decay.size(); ++i) | ||
| { | ||
| m_dst_track = nullptr; | ||
| int truth_ID = -1; | ||
| //int truth_ID = -1; |
There was a problem hiding this comment.
🎯 Functional Correctness | 🔴 Critical | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
file='offline/packages/HFTrackEfficiency/HFTrackEfficiency.cc'
echo '--- file size ---'
wc -l "$file"
echo '--- outline ---'
ast-grep outline "$file" --view expanded || true
echo '--- relevant sections ---'
sed -n '230,470p' "$file" | cat -n
echo '--- search truth_ID declarations/usages ---'
rg -n '\btruth_ID\b' "$file" || trueRepository: sPHENIX-Collaboration/coresoftware
Length of output: 11893
🏁 Script executed:
#!/bin/bash
set -euo pipefail
file='offline/packages/HFTrackEfficiency/HFTrackEfficiency.cc'
echo '--- around line 390-445 ---'
sed -n '390,445p' "$file" | cat -n
echo '--- whole repo search for truth_ID ---'
rg -n '\btruth_ID\b' offline/packages/HFTrackEfficiency || true
echo '--- class member search for trackeval / truth-reco map context ---'
rg -n 'class HFTrackEfficiency|truth_ID|m_dst_truth_reco_map|m_truthInfo|trackeval' offline/packages/HFTrackEfficiency -g '!*.o' || trueRepository: sPHENIX-Collaboration/coresoftware
Length of output: 7187
Reset daughterG4 per daughter and restore truth_ID
daughterG4needs to be cleared at the start of each daughter iteration, and them_nDaughters == 2scan shouldbreakon the first match; otherwisetrackeval->best_track_from(daughterG4)can use the last truth particle seen instead of the matched daughter.truth_IDis still assigned in them_nDaughters == 3path, but its declaration is commented out in this scope, so this file will not compile.
| //if (m_dst_truth_reco_map) | ||
| //{ | ||
| PHG4TruthInfoContainer::ConstRange range = m_truthInfo->GetParticleRange(); | ||
|
|
||
| for (PHG4TruthInfoContainer::ConstIterator iter = range.first; iter != range.second; ++iter) | ||
| { | ||
| PHG4TruthInfoContainer::ConstRange range = m_truthInfo->GetParticleRange(); | ||
| //PHG4Particle *daughterG4 = iter->second; | ||
| daughterG4 = iter->second; | ||
|
|
||
| for (PHG4TruthInfoContainer::ConstIterator iter = range.first; iter != range.second; ++iter) | ||
| if (std::abs(daughterG4->get_px() - daughterTrueLV->x()) <= 5e-3 && | ||
| std::abs(daughterG4->get_py() - daughterTrueLV->y()) <= 5e-3 && | ||
| std::abs(daughterG4->get_pz() - daughterTrueLV->z()) <= 5e-3 && daughterG4->get_pid() == decay[i].second) | ||
| { | ||
| PHG4Particle *daughterG4 = iter->second; | ||
|
|
||
| if (std::abs(daughterG4->get_px() - daughterTrueLV->x()) <= 5e-3 && | ||
| std::abs(daughterG4->get_py() - daughterTrueLV->y()) <= 5e-3 && | ||
| std::abs(daughterG4->get_pz() - daughterTrueLV->z()) <= 5e-3 && daughterG4->get_pid() == decay[i].second) | ||
| { | ||
| truth_ID = daughterG4->get_track_id(); | ||
| break; | ||
| } | ||
| //truth_ID = daughterG4->get_track_id(); | ||
| break; | ||
| } | ||
| } | ||
| //} | ||
| } | ||
| else | ||
| { | ||
| PHG4TruthInfoContainer::ConstRange range = m_truthInfo->GetParticleRange(); | ||
|
|
||
| for (PHG4TruthInfoContainer::ConstIterator iter = range.first; iter != range.second; ++iter) | ||
| { | ||
| PHG4Particle *daughterG4 = iter->second; | ||
| //PHG4Particle *daughterG4 = iter->second; | ||
| daughterG4 = iter->second; |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
Possible unguarded null dereference of m_truthInfo.
m_truthInfo is allowed to remain nullptr (only a warning is logged at line 108-115, no abort). This branch now unconditionally calls m_truthInfo->GetParticleRange() where previously (per the summary) this path was gated by if (m_dst_truth_reco_map). If a "G4TruthInfo" node is ever missing on an event, this will crash.
| SvtxEvalStack *m_svtx_evalstack{nullptr}; | ||
| SvtxTrackEval *trackeval{nullptr}; | ||
| //PHG4ParticleSvtxMap *m_dst_truth_reco_map{nullptr}; |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Map the target header and implementation first.
git ls-files 'offline/packages/HFTrackEfficiency/*' | sed -n '1,20p'
echo
echo '--- header outline ---'
ast-grep outline offline/packages/HFTrackEfficiency/HFTrackEfficiency.h --view expanded
echo
echo '--- implementation outline ---'
ast-grep outline offline/packages/HFTrackEfficiency/HFTrackEfficiency.cc --view expanded
echo
echo '--- relevant header lines ---'
sed -n '1,160p' offline/packages/HFTrackEfficiency/HFTrackEfficiency.h | nl -ba | sed -n '1,140p'
echo
echo '--- relevant implementation lines ---'
sed -n '1,260p' offline/packages/HFTrackEfficiency/HFTrackEfficiency.cc | nl -ba | sed -n '1,240p'Repository: sPHENIX-Collaboration/coresoftware
Length of output: 691
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Check for any destructor / cleanup / ownership-related code paths.
rg -n "HFTrackEfficiency::~HFTrackEfficiency|delete m_svtx_evalstack|m_svtx_evalstack|unique_ptr|SvtxEvalStack" offline/packages/HFTrackEfficiencyRepository: sPHENIX-Collaboration/coresoftware
Length of output: 774
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo '--- header excerpt ---'
sed -n '1,140p' offline/packages/HFTrackEfficiency/HFTrackEfficiency.h | cat -n
echo
echo '--- implementation excerpt ---'
sed -n '100,140p' offline/packages/HFTrackEfficiency/HFTrackEfficiency.cc | cat -nRepository: sPHENIX-Collaboration/coresoftware
Length of output: 7731
Replace m_svtx_evalstack with RAII ownership
m_svtx_evalstack is allocated with new in process_event, and HFTrackEfficiency only has a default destructor, so the stack is never released. Please switch this member to std::unique_ptr<SvtxEvalStack> (and include <memory>) to make ownership explicit and avoid a job-lifetime leak.
Source: Path instructions
| m_calculated_mother_PV_dca = motherParticle.GetDistanceFromVertex(vertex_fillbranch); | ||
| m_calculated_mother_PV_dca_sig = motherParticle.GetDeviationFromVertex(vertex_fillbranch); | ||
| m_calculated_mother_PV_dca_err = m_calculated_mother_PV_dca / std::sqrt(m_calculated_mother_PV_dca_sig); | ||
| m_calculated_mother_PV_dca_xy = motherParticle.GetDistanceFromVertexXY(vertex_fillbranch); |
There was a problem hiding this comment.
🗄️ Data Integrity & Integration | 🟠 Major | ⚡ Quick win
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Map the relevant file and nearby symbols
ast-grep outline offline/packages/KFParticle_sPHENIX/KFParticle_nTuple.cc --view expanded || true
# Locate GetDeviationFromVertex usage and related helpers
rg -n "GetDeviationFromVertex|PV_dca_err|PV_dca_sig|constrainToVertex" offline/packages/KFParticle_sPHENIX -S
# Read the relevant sections with line numbers
sed -n '390,525p' offline/packages/KFParticle_sPHENIX/KFParticle_nTuple.cc
# Inspect the helper file mentioned in the review comment
if [ -f offline/packages/KFParticle_sPHENIX/KFParticle_Tools.cc ]; then
sed -n '1,220p' offline/packages/KFParticle_sPHENIX/KFParticle_Tools.cc
fiRepository: sPHENIX-Collaboration/coresoftware
Length of output: 20101
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Inspect any local KFParticle headers or vendored API docs for GetDeviationFromVertex
fd -a "KFParticle*.h" offline packages . 2>/dev/null | sed -n '1,120p'
rg -n "GetDeviationFromVertex\\s*\\(" -S .Repository: sPHENIX-Collaboration/coresoftware
Length of output: 2657
🌐 Web query:
KFParticle GetDeviationFromVertex returns sqrt(Chi2/NDF) documentation source code
💡 Result:
The KFParticle library, commonly used in high-energy physics experiments like sPHENIX and STAR, defines the function GetDeviationFromVertex to calculate the deviation of a particle from a vertex in units of sqrt(Chi2/NDF) [1][2][3]. Documentation and source code indicate that this function calculates a normalized distance (or deviation) from a specified vertex (or another particle) [1][3][4]. The function is defined in both the KFParticle and KFParticleBase classes [2][3][5]. Key details from the source code and documentation: 1. Purpose: The function is explicitly documented as a method to "Calculate sqrt(Chi2/ndf) deviation from vertex" [3][4]. 2. Signature: It is typically available as overloads accepting a vertex position array (and an optional covariance matrix) or another KFParticle object [2][5][6]. Example signature (from KFParticleBase.h): float GetDeviationFromVertex( const float v[], const float Cv[]=0) const; float GetDeviationFromVertex( const KFParticleBase &Vtx) const; 3. Context: The deviation is calculated using the particle's state and the provided vertex's position [v] and optional covariance matrix [Cv] [3][6]. Authoritative sources for this documentation include: - The sPHENIX Collaboration's Doxygen documentation and code browser for KFParticle [7][3][8]. - The STAR Experiment's (StRoot) KFParticle class documentation and source files [2][9][6]. - The PandaRoot documentation [5]. You can find the implementation and header files in the KFParticle repository on GitHub or via the specific collaboration's software repositories [1][3][10].
Citations:
- 1: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFParticle.h
- 2: https://www.star.bnl.gov/webdata/dox/html/classKFParticle.html
- 3: https://sphenix-code-browser.sdcc.bnl.gov/lxr/source/KFParticle/KFParticle/KFParticleBase.h
- 4: https://git.jinr.ru/vvtroshin/bmnroot-sc-wall/-/blob/dev/reconstruction/KF/KFParticle/KFParticleBaseSIMD.h
- 5: http://webdocs.gsi.de/~pandacc/documentation/2022-12-23-dev/doxygen/classKFParticle.html
- 6: https://www.star.bnl.gov/webdata/dox/html/KFParticle_8h_source.html
- 7: https://sphenix-collaboration.github.io/doxygen/d0/d49/classKFParticle.html
- 8: https://sphenix-code-browser.sdcc.bnl.gov/lxr/source/KFParticle/KFParticle/KFParticle.h
- 9: https://www.star.bnl.gov/webdata/dox/html/KFParticle_8cxx_source.html
- 10: https://git.jinr.ru/myktybekov/spdroot/-/blob/6457aa343c4af0bf8640b0adb368964819e0d2de/external/KFParticle/KFParticle/KFParticleBase.h
🌐 Web query:
KFParticle GetDeviationFromVertex sqrt Chi2 ndf source code
💡 Result:
In the KFParticle software package, the GetDeviationFromVertex method is a function designed to calculate the statistical deviation of a particle from a given vertex [1][2][3]. It represents the square root of the normalized chi-square value (sqrt(Chi2/ndf)), providing a measure of how far a particle is from a vertex in terms of standard deviations [1][2][4][5]. Key Technical Details: - Mathematical Basis: The function computes sqrt(Chi2/ndf), where Chi2 is the chi-square fit value and ndf is the number of degrees of freedom [1][6][7]. - Function Purpose: It is commonly used in vertex reconstruction algorithms to assess the compatibility of a particle track with a reconstructed primary or secondary vertex [8][9]. - Implementation: The method exists within the KFParticle class (and its base class, KFParticleBase) [3][4][5]. It typically leverages the GetDistanceFromVertexXY or GetDistanceFromVertex methods to calculate the underlying chi-square, which is then divided by the number of degrees of freedom and square-rooted [2][6]. - Variations: There are variations for 3D space (GetDeviationFromVertex) and 2D projections (GetDeviationFromVertexXY), as well as overloads that accept different types of vertex representations (e.g., arrays, KFParticle objects, or KFPVertex objects) [1][3][10][5]. You can find the implementation in the source files KFParticle.cxx and KFParticleBase.h within the KFParticle repository [2][6][4][11]. The logic usually follows this pattern: 1. Calculate the distance and covariance from the vertex [2][6]. 2. Compute the chi-square difference [2][6]. 3. Return the square root of that value divided by the degrees of freedom [1][2].
Citations:
- 1: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFParticle.h
- 2: https://pandaroot.hiskp.uni-bonn.de/dev/KFParticle_8cxx_source.html
- 3: https://sphenix-collaboration.github.io/doxygen/d0/d49/classKFParticle.html
- 4: https://sphenix-code-browser.sdcc.bnl.gov/lxr/source/KFParticle/KFParticle/KFParticleBase.h
- 5: http://webdocs.gsi.de/~pandacc/documentation/2022-12-23-dev/doxygen/classKFParticle.html
- 6: https://www.star.bnl.gov/webdata/dox/html/KFParticle_8cxx_source.html
- 7: https://pandaroot.hiskp.uni-bonn.de/dev/KFParticle_8h_source.html
- 8: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFParticlePVReconstructor.cxx
- 9: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFVertex.cxx
- 10: https://www.star.bnl.gov/webdata/dox/html/classKFParticle.html
- 11: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFParticle.cxx
🏁 Script executed:
#!/bin/bash
set -euo pipefail
sed -n '1138,1156p' offline/packages/KFParticle_sPHENIX/KFParticle_Tools.cc
sed -n '1498,1518p' offline/packages/KFParticle_sPHENIX/KFParticle_truthAndDetTools.ccRepository: sPHENIX-Collaboration/coresoftware
Length of output: 2155
🌐 Web query:
"GetDeviationFromVertex" KFParticle documentation stddev
💡 Result:
In the KFParticle library, the method GetDeviationFromVertex (and its variant GetDeviationFromVertexXY) calculates the square root of the Chi-squared per degree of freedom (sqrt(Chi2/ndf)) representing the deviation of a particle from a given vertex [1][2][3][4]. This value is used to quantify the distance between the particle and the vertex in units of their combined errors (statistical significance of the separation) [5]. It is not a standard deviation in the sense of a simple Gaussian width, but rather a pull-like metric evaluating how compatible the particle's trajectory is with originating from that specific vertex, accounting for the covariance matrices of both the particle and the vertex [1][4][5]. Key details include: - Functionality: It computes a statistical distance (sqrt(Chi2/ndf)) [2][3]. - Inputs: It typically accepts either a raw array representing vertex coordinates and an optional covariance matrix, or an existing KFParticle/KFPVertex object [1][6][7]. - Related Methods: The library also provides GetDistanceFromVertex, which calculates the simple spatial distance in centimeters, distinct from the statistical deviation provided by GetDeviationFromVertex [1][4][5].
Citations:
- 1: https://github.com/sPHENIX-Collaboration/KFParticle/blob/master/KFParticle/KFParticle.h
- 2: https://sphenix-collaboration.github.io/doxygen/d0/d4c/KFParticleBase_8h_source.html
- 3: https://sphenix-collaboration.github.io/doxygen/d6/db1/KFParticle_8h_source.html
- 4: https://sphenix-code-browser.sdcc.bnl.gov/lxr/source/KFParticle/KFParticle/KFParticleBase.h
- 5: https://sphenix-code-browser.sdcc.bnl.gov/lxr/source/KFParticle/KFParticle/KFParticle.h
- 6: https://www.star.bnl.gov/webdata/dox/html/classKFParticle.html
- 7: http://webdocs.gsi.de/~pandacc/documentation/2022-12-23-dev/doxygen/classKFParticle.html
Fix the *_PV_dca_err calculation
GetDeviationFromVertex() already returns the deviation, so dividing by std::sqrt(*_PV_dca_sig) is incorrect. Use distance / *_PV_dca_sig in the mother, intermediate, and daughter blocks.
| auto minmax_PV_dca = minmax_element(ip.begin(), ip.end()); // Order the PV_DCA from small to large | ||
| minimumPV_DCA = *minmax_PV_dca.first; | ||
| auto minmax_PV_dca_stddev = minmax_element(ip_significance.begin(), ip_significance.end()); // Order the PV_DCA chi2 from small to large | ||
| minimumPV_DCA_stddev = *minmax_PV_dca_stddev.first; |
There was a problem hiding this comment.
🩺 Stability & Availability | 🟠 Major | ⚡ Quick win
Guard against an empty PVs collection.
If PVs is empty, ip/ip_significance stay empty and minmax_element returns {end, end}; dereferencing *minmax_PV_dca.first on line 517/519 is undefined behavior (likely a crash). Events with no reconstructed primary vertex would hit this path.
🛡️ Suggested guard
+ if (ip.empty())
+ {
+ minimumPV_DCA = std::numeric_limits<float>::max();
+ minimumPV_DCA_stddev = std::numeric_limits<float>::max();
+ return -1;
+ }
+
auto minmax_PV_dca = minmax_element(ip.begin(), ip.end()); // Order the PV_DCA from small to large
minimumPV_DCA = *minmax_PV_dca.first;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| auto minmax_PV_dca = minmax_element(ip.begin(), ip.end()); // Order the PV_DCA from small to large | |
| minimumPV_DCA = *minmax_PV_dca.first; | |
| auto minmax_PV_dca_stddev = minmax_element(ip_significance.begin(), ip_significance.end()); // Order the PV_DCA chi2 from small to large | |
| minimumPV_DCA_stddev = *minmax_PV_dca_stddev.first; | |
| if (ip.empty()) | |
| { | |
| minimumPV_DCA = std::numeric_limits<float>::max(); | |
| minimumPV_DCA_stddev = std::numeric_limits<float>::max(); | |
| return -1; | |
| } | |
| auto minmax_PV_dca = minmax_element(ip.begin(), ip.end()); // Order the PV_DCA from small to large | |
| minimumPV_DCA = *minmax_PV_dca.first; | |
| auto minmax_PV_dca_stddev = minmax_element(ip_significance.begin(), ip_significance.end()); // Order the PV_DCA chi2 from small to large | |
| minimumPV_DCA_stddev = *minmax_PV_dca_stddev.first; |
…tion.cc Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
Co-authored-by: coderabbitai[bot] <136622811+coderabbitai[bot]@users.noreply.github.com>
|
This PR actually drastically improves the timing. Probably as I reject combinations from different bunch crossings earlier 
|
1 participant
Types of changes
What kind of change does this PR introduce? (Bug fix, feature, ...)
As was discussed at yesterdays TG meeting, it is preferable to perform track DCA, pT and IP checks after a secondary vertex has been reconstructed. The act of creating a mother particle and associating the tracks to it causes a refit of the track which means the output nTuple variables are different to the requested cuts. This means that you can have candidates in your output that would fail the selection.
I rewrote significant chunks of the initial reconstruction to allow this. I also moved the check that tracks come from the same bunch crossing to earlier in the code in an attempt to speed up event processing. I'll post timing plots later.
I also renamed the IP branches and selections to PV_DCA as this feels less confusing. IPchi2 is now PV_DCA_StdDev to reflect that the number is actually the number of standard deviations away from the PV.
Finally, I ran the code through codex to identify logic errors. It found a few that I have patched out.
@jdosbo @adfrawley @petermic @hupereir this should cover the suggestions
TODOs (if applicable)
Links to other PRs in macros and calibration repositories (if applicable)
AI can make mistakes, so please use best judgment when reading this summary.
This PR refactors KFParticle reconstruction and output bookkeeping to use PV-based DCA quantities instead of the previous IP/IPchi2 selection logic, and also improves truth-matching and track association in HFTrackEfficiency.
Motivation / context
PV_DCA,PV_DCA_StdDev) better reflects the physics meaning of the quantities.Key changes
KFParticle_sPHENIXKFParticle_ToolsKFParticle_nTupleKFParticle_truthAndDetToolsprimaryVerticesthrough prong-finding and combination building.PHG4ParticleSvtxMap-style lookup toSvtxEvalStack/SvtxTrackEvalbest_track_from(...)for reco associationsetMinTPOThitstarget member).Potential risk areas
Possible future improvements