|
| 1 | +// Copyright 2019-2020 CERN and copyright holders of ALICE O2. |
| 2 | +// See https://alice-o2.web.cern.ch/copyright for details of the copyright holders. |
| 3 | +// All rights not expressly granted are reserved. |
| 4 | +// |
| 5 | +// This software is distributed under the terms of the GNU General Public |
| 6 | +// License v3 (GPL Version 3), copied verbatim in the file "COPYING". |
| 7 | +// |
| 8 | +// In applying this license CERN does not waive the privileges and immunities |
| 9 | +// granted to it by virtue of its status as an Intergovernmental Organization |
| 10 | +// or submit itself to any jurisdiction. |
| 11 | +/// |
| 12 | +/// \file alice3PidEvaluation.cxx |
| 13 | +/// |
| 14 | +/// \brief This task computes purity and efficiency from the OTF PID tables for multiple detectors. |
| 15 | +/// Analyzes individual detectors (Tracker, TOF Inner, TOF Outer, RICH), |
| 16 | +/// as well as combined detector performance using quadrature combination |
| 17 | +/// of nSigma values. |
| 18 | +/// |
| 19 | +/// \author Henrik Fribert TUM |
| 20 | +/// \since August 14, 2025 |
| 21 | +/// |
| 22 | + |
| 23 | +#include "ALICE3/DataModel/OTFPIDTrk.h" |
| 24 | +#include "ALICE3/DataModel/OTFRICH.h" |
| 25 | +#include "ALICE3/DataModel/OTFTOF.h" |
| 26 | +#include "Common/DataModel/TrackSelectionTables.h" |
| 27 | + |
| 28 | +#include "CommonUtils/NameConf.h" |
| 29 | +#include "Framework/ASoAHelpers.h" |
| 30 | +#include "Framework/AnalysisDataModel.h" |
| 31 | +#include "Framework/AnalysisTask.h" |
| 32 | +#include "Framework/HistogramRegistry.h" |
| 33 | +#include "Framework/RunningWorkflowInfo.h" |
| 34 | +#include "Framework/runDataProcessing.h" |
| 35 | +#include "ReconstructionDataFormats/DCA.h" |
| 36 | + |
| 37 | +#include "TH1F.h" |
| 38 | +#include "TH2F.h" |
| 39 | +#include "TProfile.h" |
| 40 | +#include "TVector3.h" |
| 41 | + |
| 42 | +#include <array> |
| 43 | +#include <vector> |
| 44 | + |
| 45 | +using namespace o2; |
| 46 | +using namespace o2::framework; |
| 47 | + |
| 48 | +struct Alice3PidEvaluation { |
| 49 | + |
| 50 | + HistogramRegistry histos{"histos", {}, OutputObjHandlingPolicy::AnalysisObject}; |
| 51 | + |
| 52 | + static constexpr float kInvalidNSigmaValue = 999.0f; |
| 53 | + |
| 54 | + Configurable<float> maxNSigmaForIdentification{"maxNSigmaForIdentification", 3.0f, "Maximum |nSigma| allowed for particle identification (closest-hypothesis rule)"}; |
| 55 | + Configurable<int> numLogBins{"numLogBins", 200, "Number of logarithmic momentum bins"}; |
| 56 | + Configurable<bool> useClosestHypothesisRule{"useClosestHypothesisRule", true, "Use closest-hypothesis rule: assign track to hypothesis with smallest |nSigma|"}; |
| 57 | + Configurable<bool> useMinimalIdentification{"useMinimalIdentification", false, "Require that only one hypothesis is within the cutoff"}; |
| 58 | + Configurable<bool> includeTrackerInCombined{"includeTrackerInCombined", true, "Include Tracker in combined analysis"}; |
| 59 | + Configurable<bool> includeTofInnerInCombined{"includeTofInnerInCombined", true, "Include TOF Inner in combined analysis"}; |
| 60 | + Configurable<bool> includeTofOuterInCombined{"includeTofOuterInCombined", true, "Include TOF Outer in combined analysis"}; |
| 61 | + Configurable<bool> includeRichInCombined{"includeRichInCombined", false, "Include RICH in combined analysis"}; |
| 62 | + |
| 63 | + std::vector<double> mLogBins; |
| 64 | + |
| 65 | + enum PidHypothesis { kElectron, |
| 66 | + kMuon, |
| 67 | + kPion, |
| 68 | + kKaon, |
| 69 | + kProton, |
| 70 | + kDeuteron, |
| 71 | + kTriton, |
| 72 | + kHelium3, |
| 73 | + kAlpha, |
| 74 | + kCount }; |
| 75 | + static constexpr std::array<int, PidHypothesis::kCount> kHypothesisPdg = {11, 13, 211, 321, 2212, 1000010020, 1000010030, 1000020030, 1000020040}; |
| 76 | + static constexpr std::array<const char*, PidHypothesis::kCount> kHypothesisNames = {"Electron", "Muon", "Pion", "Kaon", "Proton", "Deuteron", "Triton", "Helium3", "Alpha"}; |
| 77 | + |
| 78 | + struct DetectorHistograms { |
| 79 | + std::array<std::shared_ptr<TH1>, PidHypothesis::kCount> hTotalTrue; |
| 80 | + std::array<std::shared_ptr<TProfile>, PidHypothesis::kCount> hEfficiency; |
| 81 | + std::array<std::shared_ptr<TProfile>, PidHypothesis::kCount> hPurityAsHypothesis; |
| 82 | + }; |
| 83 | + |
| 84 | + std::shared_ptr<TH2> hDetectorParticipation2D; |
| 85 | + |
| 86 | + DetectorHistograms trackerHists; |
| 87 | + DetectorHistograms tofInnerHists; |
| 88 | + DetectorHistograms tofOuterHists; |
| 89 | + DetectorHistograms richHists; |
| 90 | + DetectorHistograms combinedHists; |
| 91 | + DetectorHistograms combinedNoTrackerHists; |
| 92 | + void init(o2::framework::InitContext&) |
| 93 | + { |
| 94 | + LOG(info) << "Initializing multi-detector PID evaluation using closest-hypothesis rule"; |
| 95 | + LOG(info) << "Maximum |nSigma| for identification: " << maxNSigmaForIdentification.value; |
| 96 | + LOG(info) << "Closest-hypothesis rule: " << (useClosestHypothesisRule.value ? "ENABLED" : "DISABLED"); |
| 97 | + LOG(info) << "Require unique identification: " << (useMinimalIdentification.value ? "YES" : "NO"); |
| 98 | + LOG(info) << "Combined analysis includes: " |
| 99 | + << (includeTrackerInCombined.value ? "Tracker " : "") |
| 100 | + << (includeTofInnerInCombined.value ? "TOF_Inner " : "") |
| 101 | + << (includeTofOuterInCombined.value ? "TOF_Outer " : "") |
| 102 | + << (includeRichInCombined.value ? "RICH " : ""); |
| 103 | + |
| 104 | + mLogBins.clear(); |
| 105 | + double pMin = 0.05; |
| 106 | + double pMax = 10; |
| 107 | + double logMin = std::log10(pMin); |
| 108 | + double logMax = std::log10(pMax); |
| 109 | + double dLog = (logMax - logMin) / numLogBins.value; |
| 110 | + for (int i = 0; i <= numLogBins.value; ++i) { |
| 111 | + mLogBins.push_back(std::pow(10, logMin + i * dLog)); |
| 112 | + } |
| 113 | + const AxisSpec axisMomentum{mLogBins, "#it{p} (GeV/#it{c})"}; |
| 114 | + |
| 115 | + auto createDetectorHistograms = [&](DetectorHistograms& detHists, const std::string& detectorName) { |
| 116 | + for (int trueIdx = 0; trueIdx < PidHypothesis::kCount; ++trueIdx) { |
| 117 | + const auto& trueName = kHypothesisNames[trueIdx]; |
| 118 | + detHists.hTotalTrue[trueIdx] = histos.add<TH1>(Form("%s/hTotalTrue%s", detectorName.c_str(), trueName), |
| 119 | + Form("%s: Total True %s; #it{p} (GeV/#it{c})", detectorName.c_str(), trueName), |
| 120 | + kTH1F, {axisMomentum}); |
| 121 | + detHists.hEfficiency[trueIdx] = histos.add<TProfile>(Form("%s/hEfficiency%s", detectorName.c_str(), trueName), |
| 122 | + Form("%s: PID Efficiency for %s; #it{p} (GeV/#it{c}); Efficiency", detectorName.c_str(), trueName), |
| 123 | + kTProfile, {axisMomentum}); |
| 124 | + } |
| 125 | + |
| 126 | + for (int hypIdx = 0; hypIdx < PidHypothesis::kCount; ++hypIdx) { |
| 127 | + const auto& hypName = kHypothesisNames[hypIdx]; |
| 128 | + detHists.hPurityAsHypothesis[hypIdx] = histos.add<TProfile>(Form("%s/hPurityAs%s", detectorName.c_str(), hypName), |
| 129 | + Form("%s: Purity when selecting as %s; #it{p} (GeV/#it{c}); Purity", detectorName.c_str(), hypName), |
| 130 | + kTProfile, {axisMomentum}); |
| 131 | + } |
| 132 | + }; |
| 133 | + |
| 134 | + createDetectorHistograms(trackerHists, "Tracker"); |
| 135 | + createDetectorHistograms(tofInnerHists, "TOF_Inner"); |
| 136 | + createDetectorHistograms(tofOuterHists, "TOF_Outer"); |
| 137 | + createDetectorHistograms(richHists, "RICH"); |
| 138 | + createDetectorHistograms(combinedHists, "Combined"); |
| 139 | + createDetectorHistograms(combinedNoTrackerHists, "Combined_NoTracker"); |
| 140 | + |
| 141 | + const AxisSpec axisDetectorCount{5, -0.5, 4.5, "Number of detectors"}; |
| 142 | + hDetectorParticipation2D = histos.add<TH2>("Combined/hDetectorParticipation2D", |
| 143 | + "Detector participation vs momentum; #it{p} (GeV/#it{c}); Number of detectors", |
| 144 | + kTH2F, {axisMomentum, axisDetectorCount}); |
| 145 | + } |
| 146 | + |
| 147 | + void process(soa::Join<aod::Tracks, aod::TracksCov, aod::McTrackLabels, aod::UpgradeTrkPids, aod::UpgradeTrkPidSignals, aod::UpgradeTofs, aod::UpgradeRichs> const& tracks, |
| 148 | + aod::McParticles const& /*mcParticles*/) |
| 149 | + { |
| 150 | + int totalTracks = 0; |
| 151 | + int analyzedTracks = 0; |
| 152 | + |
| 153 | + auto isValidNSigma = [](float nSigma) -> bool { |
| 154 | + return (nSigma < kInvalidNSigmaValue && nSigma > -kInvalidNSigmaValue); |
| 155 | + }; |
| 156 | + |
| 157 | + auto computeCombinedNSigma = [&](const std::vector<std::array<float, PidHypothesis::kCount>>& detectorNSigmas, float p) -> std::array<float, PidHypothesis::kCount> { |
| 158 | + std::array<float, PidHypothesis::kCount> combinedNSigma; |
| 159 | + int totalValidDetectors = 0; |
| 160 | + for (const auto& detNSigma : detectorNSigmas) { |
| 161 | + bool detectorHasValidMeasurement = false; |
| 162 | + for (int hypIdx = 0; hypIdx < PidHypothesis::kCount; ++hypIdx) { |
| 163 | + if (isValidNSigma(detNSigma[hypIdx])) { |
| 164 | + detectorHasValidMeasurement = true; |
| 165 | + break; |
| 166 | + } |
| 167 | + } |
| 168 | + if (detectorHasValidMeasurement) { |
| 169 | + totalValidDetectors++; |
| 170 | + } |
| 171 | + } |
| 172 | + |
| 173 | + for (int hypIdx = 0; hypIdx < PidHypothesis::kCount; ++hypIdx) { |
| 174 | + float sumSquares = 0.0f; |
| 175 | + int validDetectors = 0; |
| 176 | + |
| 177 | + for (const auto& detNSigma : detectorNSigmas) { |
| 178 | + if (isValidNSigma(detNSigma[hypIdx])) { |
| 179 | + sumSquares += detNSigma[hypIdx] * detNSigma[hypIdx]; |
| 180 | + validDetectors++; |
| 181 | + } |
| 182 | + } |
| 183 | + |
| 184 | + if (validDetectors > 0) { |
| 185 | + combinedNSigma[hypIdx] = std::sqrt(sumSquares); |
| 186 | + } else { |
| 187 | + combinedNSigma[hypIdx] = kInvalidNSigmaValue; |
| 188 | + } |
| 189 | + } |
| 190 | + if (totalValidDetectors > 0) { |
| 191 | + hDetectorParticipation2D->Fill(p, totalValidDetectors); |
| 192 | + } |
| 193 | + |
| 194 | + return combinedNSigma; |
| 195 | + }; |
| 196 | + |
| 197 | + auto analyzeDetector = [&](DetectorHistograms& detHists, const std::array<float, PidHypothesis::kCount>& nSigmaValues, |
| 198 | + int trueParticleIndex, float p) { |
| 199 | + detHists.hTotalTrue[trueParticleIndex]->Fill(p); |
| 200 | + bool hasValidNSigma = false; |
| 201 | + for (int i = 0; i < PidHypothesis::kCount; ++i) { |
| 202 | + if (isValidNSigma(nSigmaValues[i])) { |
| 203 | + hasValidNSigma = true; |
| 204 | + break; |
| 205 | + } |
| 206 | + } |
| 207 | + if (!hasValidNSigma) { |
| 208 | + return; |
| 209 | + } |
| 210 | + |
| 211 | + bool correctlyIdentified = false; |
| 212 | + int selectedHypothesis = -1; |
| 213 | + |
| 214 | + if (useClosestHypothesisRule.value) { |
| 215 | + float minAbsNSigma = kInvalidNSigmaValue; |
| 216 | + int bestHypothesis = -1; |
| 217 | + int validHypothesesCount = 0; |
| 218 | + |
| 219 | + for (int hypIdx = 0; hypIdx < PidHypothesis::kCount; ++hypIdx) { |
| 220 | + if (isValidNSigma(nSigmaValues[hypIdx])) { |
| 221 | + float absNSigma = std::fabs(nSigmaValues[hypIdx]); |
| 222 | + if (absNSigma < minAbsNSigma) { |
| 223 | + minAbsNSigma = absNSigma; |
| 224 | + bestHypothesis = hypIdx; |
| 225 | + } |
| 226 | + if (absNSigma < maxNSigmaForIdentification.value) { |
| 227 | + validHypothesesCount++; |
| 228 | + } |
| 229 | + } |
| 230 | + } |
| 231 | + |
| 232 | + if (bestHypothesis >= 0 && minAbsNSigma < maxNSigmaForIdentification.value) { |
| 233 | + if (useMinimalIdentification.value && validHypothesesCount > 1) { |
| 234 | + selectedHypothesis = -1; |
| 235 | + } else { |
| 236 | + selectedHypothesis = bestHypothesis; |
| 237 | + } |
| 238 | + } |
| 239 | + correctlyIdentified = (selectedHypothesis == trueParticleIndex); |
| 240 | + } else { |
| 241 | + correctlyIdentified = (std::fabs(nSigmaValues[trueParticleIndex]) < maxNSigmaForIdentification.value); |
| 242 | + for (int hypIdx = 0; hypIdx < PidHypothesis::kCount; ++hypIdx) { |
| 243 | + if (std::fabs(nSigmaValues[hypIdx]) < maxNSigmaForIdentification.value) { |
| 244 | + bool isCorrect = (hypIdx == trueParticleIndex); |
| 245 | + detHists.hPurityAsHypothesis[hypIdx]->Fill(p, isCorrect ? 1.0 : 0.0); |
| 246 | + } |
| 247 | + } |
| 248 | + } |
| 249 | + |
| 250 | + detHists.hEfficiency[trueParticleIndex]->Fill(p, correctlyIdentified ? 1.0 : 0.0); |
| 251 | + |
| 252 | + if (useClosestHypothesisRule.value && selectedHypothesis >= 0) { |
| 253 | + bool isCorrect = (selectedHypothesis == trueParticleIndex); |
| 254 | + detHists.hPurityAsHypothesis[selectedHypothesis]->Fill(p, isCorrect ? 1.0 : 0.0); |
| 255 | + } |
| 256 | + }; |
| 257 | + |
| 258 | + for (const auto& track : tracks) { |
| 259 | + totalTracks++; |
| 260 | + |
| 261 | + if (!track.has_mcParticle()) { |
| 262 | + continue; |
| 263 | + } |
| 264 | + |
| 265 | + const auto& mcParticle = track.mcParticle(); |
| 266 | + const float p = mcParticle.p(); |
| 267 | + const int truePdg = std::abs(mcParticle.pdgCode()); |
| 268 | + |
| 269 | + int trueParticleIndex = -1; |
| 270 | + for (int i = 0; i < PidHypothesis::kCount; ++i) { |
| 271 | + if (kHypothesisPdg[i] == truePdg) { |
| 272 | + trueParticleIndex = i; |
| 273 | + break; |
| 274 | + } |
| 275 | + } |
| 276 | + if (trueParticleIndex == -1) { |
| 277 | + continue; |
| 278 | + } |
| 279 | + |
| 280 | + std::array<float, PidHypothesis::kCount> trackerNSigma; |
| 281 | + trackerNSigma[kElectron] = track.nSigmaTrkEl(); |
| 282 | + trackerNSigma[kMuon] = track.nSigmaTrkMu(); |
| 283 | + trackerNSigma[kPion] = track.nSigmaTrkPi(); |
| 284 | + trackerNSigma[kKaon] = track.nSigmaTrkKa(); |
| 285 | + trackerNSigma[kProton] = track.nSigmaTrkPr(); |
| 286 | + trackerNSigma[kDeuteron] = track.nSigmaTrkDe(); |
| 287 | + trackerNSigma[kTriton] = track.nSigmaTrkTr(); |
| 288 | + trackerNSigma[kHelium3] = track.nSigmaTrkHe(); |
| 289 | + trackerNSigma[kAlpha] = track.nSigmaTrkAl(); |
| 290 | + |
| 291 | + analyzedTracks++; |
| 292 | + |
| 293 | + analyzeDetector(trackerHists, trackerNSigma, trueParticleIndex, p); |
| 294 | + |
| 295 | + std::array<float, PidHypothesis::kCount> tofInnerNSigma; |
| 296 | + tofInnerNSigma[kElectron] = track.nSigmaElectronInnerTOF(); |
| 297 | + tofInnerNSigma[kMuon] = track.nSigmaMuonInnerTOF(); |
| 298 | + tofInnerNSigma[kPion] = track.nSigmaPionInnerTOF(); |
| 299 | + tofInnerNSigma[kKaon] = track.nSigmaKaonInnerTOF(); |
| 300 | + tofInnerNSigma[kProton] = track.nSigmaProtonInnerTOF(); |
| 301 | + tofInnerNSigma[kDeuteron] = track.nSigmaDeuteronInnerTOF(); |
| 302 | + tofInnerNSigma[kTriton] = track.nSigmaTritonInnerTOF(); |
| 303 | + tofInnerNSigma[kHelium3] = track.nSigmaHelium3InnerTOF(); |
| 304 | + tofInnerNSigma[kAlpha] = track.nSigmaAlphaInnerTOF(); |
| 305 | + |
| 306 | + analyzeDetector(tofInnerHists, tofInnerNSigma, trueParticleIndex, p); |
| 307 | + |
| 308 | + std::array<float, PidHypothesis::kCount> tofOuterNSigma; |
| 309 | + tofOuterNSigma[kElectron] = track.nSigmaElectronOuterTOF(); |
| 310 | + tofOuterNSigma[kMuon] = track.nSigmaMuonOuterTOF(); |
| 311 | + tofOuterNSigma[kPion] = track.nSigmaPionOuterTOF(); |
| 312 | + tofOuterNSigma[kKaon] = track.nSigmaKaonOuterTOF(); |
| 313 | + tofOuterNSigma[kProton] = track.nSigmaProtonOuterTOF(); |
| 314 | + tofOuterNSigma[kDeuteron] = track.nSigmaDeuteronOuterTOF(); |
| 315 | + tofOuterNSigma[kTriton] = track.nSigmaTritonOuterTOF(); |
| 316 | + tofOuterNSigma[kHelium3] = track.nSigmaHelium3OuterTOF(); |
| 317 | + tofOuterNSigma[kAlpha] = track.nSigmaAlphaOuterTOF(); |
| 318 | + |
| 319 | + analyzeDetector(tofOuterHists, tofOuterNSigma, trueParticleIndex, p); |
| 320 | + |
| 321 | + std::array<float, PidHypothesis::kCount> richNSigma; |
| 322 | + richNSigma[kElectron] = track.nSigmaElectronRich(); |
| 323 | + richNSigma[kMuon] = track.nSigmaMuonRich(); |
| 324 | + richNSigma[kPion] = track.nSigmaPionRich(); |
| 325 | + richNSigma[kKaon] = track.nSigmaKaonRich(); |
| 326 | + richNSigma[kProton] = track.nSigmaProtonRich(); |
| 327 | + richNSigma[kDeuteron] = track.nSigmaDeuteronRich(); |
| 328 | + richNSigma[kTriton] = track.nSigmaTritonRich(); |
| 329 | + richNSigma[kHelium3] = track.nSigmaHelium3Rich(); |
| 330 | + richNSigma[kAlpha] = track.nSigmaAlphaRich(); |
| 331 | + |
| 332 | + analyzeDetector(richHists, richNSigma, trueParticleIndex, p); |
| 333 | + |
| 334 | + std::vector<std::array<float, PidHypothesis::kCount>> allDetectorNSigmas; |
| 335 | + |
| 336 | + if (includeTrackerInCombined.value) { |
| 337 | + allDetectorNSigmas.push_back(trackerNSigma); |
| 338 | + } |
| 339 | + if (includeTofInnerInCombined.value) { |
| 340 | + allDetectorNSigmas.push_back(tofInnerNSigma); |
| 341 | + } |
| 342 | + if (includeTofOuterInCombined.value) { |
| 343 | + allDetectorNSigmas.push_back(tofOuterNSigma); |
| 344 | + } |
| 345 | + if (includeRichInCombined.value) { |
| 346 | + allDetectorNSigmas.push_back(richNSigma); |
| 347 | + } |
| 348 | + if (!allDetectorNSigmas.empty()) { |
| 349 | + std::array<float, PidHypothesis::kCount> combinedNSigma = computeCombinedNSigma(allDetectorNSigmas, p); |
| 350 | + analyzeDetector(combinedHists, combinedNSigma, trueParticleIndex, p); |
| 351 | + } |
| 352 | + |
| 353 | + std::vector<std::array<float, PidHypothesis::kCount>> noTrackerDetectorNSigmas; |
| 354 | + |
| 355 | + if (includeTofInnerInCombined.value) { |
| 356 | + noTrackerDetectorNSigmas.push_back(tofInnerNSigma); |
| 357 | + } |
| 358 | + if (includeTofOuterInCombined.value) { |
| 359 | + noTrackerDetectorNSigmas.push_back(tofOuterNSigma); |
| 360 | + } |
| 361 | + if (includeRichInCombined.value) { |
| 362 | + noTrackerDetectorNSigmas.push_back(richNSigma); |
| 363 | + } |
| 364 | + if (!noTrackerDetectorNSigmas.empty()) { |
| 365 | + std::array<float, PidHypothesis::kCount> combinedNoTrackerNSigma = computeCombinedNSigma(noTrackerDetectorNSigmas, p); |
| 366 | + analyzeDetector(combinedNoTrackerHists, combinedNoTrackerNSigma, trueParticleIndex, p); |
| 367 | + } |
| 368 | + } |
| 369 | + } |
| 370 | +}; |
| 371 | + |
| 372 | +WorkflowSpec defineDataProcessing(ConfigContext const& cfgc) |
| 373 | +{ |
| 374 | + return WorkflowSpec{adaptAnalysisTask<Alice3PidEvaluation>(cfgc)}; |
| 375 | +} |
0 commit comments