51{
53
54 double clusterE = 0;
55 double clusterEta = 0;
56 double cluster_SIGNIFICANCE = 0;
57 double cluster_time = 0;
58 double cluster_SECOND_TIME = 0;
59 double cluster_CENTER_LAMBDA = 0;
60 double cluster_CENTER_MAG = 0;
61 double cluster_ENG_FRAC_EM_INCL = 0;
62 double cluster_FIRST_ENG_DENS = 0;
63 double cluster_LONGITUDINAL = 0;
64 double cluster_LATERAL = 0;
65 double cluster_PTD = 0;
66 double cluster_ISOLATION = 0;
67
68 std::vector<float> transformedFeatures;
69 std::vector<bool> clusterInputsValid;
70 clusterInputsValid.reserve(
clusters.size());
71 bool ok{};
72
74 {
79 cluster_SECOND_TIME /= (Gaudi::Units::nanosecond * Gaudi::Units::nanosecond);
81 cluster_CENTER_LAMBDA /= Gaudi::Units::millimeter;
84 cluster_FIRST_ENG_DENS /= (Gaudi::Units::GeV / Gaudi::Units::millimeter3);
89 cluster_time = cluster->time() / Gaudi::Units::nanosecond;
90
91 if (!ok) {
93 return StatusCode::FAILURE;
94 }
95
96 float e_EM = 0.0;
97 for (
size_t s = CaloSampling::PreSamplerB;
s < CaloSampling::Unknown;
s++)
98 {
99 if (s == CaloSampling::EMB1 || s == CaloSampling::EMB2 || s == CaloSampling::EMB3 || s == CaloSampling::EME1 || s == CaloSampling::EME2 || s == CaloSampling::EME3 || s == CaloSampling::FCAL0)
100 {
102 }
103 }
104 cluster_ENG_FRAC_EM_INCL = e_EM / cluster->rawE();
105
106 std::vector<float> rawValues = {
107 static_cast<float>(clusterE),
108 static_cast<float>(clusterEta),
109 static_cast<float>(cluster_SIGNIFICANCE),
110 static_cast<float>(cluster_time),
111 static_cast<float>(cluster_SECOND_TIME),
112 static_cast<float>(cluster_CENTER_LAMBDA),
113 static_cast<float>(cluster_CENTER_MAG),
114 static_cast<float>(cluster_ENG_FRAC_EM_INCL),
115 static_cast<float>(cluster_FIRST_ENG_DENS),
116 static_cast<float>(cluster_LONGITUDINAL),
117 static_cast<float>(cluster_LATERAL),
118 static_cast<float>(cluster_PTD),
119 static_cast<float>(cluster_ISOLATION),
120 static_cast<float>(nPrimVtx),
121 avgMu
122 };
123
124 bool inputsValid = true;
126 {
128 const float raw = rawValues.at(i);
130 if (!std::isfinite(transformed)) {
131 inputsValid = false;
132
133
134 transformed = 0.0F;
135 }
136 transformedFeatures.push_back(transformed);
137 }
138 clusterInputsValid.push_back(inputsValid);
139 }
140
142 std::vector<int64_t> inputShape = {numClusters, 15};
143
145 inputData["features"] = std::make_pair(
146 inputShape, std::move(transformedFeatures));
147
149
150 outputData["mus"] = std::make_pair(
151 std::vector<int64_t>{numClusters, 3}, std::vector<float>{});
152 outputData["sigmas"] = std::make_pair(
153 std::vector<int64_t>{numClusters, 3}, std::vector<float>{});
154 outputData["alphas"] = std::make_pair(
155 std::vector<int64_t>{numClusters, 3}, std::vector<float>{});
156
158
159 std::vector<float> &onnx_mus = std::get<std::vector<float>>(outputData["mus"].second);
160 std::vector<float> &onnx_sigma2s = std::get<std::vector<float>>(outputData["sigmas"].second);
161 std::vector<float> &onnx_alphas = std::get<std::vector<float>>(outputData["alphas"].second);
162
163 if (msgLvl(MSG::DEBUG)) {
164 int nan_in_mus = 0;
165 int nan_in_sigma2s = 0;
166 int nan_in_alphas = 0;
167
168 for (float val : onnx_mus)
169 if (std::isnan(val))
170 nan_in_mus++;
171
172 for (float val : onnx_sigma2s)
173 if (std::isnan(val))
174 nan_in_sigma2s++;
175
176 for (float val : onnx_alphas)
177 if (std::isnan(val))
178 nan_in_alphas++;
179
180 if (nan_in_mus > 0)
181 ATH_MSG_DEBUG(nan_in_mus <<
" NaN value found in `mus` output layer during ONNX inference");
182
183 if (nan_in_sigma2s)
184 ATH_MSG_DEBUG(nan_in_sigma2s <<
" NaN value found in `sigmas` output layer during ONNX inference");
185
186 if (nan_in_alphas)
187 ATH_MSG_DEBUG(nan_in_alphas <<
" NaN value found in `alphas` output layer during ONNX inference");
188 }
189
190 clusterE_ML_vec.clear();
191 clusterE_ML_Unc_vec.clear();
192 clusterE_ML_vec.reserve(numClusters);
193 clusterE_ML_Unc_vec.reserve(numClusters);
194
195 for (
int i = 0;
i < numClusters; ++
i)
196 {
197 bool calibrateCluster = clusterInputsValid.at(i);
198 for (
size_t j=0;
j<3; ++
j) {
199 if (std::isnan(onnx_mus[i*3+j]) || std::isnan(onnx_sigma2s[i*3+j]) || std::isnan(onnx_alphas[i*3+j])) {
200 calibrateCluster = false;
201 break;
202 }
203 }
206 if (calibrateCluster) {
207 std::vector<float> current_mus = {onnx_mus[
i * 3], onnx_mus[
i * 3 + 1], onnx_mus[
i * 3 + 2]};
208 std::vector<float> current_sigma2s = {onnx_sigma2s[
i * 3], onnx_sigma2s[
i * 3 + 1], onnx_sigma2s[
i * 3 + 2]};
209 std::vector<float> current_alphas = {onnx_alphas[
i * 3], onnx_alphas[
i * 3 + 1], onnx_alphas[
i * 3 + 2]};
210
212 r = std::pow(10, mode);
214 s = std::abs(std::log(10) *
r) * onnx_s;
215
216 if (!std::isfinite(
r) || std::abs(
r) < 1e-6) {
217 ATH_MSG_WARNING(
"ML-correction factor to cluster energy (used as denominator) is " <<
r <<
"; The ML-correction factor is reset to 1. Uncertainty is set to 0.");
220 }
221 }
222
224
225 clusterE_ML_vec.push_back(cluster_energy);
226 clusterE_ML_Unc_vec.push_back(static_cast<double>(s));
227 }
228
229 return StatusCode::SUCCESS;
230}
#define ATH_CHECK
Evaluate an expression and check for errors.
#define ATH_MSG_ERROR(x,...)
#define ATH_MSG_WARNING(x,...)
@ PTD
relative spread of pT of constiuent cells = sqrt(n)*RMS/Mean
@ SECOND_TIME
Second moment of cell time distribution in cluster.
@ LATERAL
Normalized lateral moment.
@ LONGITUDINAL
Normalized longitudinal moment.
@ FIRST_ENG_DENS
First Moment in E/V.
@ CENTER_LAMBDA
Shower depth at Cluster Centroid.
@ SIGNIFICANCE
Cluster significance.
@ CENTER_MAG
Cluster Centroid ( ).
@ ISOLATION
Energy weighted fraction of non-clustered perimeter cells.
CaloSampling::CaloSample CaloSample
Amg::Vector3D transform(Amg::Vector3D &v, Amg::Transform3D &tr)
Transform a point from a Trasformation3D.
std::map< std::string, InferenceData > OutputDataMap
std::map< std::string, InferenceData > InputDataMap
float modes(const std::vector< float > &mus, const std::vector< float > &log_sigma2s, const std::vector< float > &alphas)
float sigma_stoch(const std::vector< float > &mus, const std::vector< float > &log_sigma2s, const std::vector< float > &alphas)
float j(const xAOD::IParticle &, const xAOD::TrackMeasurementValidation &hit, const Eigen::Matrix3d &jab_inv)
CaloCluster_v1 CaloCluster
Define the latest version of the calorimeter cluster class.
1.17.0