Various seeding performance writer improvements (#1104)

stephenswat · web-flow · commit ed4e53dc1ac4 · 2025-07-29T21:44:10.000+02:00
This commit makes the following changes to the seeding performance
writer (and some related code):

1. The class now accepts the same truth matching configuration as the
   finding efficiency writer.
2. Relevant example executables were updated to forward the
   aforementioned configuration to the efficiency writer.
3. ROOT plots outputted now have clearer names.
4. Printouts have been added of the seeding performance.
5. In tandem with point 4, the finding efficiency writer has had its
   printouts more clearly differentiated, and the cut optimiser has been
   updated accordingly.
diff --git a/examples/run/alpaka/seeding_example_alpaka.cpp b/examples/run/alpaka/seeding_example_alpaka.cpp
@@ -79,7 +79,8 @@ int seq_run(const traccc::opts::track_seeding& seeding_opts,
 
     // Performance writer
     traccc::seeding_performance_writer sd_performance_writer(
-        traccc::seeding_performance_writer::config{},
+        traccc::seeding_performance_writer::config{.truth_config =
+                                                       truth_finding_opts},
         logger().clone("SeedingPerformanceWriter"));
     traccc::finding_performance_writer find_performance_writer(
         traccc::finding_performance_writer::config{.truth_config =
diff --git a/examples/run/cpu/seeding_example.cpp b/examples/run/cpu/seeding_example.cpp
@@ -77,7 +77,8 @@ int seq_run(const traccc::opts::track_seeding& seeding_opts,
 
     // Performance writer
     traccc::seeding_performance_writer sd_performance_writer(
-        traccc::seeding_performance_writer::config{},
+        traccc::seeding_performance_writer::config{.truth_config =
+                                                       truth_finding_opts},
         logger().clone("SeedingPerformanceWriter"));
     traccc::finding_performance_writer find_performance_writer(
         traccc::finding_performance_writer::config{.truth_config =
diff --git a/examples/run/cpu/seq_example.cpp b/examples/run/cpu/seq_example.cpp
@@ -158,7 +158,8 @@ int seq_run(const traccc::opts::input_data& input_opts,
 
     // performance writer
     traccc::seeding_performance_writer sd_performance_writer(
-        traccc::seeding_performance_writer::config{},
+        traccc::seeding_performance_writer::config{.truth_config =
+                                                       truth_finding_opts},
         logger().clone("SeedingPerformanceWriter"));
     traccc::finding_performance_writer find_performance_writer(
         traccc::finding_performance_writer::config{.truth_config =
diff --git a/examples/run/cuda/seeding_example_cuda.cpp b/examples/run/cuda/seeding_example_cuda.cpp
@@ -83,7 +83,8 @@ int seq_run(const traccc::opts::track_seeding& seeding_opts,
 
     // Performance writer
     traccc::seeding_performance_writer sd_performance_writer(
-        traccc::seeding_performance_writer::config{},
+        traccc::seeding_performance_writer::config{.truth_config =
+                                                       truth_finding_opts},
         logger().clone("SeedingPerformanceWriter"));
     traccc::finding_performance_writer find_performance_writer(
         traccc::finding_performance_writer::config{.truth_config =
diff --git a/extras/cut_optimiser/main.py b/extras/cut_optimiser/main.py
@@ -275,19 +275,23 @@ def main():
 
                 stdout = result.stdout.decode("utf-8")
 
-                if match := re.search(r"Total efficiency was (\d+(?:\.\d*)?)%", stdout):
+                if match := re.search(
+                    r"Total track efficiency was (\d+(?:\.\d*)?)%", stdout
+                ):
                     result_efficiency = float(match.group(1)) / 100.0
                 else:
                     raise ValueError("Efficiency could not be parsed from stdout!")
 
                 if match := re.search(
-                    r"Total duplicate rate was (\d+(?:\.\d*)?)", stdout
+                    r"Total track duplicate rate was (\d+(?:\.\d*)?)", stdout
                 ):
                     result_duplicate_rate = float(match.group(1))
                 else:
                     raise ValueError("Dupliacate rate could not be parsed from stdout!")
 
-                if match := re.search(r"Total fake rate was (\d+(?:\.\d*)?)", stdout):
+                if match := re.search(
+                    r"Total track fake rate was (\d+(?:\.\d*)?)", stdout
+                ):
                     result_fake_rate = float(match.group(1))
                 else:
                     raise ValueError("Fake rate could not be parsed from stdout!")
diff --git a/performance/include/traccc/efficiency/seeding_performance_writer.hpp b/performance/include/traccc/efficiency/seeding_performance_writer.hpp
@@ -10,6 +10,7 @@
 // Local include(s).
 #include "traccc/utils/helpers.hpp"
 #include "traccc/utils/messaging.hpp"
+#include "traccc/utils/truth_matching_config.hpp"
 
 // Project include(s).
 #include "traccc/edm/measurement.hpp"
@@ -51,11 +52,7 @@ class seeding_performance_writer : public messaging {
             {"Num", plot_helpers::binning("N", 30, -0.5f, 29.5f)}};
 
         /// Cut values
-        scalar pT_cut = 0.5f * traccc::unit<scalar>::GeV;
-        scalar z_min = -500.f * traccc::unit<scalar>::mm;
-        scalar z_max = 500.f * traccc::unit<scalar>::mm;
-        scalar r_max = 200.f * traccc::unit<scalar>::mm;
-        scalar matching_ratio = 0.5f;
+        truth_matching_config truth_config;
     };
 
     /// Construct from configuration and log level.
diff --git a/performance/src/efficiency/eff_plot_tool.hpp b/performance/src/efficiency/eff_plot_tool.hpp
@@ -63,18 +63,28 @@ class eff_plot_tool {
         (void)cache;
 
 #ifdef TRACCC_HAVE_ROOT
+        std::string header;
+
+        if (name == "finding") {
+            header = "Track finding efficiency";
+        } else if (name == "seeding") {
+            header = "Seed finding efficiency";
+        } else {
+            header = "Tracking efficiency";
+        }
+
         // efficiency vs pT
         cache.track_eff_vs_pT = plot_helpers::book_eff(
             TString(name) + "_trackeff_vs_pT",
-            "Tracking efficiency;Truth pT [GeV/c];Efficiency", b_pt);
+            header + ";Truth pT [GeV/c];Efficiency", b_pt);
         // efficiency vs eta
-        cache.track_eff_vs_eta = plot_helpers::book_eff(
-            TString(name) + "_trackeff_vs_eta",
-            "Tracking efficiency;Truth #eta;Efficiency", b_eta);
+        cache.track_eff_vs_eta =
+            plot_helpers::book_eff(TString(name) + "_trackeff_vs_eta",
+                                   header + ";Truth #eta;Efficiency", b_eta);
         // efficiency vs phi
-        cache.track_eff_vs_phi = plot_helpers::book_eff(
-            TString(name) + "_trackeff_vs_phi",
-            "Tracking efficiency;Truth #phi;Efficiency", b_phi);
+        cache.track_eff_vs_phi =
+            plot_helpers::book_eff(TString(name) + "_trackeff_vs_phi",
+                                   header + ";Truth #phi;Efficiency", b_phi);
 #endif  // TRACCC_HAVE_ROOT
     }
 
diff --git a/performance/src/efficiency/finding_performance_writer.cpp b/performance/src/efficiency/finding_performance_writer.cpp
@@ -287,19 +287,19 @@ void finding_performance_writer::write_common(
     TRACCC_INFO("Total number of truth particles was "
                 << total_truth_particles);
     TRACCC_INFO("Total number of found tracks was " << n_tracks);
-    TRACCC_INFO("Total number of matched particles was "
+    TRACCC_INFO("Total number of track-matched particles was "
                 << total_matched_truth_particles);
     TRACCC_INFO("Total number of duplicated tracks was "
                 << total_duplicate_tracks);
     TRACCC_INFO("Total number of fake tracks was " << total_fake_tracks);
-    TRACCC_INFO("Total efficiency was "
+    TRACCC_INFO("Total track efficiency was "
                 << (100. * static_cast<double>(total_matched_truth_particles) /
                     static_cast<double>(total_truth_particles))
                 << "%");
-    TRACCC_INFO("Total duplicate rate was "
+    TRACCC_INFO("Total track duplicate rate was "
                 << (static_cast<double>(total_duplicate_tracks) /
                     static_cast<double>(total_matched_truth_particles)));
-    TRACCC_INFO("Total fake rate was "
+    TRACCC_INFO("Total track fake rate was "
                 << (static_cast<double>(total_fake_tracks) /
                     static_cast<double>(total_truth_particles)));
 }
diff --git a/performance/src/efficiency/seeding_performance_writer.cpp b/performance/src/efficiency/seeding_performance_writer.cpp
@@ -77,6 +77,10 @@ void seeding_performance_writer::write(
     const measurement_collection_types::const_device measurements(
         measurements_view);
 
+    const std::size_t n_seeds = seeds.size();
+
+    std::size_t total_fake_seeds = 0;
+
     // Iterate over the seeds.
     for (edm::seed_collection::const_device::size_type i = 0u; i < seeds.size();
          ++i) {
@@ -143,34 +147,75 @@ void seeding_performance_writer::write(
         assert(seed_measurements.size() > 0u);
         if (static_cast<double>(particle_hit_counts.at(0).hit_counts) /
                 static_cast<double>(seed_measurements.size()) >
-            m_cfg.matching_ratio) {
+            m_cfg.truth_config.matching_ratio) {
             auto pid = particle_hit_counts.at(0).ptc.particle_id;
             match_counter[pid]++;
+        } else {
+            total_fake_seeds++;
         }
     }
 
+    std::size_t total_truth_particles = 0;
+    std::size_t total_matched_truth_particles = 0;
+    std::size_t total_duplicate_seeds = 0;
+
     for (auto const& [pid, ptc] : evt_data.m_particle_map) {
+        std::size_t num_measurements = 0;
+
+        // Find the number of measurements belonging to this track
+        if (auto it = evt_data.m_ptc_to_meas_map.find(ptc);
+            it != evt_data.m_ptc_to_meas_map.end()) {
+            num_measurements = it->second.size();
+        } else {
+            continue;
+        }
 
         // Count only charged particles which satisfiy pT_cut
-        if (ptc.charge == 0 || vector::perp(ptc.momentum) < m_cfg.pT_cut ||
-            ptc.vertex[2] < m_cfg.z_min || ptc.vertex[2] > m_cfg.z_max ||
-            vector::perp(ptc.vertex) > m_cfg.r_max) {
+        if (ptc.charge == 0 ||
+            vector::perp(ptc.momentum) < m_cfg.truth_config.pT_min ||
+            ptc.vertex[2] < m_cfg.truth_config.z_min ||
+            ptc.vertex[2] > m_cfg.truth_config.z_max ||
+            vector::perp(ptc.vertex) > m_cfg.truth_config.r_max ||
+            num_measurements < m_cfg.truth_config.min_track_candidates) {
             continue;
         }
 
+        total_truth_particles++;
+
         bool is_matched = false;
         std::size_t n_matched_seeds_for_particle = 0;
         auto it = match_counter.find(pid);
         if (it != match_counter.end()) {
             is_matched = true;
             n_matched_seeds_for_particle = it->second;
+            total_matched_truth_particles++;
+            total_duplicate_seeds += n_matched_seeds_for_particle - 1;
         }
 
         m_data->m_eff_plot_tool.fill(m_data->m_eff_plot_cache, ptc, is_matched);
         m_data->m_duplication_plot_tool.fill(m_data->m_duplication_plot_cache,
                                              ptc,
                                              n_matched_seeds_for_particle - 1);
     }
+
+    TRACCC_INFO("Total number of truth particles was "
+                << total_truth_particles);
+    TRACCC_INFO("Total number of found seeds was " << n_seeds);
+    TRACCC_INFO("Total number of seed-matched particles was "
+                << total_matched_truth_particles);
+    TRACCC_INFO("Total number of duplicate seeds was "
+                << total_duplicate_seeds);
+    TRACCC_INFO("Total number of fake seeds was " << total_fake_seeds);
+    TRACCC_INFO("Total seed efficiency was "
+                << (100. * static_cast<double>(total_matched_truth_particles) /
+                    static_cast<double>(total_truth_particles))
+                << "%");
+    TRACCC_INFO("Total seed duplicate rate was "
+                << (static_cast<double>(total_duplicate_seeds) /
+                    static_cast<double>(total_matched_truth_particles)));
+    TRACCC_INFO("Total seed fake rate was "
+                << (static_cast<double>(total_fake_seeds) /
+                    static_cast<double>(total_truth_particles)));
 }
 
 void seeding_performance_writer::finalize() {
