Updated pipeline to run on the **pre_summer2026** production (#524)

Author: selvaggi

analyzers/dataframe/src/JetClusteringUtils.cc

@@ -258,7 +258,7 @@ std::vector<float> exclusive_dmerge(fastjet::ClusterSequence &cs,
 }
 
 bool check(unsigned int n, int exclusive, float cut) {
-  if (exclusive > 0 && n <= int(cut))
+  if (exclusive > 0 && n < int(cut))
     return false;
   return true;
 }

analyzers/dataframe/src/JetConstituentsUtils.cc

@@ -817,11 +817,11 @@ namespace FCCAnalyses
 
           if (ct.at(j).tracks_begin < trackdata.size())
           {
-            if (abs(ct.at(j).charge) > 0 and abs(ct.at(j).mass - 0.000510999) < 1.e-05)
+            if (std::abs(ct.at(j).charge) > 0 and std::abs(ct.at(j).mass - 0.000510999) < 1.e-05)
             {
               tmp.push_back(0.000510999);
             }
-            else if (abs(ct.at(j).charge) > 0 and abs(ct.at(j).mass - 0.105658) < 1.e-03)
+            else if (std::abs(ct.at(j).charge) > 0 and std::abs(ct.at(j).mass - 0.105658) < 1.e-03)
             {
               tmp.push_back(0.105658);
             }
@@ -1094,7 +1094,7 @@ namespace FCCAnalyses
         FCCAnalysesJetConstituents ct = jcs.at(i);
         for (int j = 0; j < ct.size(); ++j)
         {
-          if (abs(ct.at(j).charge) > 0 and abs(ct.at(j).mass - 0.000510999) < 1.e-05)
+          if (std::abs(ct.at(j).charge) > 0 and std::abs(ct.at(j).mass - 0.000510999) < 1.e-05)
           {
             is_El.push_back(1.);
           }
@@ -1118,7 +1118,7 @@ namespace FCCAnalyses
         FCCAnalysesJetConstituents ct = jcs.at(i);
         for (int j = 0; j < ct.size(); ++j)
         {
-          if (abs(ct.at(j).charge) > 0 and abs(ct.at(j).mass - 0.105658) < 1.e-03)
+          if (std::abs(ct.at(j).charge) > 0 and std::abs(ct.at(j).mass - 0.105658) < 1.e-03)
           {
             is_Mu.push_back(1.);
           }
@@ -1142,7 +1142,7 @@ namespace FCCAnalyses
         FCCAnalysesJetConstituents ct = jcs.at(i);
         for (int j = 0; j < ct.size(); ++j)
         {
-          if (abs(ct.at(j).charge) > 0 and abs(ct.at(j).mass - 0.13957) < 1.e-03)
+          if (std::abs(ct.at(j).charge) > 0 and std::abs(ct.at(j).mass - 0.13957) < 1.e-03)
           {
             is_ChargedHad.push_back(1.);
           }

analyzers/dataframe/src/MCParticle.cc

@@ -103,6 +103,7 @@ get_EventPrimaryVertexP4::get_EventPrimaryVertexP4() {};
 TLorentzVector get_EventPrimaryVertexP4::operator() ( ROOT::VecOps::RVec<edm4hep::MCParticleData> in )  {
   TLorentzVector result(-1e12,-1e12,-1e12,-1e12);
   Bool_t found_py8 = false;
+  Bool_t found_beam = false;
   //std::cout<<"-------------------------------------------"<<std::endl;
   // first try pythia8 gen status == 21 code;
   for (auto & p: in) {
@@ -117,8 +118,19 @@ TLorentzVector get_EventPrimaryVertexP4::operator() ( ROOT::VecOps::RVec<edm4hep
 
    if (!found_py8) {
      for (auto & p: in) {
-        // std::cout<< p.generatorStatus<<", "<<p.PDG<<", "<<p.momentum.x<<", "<<p.momentum.y<<",     "<< p.vertex.y<<", "<< p.vertex.z<<", "<< p.time * 1.0e3 * 2.99792458e+8<<std::endl;
-        if ( p.generatorStatus == 2 and abs(p.vertex.z) > 1.e-12 ) {   // generator status code for the incoming particles of the hardest subprocess
+        if ( p.generatorStatus == 4 ) {   // HepMC beam particles (Herwig, Sherpa, ...)
+          // vertex.time is in s, convert in mm here.
+          TLorentzVector res( p.vertex.x, p.vertex.y, p.vertex.z, p.time * 1.0e3 * 2.99792458e+8);
+          result = res;
+          found_beam = true;
+          break;
+        }
+      }
+   }
+
+   if (!found_py8 && !found_beam) {
+     for (auto & p: in) {
+        if ( p.parents_begin == p.parents_end ) {   // generators with neither convention (Whizard/STDHEP has only status 1 and 2): the beam particles are the only parentless particles
           // vertex.time is in s, convert in mm here.
           TLorentzVector res( p.vertex.x, p.vertex.y, p.vertex.z, p.time * 1.0e3 * 2.99792458e+8);
           result = res;

examples/FCCee/weaver/README.md

@@ -1,38 +1,15 @@
-import os
-import urllib.request
-
-# ____________________________________________________________
-def get_file_path(url, filename):
-    if os.path.exists(filename):
-        return os.path.abspath(filename)
-    else:
-        urllib.request.urlretrieve(url, os.path.basename(url))
-        return os.path.basename(url)
-
-# ____________________________________________________________
-
 ## input file needed for unit test in CI
 testFile = "https://fccsw.web.cern.ch/fccsw/testsamples/wzp6_ee_nunuH_Hss_ecm240.root"
 
 ## output directory
 outputDir   = "outputs/inference"
 
-## latest particle transformer model, trainied on 9M jets in winter2023 samples
-model_name = "fccee_flavtagging_edm4hep_wc_v1"
-
-## model files needed for unit testing in CI
-url_model_dir = "https://fccsw.web.cern.ch/fccsw/testsamples/jet_flavour_tagging/winter2023/wc_pt_13_01_2022/"
-url_preproc = "{}/{}.json".format(url_model_dir, model_name)
-url_model = "{}/{}.onnx".format(url_model_dir, model_name)
-
-## model files locally stored on /eos
-model_dir = "/eos/experiment/fcc/ee/jet_flavour_tagging/winter2023/wc_pt_13_01_2022/"
-local_preproc = "{}/{}.json".format(model_dir, model_name)
-local_model = "{}/{}.onnx".format(model_dir, model_name)
+## pre_summer2026 7-class (G/U/D/S/C/B/TAU) particle transformer, 70M jets
+model_dir = "/eos/experiment/fcc/ee/jet_flavour_tagging/pre_summer2026/models/IDEA_240_andrea_30_06_2026/TRAINING_70M_7labels_4GPUs_newKey4Hep_LongJob2"
+model_name = "TRAINING_70M_7labels_4GPUs_newKey4Hep_LongJob2"
 
-## get local file, else download from url
-weaver_preproc = get_file_path(url_preproc, local_preproc)
-weaver_model = get_file_path(url_model, local_model)
+weaver_preproc = "{}/{}_preprocess.json".format(model_dir, model_name)
+weaver_model = "{}/{}_best_epoch_state.onnx".format(model_dir, model_name)
 
 from addons.ONNXRuntime.jetFlavourHelper import JetFlavourHelper
 from addons.FastJet.jetClusteringHelper import ExclusiveJetClusteringHelper

examples/FCCee/weaver/analysis_inference.py

@@ -23,8 +23,9 @@
     "Bz": "magFieldBz",
 }
 
-#### list of flavors f = g, q, c, s, ...(will look for input file name ccontaining "[Hff]")
-flavors = ["g", "q", "s", "c", "b", "tau"]
+#### list of flavors (7 classes, matching the G/U/D/S/C/B/TAU tagger outputs);
+#### the jet label is inferred from the input file name ("Hff" or "_ee_ff_")
+flavors = ["g", "u", "d", "s", "c", "b", "tau"]
 
 ## define here the branches to be stored in the output root files in addition to the predefined one
 ## only the name of the var is used here, the metadata is used in stage_plots