EventWeightProvider::EventWeightProvider(string datasetInformationFile) :
datasetInfo_(datasetInformationFile), //
xsection(datasetInfo_.getCrossSections()), //
numberOfProcessedEvents(datasetInfo_.getArrayOfProcessedEvents()), //
estimatedPileUp(Globals::estimatedPileup), //
DATAdistribution(), //
pileUpWeights(), //
numberOfEventsWithTooHighPileUp(0) {
generate_weights();
}
void fit_complete_signal(matrix* ols_fit, matrix* design_matrix, matrix* signal, double min_signal,
double min_diffusivity, tensor** tensor_output) {
int number_of_signals = signal->rows;
int signal_elements = signal->columns;
matrix* processed_signal_gpu = process_signal(signal, min_signal);
//Switching dimensions to simulator transpose and conversion to column major
processed_signal_gpu->rows = signal_elements;
processed_signal_gpu->columns = number_of_signals;
matrix* column_major_weights_gpu = generate_weights(ols_fit, processed_signal_gpu);
matrix* column_major_design_matrix_gpu = process_matrix(design_matrix);
//Switching dimensions back
processed_signal_gpu->rows = number_of_signals;
processed_signal_gpu->columns = signal_elements;
double* tensors_gpu = cuda_fitter(column_major_design_matrix_gpu, column_major_weights_gpu,
processed_signal_gpu);
double* padded_eigendecompositions_gpu = cuda_decompose_tensors(tensors_gpu, column_major_design_matrix_gpu->columns , number_of_signals);
cudaDeviceSynchronize();
extract_eigendecompositions(padded_eigendecompositions_gpu, tensor_output, number_of_signals, min_diffusivity);
}
