Vector_double
stf::linCorr(const Vector_double& data, const Vector_double& templ, stfio::ProgressInfo& progDlg)
{
bool skipped = false;
// the template has to be smaller than the data waveform:
if (data.size()<templ.size()) {
throw std::runtime_error("Template larger than data in stf::crossCorr");
}
if (data.size()==0 || templ.size()==0) {
throw std::runtime_error("Array of size 0 in stf::crossCorr");
}
Vector_double Corr(data.size()-templ.size());
// Optimal scaling & offset:
// avoid redundant computations:
double sum_templ_data=0.0, sum_templ=0.0, sum_templ_sqr=0.0, sum_data=0.0, sum_data_sqr=0.0;
for (int n_templ=0; n_templ<(int)templ.size();++n_templ) {
sum_templ_data+=templ[n_templ]*data[0+n_templ];
sum_data+=data[0+n_templ];
sum_data_sqr+=data[0+n_templ]*data[0+n_templ];
sum_templ+=templ[n_templ];
sum_templ_sqr+=templ[n_templ]*templ[n_templ];
}
double y_old=0.0;
double y2_old=0.0;
int progCounter=0;
double progFraction=(data.size()-templ.size())/100;
for (unsigned n_data=0; n_data<data.size()-templ.size(); ++n_data) {
if (n_data/progFraction>progCounter) {
progDlg.Update( (int)((double)n_data/(double)(data.size()-templ.size())*100.0),
"Calculating correlation coefficient", &skipped );
if (skipped) {
Corr.resize(0);
return Corr;
}
progCounter++;
}
if (n_data!=0) {
sum_templ_data=0.0;
// The product has to be computed in full length:
for (int n_templ=0; n_templ<(int)templ.size();++n_templ) {
sum_templ_data+=templ[n_templ]*data[n_data+n_templ];
}
// The new value that will be added is:
double y_new=data[n_data+templ.size()-1];
double y2_new=data[n_data+templ.size()-1]*data[n_data+templ.size()-1];
sum_data+=y_new-y_old;
sum_data_sqr+=y2_new-y2_old;
}
// The first value that was added (and will have to be subtracted during
// the next loop):
y_old=data[n_data+0];
y2_old=data[n_data+0]*data[n_data+0];
double scale=(sum_templ_data-sum_templ*sum_data/templ.size())/
(sum_templ_sqr-sum_templ*sum_templ/templ.size());
double offset=(sum_data-scale*sum_templ)/templ.size();
// Now that the optimal template has been found,
// compute the correlation between data and optimal template.
// The correlation coefficient is computed in a way that avoids
// numerical instability; therefore, the sum of squares
// computed above can't be re-used.
// Get the means:
double mean_data=sum_data/templ.size();
double sum_optTempl=sum_templ*scale+offset*templ.size();
double mean_optTempl=sum_optTempl/templ.size();
// Get SDs:
double sd_data=0.0;
double sd_templ=0.0;
for (int i=0;i<(int)templ.size();++i) {
sd_data+=SQR(data[i+n_data]-mean_data);
sd_templ+=SQR(templ[i]*scale+offset-mean_optTempl);
}
sd_data=sqrt(sd_data/templ.size());
sd_templ=sqrt(sd_templ/templ.size());
// Get correlation:
double r=0.0;
for (int i=0;i<(int)templ.size();++i) {
r+=(data[i+n_data]-mean_data)*(templ[i]*scale+offset-mean_optTempl);
}
r/=((templ.size()-1)*sd_data*sd_templ);
Corr[n_data]=r;
}
return Corr;
}
int main (int ac, char* av[]) {
// ***************************************************************************
// ***************************************************************************
// initialization ************************************************************
// ***************************************************************************
// ***************************************************************************
// initialization of eigen OMP paralization
Eigen::initParallel();
// set numer of threads used by eigen
// first line sets number of threads directly
// second line lets OMP decide on the number of threads,
// e. g. via OMP_NUM_THREADS
//Eigen::setNbThreads(4);
Eigen::setNbThreads(0);
//check the number of threads used
const int nthreads = Eigen::nbThreads();
std::cout << "contraction code for stochastic dilution" << std::endl;
std::cout << "using " << nthreads << " threads for eigen\n" << std::endl;
// reading in global parameters from input file
GlobalData* global_data = GlobalData::Instance();
global_data->read_parameters(ac, av);
// reading in of data
ReadWrite* rewr = new ReadWrite;
// everything for operator handling
BasicOperator* basic = new BasicOperator();
// global variables from input file needed in main function
const int Lt = global_data->get_Lt();
const int end_config = global_data->get_end_config();
const int delta_config = global_data->get_delta_config();
const int start_config = global_data->get_start_config();
const int number_of_eigen_vec = global_data->get_number_of_eigen_vec();
const int number_of_max_mom = global_data->get_number_of_max_mom();
const int max_mom_squared = number_of_max_mom * number_of_max_mom;
const int number_of_momenta = global_data->get_number_of_momenta();
const std::vector<int> mom_squared = global_data->get_momentum_squared();
const std::vector<quark> quarks = global_data->get_quarks();
const int number_of_rnd_vec = quarks[0].number_of_rnd_vec;
const int dirac_min = global_data->get_dirac_min();
const int dirac_max = global_data->get_dirac_max();
const int number_of_dirac = dirac_max - dirac_min + 1;
const int displ_min = global_data->get_displ_min();
const int displ_max = global_data->get_displ_max();
const int number_of_displ = displ_max - displ_min + 1;
const int p_min = 0; //number_of_momenta/2;
const int p_max = number_of_momenta;
// TODO decide on path
std::string outpath = global_data->get_output_path() + "/";
// other variables
clock_t time;
const std::complex<double> I(0.0, 1.0);
char outfile[400];
FILE *fp = NULL;
// ***************************************************************************
// ***************************************************************************
// memory allocation *********************************************************
// ***************************************************************************
// ***************************************************************************
// abbreviations for clearer memory allocation. Wont be used in loops and
// when building the contractions
// CJ: but it is a little bit ugly...
const size_t nmom = number_of_momenta;
const size_t nrnd = number_of_rnd_vec;
const size_t ndir = number_of_dirac;
const size_t ndis = number_of_displ;
// memory for the correlation function
array_cd_d7 C2_mes(boost::extents[nmom][nmom][ndir][ndir][ndis][ndis][Lt]);
//TODO: dont need the memory for p_u^2 > p_d^2
array_cd_d10 Corr(boost::extents[nmom][nmom][ndir][ndir][ndis][ndis][Lt][Lt][nrnd][nrnd]);
int norm = 0;
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd3 = rnd1 + 1; rnd3 < number_of_rnd_vec; ++rnd3){
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; ++rnd2){
if((rnd2 != rnd1) && (rnd2 != rnd3)){
for(int rnd4 = rnd2 + 1; rnd4 < number_of_rnd_vec; ++rnd4){
if((rnd4 != rnd1) && (rnd4 != rnd3)){
norm++;
//std::cout << "\n\nnorm: " << norm << rnd1 << rnd3 << rnd2 << rnd4 << std::endl;
}
}
}
}
}
}
std::cout << "\n\tNumber of contraction combinations: " << norm << std::endl;
// const double norm1 = Lt * norm;
// Memory for propagation matrices (is that a word?) from t_source to t_sink
// (op_1) and vice versa (op_2)
// additional t_source to t_sink (op_3) and t_sink to t_source (op_4) for
// 4-point functions
// 1, 3 -> u-quarks; 2, 4 -> d-quarks; 5, 6 -> u quarks for neutral particle
array_Xcd_d2_eigen op_1(boost::extents[nrnd][nrnd]);
array_Xcd_d2_eigen op_3(boost::extents[nrnd][nrnd]);
vec_Xcd_eigen op_2(number_of_rnd_vec);
vec_Xcd_eigen op_4(number_of_rnd_vec);
vec_Xcd_eigen op_5(number_of_rnd_vec);
vec_Xcd_eigen op_6(number_of_rnd_vec);
for(int rnd_i = 0; rnd_i < number_of_rnd_vec; ++rnd_i){
for(int rnd_j = 0; rnd_j < number_of_rnd_vec; ++rnd_j){
op_1[rnd_i][rnd_j] = Eigen::MatrixXcd(4 * number_of_eigen_vec,
4 * quarks[0].number_of_dilution_E);
}
op_2[rnd_i] = Eigen::MatrixXcd(4 * quarks[0].number_of_dilution_E,
4 * number_of_eigen_vec);
}
// ***************************************************************************
// ***************************************************************************
// Loop over all configurations **********************************************
// ***************************************************************************
// ***************************************************************************
for(int config_i = start_config; config_i <= end_config; config_i +=
delta_config){
std::cout << "\nprocessing configuration: " << config_i << "\n\n";
rewr->read_perambulators_from_file(config_i);
rewr->read_rnd_vectors_from_file(config_i);
// rewr->read_eigenvectors_from_file(config_i);
rewr->read_lime_gauge_field_doubleprec_timeslices(config_i);
rewr->build_source_matrix(config_i);
// *************************************************************************
// TWO PT CONTRACTION 1 ****************************************************
// *************************************************************************
// setting the correlation function to zero
std::cout << "\n\tcomputing the connected contribution of pi_+/-:\n";
time = clock();
// setting the correlation function to zero
for(int p1 = 0; p1 < number_of_momenta; ++p1)
for(int p2 = 0; p2 < number_of_momenta; ++p2)
for(int dirac1 = 0; dirac1 < number_of_dirac; ++dirac1)
for(int dirac2 = 0; dirac2 < number_of_dirac; ++dirac2)
for(int displ1 = 0; displ1 < number_of_displ; ++displ1)
for(int displ2 = 0; displ2 < number_of_displ; ++displ2)
for(int t1 = 0; t1 < Lt; ++t1)
for(int t1 = 0; t1 < Lt; ++t1)
C2_mes[p1][p2][dirac1][dirac2][displ1][displ2][t1] = std::complex<double>(0.0, 0.0);
for(int p1 = 0; p1 < number_of_momenta; ++p1)
for(int p2 = 0; p2 < number_of_momenta; ++p2)
for(int dirac1 = 0; dirac1 < number_of_dirac; ++dirac1)
for(int dirac2 = 0; dirac2 < number_of_dirac; ++dirac2)
for(int displ1 = 0; displ1 < number_of_displ; ++displ1)
for(int displ2 = 0; displ2 < number_of_displ; ++displ2)
for(int t1 = 0; t1 < Lt; ++t1)
for(int t2 = 0; t2 < Lt; ++t2)
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; rnd1++)
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; rnd2++)
Corr[p1][p2][dirac1][dirac2][displ1][displ2][t1][t2][rnd1][rnd2] =
std::complex<double>(0.0, 0.0);
#if 1 // PI^+/-
// initializing of Corr: calculate all two-operator traces of the form tr(u \Gamma \bar{d})
// build all combinations of momenta, dirac_structures and displacements as specified in
// infile
for(int displ_u = 0; displ_u < number_of_displ; displ_u++){
for(int displ_d = 0; displ_d < number_of_displ; displ_d++){
for(int t_source = 0; t_source < Lt; ++t_source){
for(int t_sink = 0; t_sink < Lt; ++t_sink){
for(int p = p_min; p < p_max; ++p) {
// initialize contraction[rnd_i] = perambulator * basicoperator
// = D_u^-1
// choose 'i' for interlace or 'b' for block time dilution scheme
// TODO: get that from input file
// choose 'c' for charged or 'u' for uncharged particles
basic->init_operator_u(0, t_source, t_sink, rewr, 'b', p, displ_min + displ_u);
basic->init_operator_d(0, t_source, t_sink, rewr, 'b', p, displ_min + displ_d);
}
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int p_u = p_min; p_u < p_max; ++p_u) {
// code for pi+-
// "multiply contraction[rnd_i] with gamma structure"
// contraction[rnd_i] are the columns of D_u^-1 which get
// reordered by gamma multiplication. No actual multiplication
// is carried out
basic->get_operator_charged(op_1, 0, t_sink, rewr, dirac_min + dirac_u, p_u);
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
// same as get_operator but with gamma_5 trick. D_u^-1 is
// daggered and multipied with gamma_5 from left and right
// the momentum is changed to reflect the switched sign in
// the momentum exponential for pi_+-
basic->get_operator_g5(op_2, 0, dirac_min + dirac_d, p_d);
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd2 = rnd1 + 1; rnd2 < number_of_rnd_vec; ++rnd2){
// build all 2pt traces leading to C2_mes
// Corr = tr(D_d^-1(t_sink) Gamma
// D_u^-1(t_source) Gamma)
Corr[p_u][p_d][dirac_u][dirac_d][displ_u][displ_d]
[t_source][t_sink][rnd1][rnd2] =
(op_2[rnd2] * op_1[rnd1][rnd2]).trace();
// std::cout << "p" << p_u << p_d << "dirac" << dirac_u << dirac_d << "\nCorr "
// << Corr[p_u][p_d][dirac_u][dirac_d][displ_u][displ_d]
// [t_source][t_sink][rnd1][rnd2] << std::endl;
}
}
}
}
}
}
}
}
}
}
}
// build 2pt-function C2_mes for pi^+ from Corr. Equivalent two just summing
// up traces with same time difference between source and sink (all to all)
// for every dirac structure, momentum, displacement
// build 2pt-function C2_mes for pi^+ from Corr. Equivalent two just summing
// up traces with same time difference between source and sink (all to all)
// for every dirac structure, momentum, displacement
for(int t_source = 0; t_source < Lt; ++t_source){
for(int t_sink = 0; t_sink < Lt; ++t_sink){
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int displ_u = 0; displ_u < number_of_displ; displ_u++){
for(int displ_d = 0; displ_d < number_of_displ; displ_d++){
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd2 = rnd1 + 1; rnd2 < number_of_rnd_vec; ++rnd2){
// building Correlation function
// C2 = tr(D_d^-1 Gamma D_u^-1 Gamma)
// TODO: find signflip of imaginary part
// TODO: is C2_mes[dirac][p] better?
C2_mes[p_u][p_d][dirac_u][dirac_d][displ_u][displ_d]
[abs((t_sink - t_source - Lt) % Lt)] +=
Corr[p_u][number_of_momenta - p_d - 1]
[dirac_u][dirac_d][displ_u][displ_d]
[t_source][t_sink][rnd1][rnd2];
}
}
}
}
}
}
}
}
}
}
}
// normalization of correlation function
double norm3 = Lt * number_of_rnd_vec * (number_of_rnd_vec - 1) * 0.5;
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int displ_u = 0; displ_u < number_of_displ; ++displ_u){
for(int displ_d = 0; displ_d < number_of_displ; ++displ_d){
for(int t = 0; t < Lt; ++t){
C2_mes[p_u][p_d][dirac_u][dirac_d][displ_u][displ_d][t] /= norm3;
}
}
}
}
}
}
}
}
#endif
// output to binary file
// to build a GEVP, the correlators are written into a seperate folder
// for every dirac structure, momentum, displacement (entry of the GEVP
// matrix). In the folders a file is created for every configuration which
// contains all momentum combinations with same momentum squared
// The folders are created when running create_runs.sh. If they dont exist,
// a segmentation fault will occur
// TODO: implement check for existence of folders
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p = 0; p <= max_mom_squared; p++){
for(int displ_u = 0; displ_u < number_of_displ; ++displ_u){
for(int displ_d = 0; displ_d < number_of_displ; ++displ_d){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d_unsuppressed/"
"C2_pi+-_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d, p, p,
displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(rewr.mom_squared[p_u] == p){
fwrite((double*) &(C2_mes[p_u][p_u][dirac_u][dirac_d]
[displ_u][displ_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
fclose(fp);
}
}
}
}
}
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p1 = 0; p1 <= max_mom_squared; p1++){
for(int p2 = p1; p2 <= max_mom_squared; p2++){
for(int displ_u = 0; displ_u < number_of_displ; ++displ_u){
for(int displ_d = 0; displ_d < number_of_displ; ++displ_d){
printf("Writing to file: ");
// sprintf(outfile,
// "%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d/"
// "C2_pi+-_conf%04d.dat",
// outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d,
// p1, p2, displ_min + displ_u, displ_min + displ_d, config_i);
sprintf(outfile, "%s/C2_pi+-_conf%04d.dat", outpath.c_str(), config_i);
printf("%s\n", outfile);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p1){
for(int p_d = p_min; p_d < p_max; ++p_d){
if(mom_squared[p_d] == p2){
fwrite((double*) &(C2_mes[p_u][p_d][dirac_u][dirac_d][displ_u][displ_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
}
}
fclose(fp);
}
}
}
}
}
}
#if 0 // (old?) output routine and output to terminal
sprintf(outfile,
"%s/dirac_%02d_%02d_p_0_%01d_displ_%01d_%01d/C2_pi+-_conf%04d.dat",
outpath.c_str(), dirac_min, dirac_max, 0,
displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac)
fwrite((double*) C2_mes[number_of_momenta/2]
[number_of_momenta/2][dirac], sizeof(double), 2 * Lt, fp);
fclose(fp);
for(int rnd_i = 0; rnd_i < number_of_rnd_vec; ++rnd_i) {
sprintf(outfile,
"%s/dirac_%02d_%02d_p_0_%01d_displ_%01d_%01d/C2_dis_u_rnd%02d_conf%04d.dat",
outpath.c_str(), dirac_min, dirac_max, number_of_max_mom, displ_min,
displ_max, rnd_i, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac)
for(int p = 0; p < number_of_momenta; ++p)
fwrite((double*) C2_dis[p][dirac][rnd_i], sizeof(double), 2 * Lt, fp);
fclose(fp);
}
// output to terminal
// printf("\n");
// for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac){
// printf("\tdirac = %02d\n", dirac);
// for(int p = 0; p <= max_mom_squared; p++){
// printf("\tmomentum_u = %02d\n", p);
// printf("\tmomentum_d = %02d\n", p);
// for(int p_u = p_min; p_u < p_max; ++p_u){
// if((mom_squared[p_u] == p)){
// //printf(
// // "\t t\tRe(C2_con)\tIm(C2_con)\n\t----------------------------------\n");
//// for(int t1 = 0; t1 < Lt; ++t1){
//// printf("\t%02d\t%.5e\t%.5e\n", t1, real(C2_mes[p_u][p_u][dirac][t1]),
//// imag(C2_mes[p_u][p_u][dirac][t1]));
//// }
// printf("\n");
// printf("p_u = %02d\n", p_u);
// }
// }
// }
//
// for(int p = 1; p <= max_mom_squared; p++){
// printf("\tmomentum_u = %02d\n", 0);
// printf("\tmomentum_d = %02d\n", p);
// for(int p_u = p_min; p_u < p_max; ++p_u){
// if((mom_squared[p_u] == p)){
// //printf(
// // "\t t\tRe(C2_con)\tIm(C2_con)\n\t----------------------------------\n");
//// for(int t1 = 0; t1 < Lt; ++t1){
//// printf("\t%02d\t%.5e\t%.5e\n", t1, real(C2_mes[p_u][p_u][dirac][t1]),
//// imag(C2_mes[p_u][p_u][dirac][t1]));
//// }
// printf("\n");
// printf("p_u = %02d\n", p_u);
// }
// }
// }
//
// }
#endif // (old?) output routine and output to terminal
time = clock() - time;
std::cout << "\t\tSUCCESS - " << std::fixed << std::setprecision(1)
<< ((float) time)/CLOCKS_PER_SEC << " seconds" << std::endl;
// *************************************************************************
// FOUR PT CONTRACTION 1 ***************************************************
// *************************************************************************
#if 0 //4-point contraction 1
// setting the correlation function to zero
std::cout << "\n\tcomputing the connected contribution of C4_1:\n";
time = clock();
// displacement not supported for 4pt functions atm
displ_min = 0;
displ_max = 0;
std::cout << "\n\tcomputing the connected contribution of C4_1:\n";
time = clock();
// setting the correlation function to zero
for(int p_u = 0; p_u < number_of_momenta; ++p_u)
for(int p_d = 0; p_d < number_of_momenta; ++p_d)
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u)
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d)
for(int t1 = 0; t1 < Lt; ++t1)
C4_mes[p_u][p_d][dirac_u][dirac_d][t1] =
std::complex<double>(0.0, 0.0);
for(int t_source = 0; t_source < Lt; ++t_source){
for(int t_sink = 0; t_sink < Lt; ++t_sink){
int t_source_1 = (t_source + 1) % Lt;
int t_sink_1 = (t_sink + 1) % Lt;
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
// complete diagramm
// every quark line must have its own random vec
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd3 = rnd1 + 1; rnd3 < number_of_rnd_vec; ++rnd3){
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; ++rnd2){
if((rnd2 != rnd1) && (rnd2 != rnd3)){
for(int rnd4 = rnd2 + 1; rnd4 < number_of_rnd_vec; ++rnd4){
if((rnd4 != rnd1) && (rnd4 != rnd3)){
C4_mes[p_u][p_d][dirac_u][dirac_d]
[abs((t_sink - t_source - Lt) % Lt)] +=
(Corr[p_u]
[number_of_momenta - p_d - 1]
[dirac_u][dirac_d][0][0]
[t_source_1][t_sink_1][rnd1][rnd3]) *
(Corr[number_of_momenta - p_u - 1]
[p_d][dirac_u][dirac_d][0][0]
[t_source][t_sink][rnd2][rnd4]);
}
}
}
}
}
}
}
}
}
}
}
}
}
// Normalization of 4pt-function. Accounts for all rnd-number combinations
for(int t = 0; t < Lt; ++t){
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
C4_mes[p_u][p_d][dirac_u][dirac_d][t] /= norm1;
}
}
}
}
}
}
// output to binary file
// see output to binary file for C2.
// write into folders with suffix "_unsuppressed". These only include
// correlators of the diagonal matrix elements of the GEVP for which
// the three-momentum remains unchanged for both quarks. Because the
// quarks have to be back-to-back, for the offdiagonal elements this
// cannot occur. The suppression can be interpreted as Zweig-suppressed
// gluon exchange
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p = 0; p <= max_mom_squared; p++){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d_unsuppressed/"
"C4_1_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d, p, p,
displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p){
fwrite((double*) &(C4_mes[p_u][p_u][dirac_u][dirac_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
fclose(fp);
}
}
}
// to build a GEVP, the correlators are written into a seperate folder
// for every dirac structure, momentum, (entry of the GEVP matrix).
// displacement is not supported at the moment
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p1 = 0; p1 <= max_mom_squared; p1++){
for(int p2 = p1; p2 <= max_mom_squared; p2++){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d/"
"C4_1_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d,
p1, p2, displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p1){
for(int p_d = p_min; p_d < p_max; ++p_d){
if(mom_squared[p_d] == p2){
fwrite((double*) &(C4_mes[p_u][p_d][dirac_u][dirac_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
}
}
fclose(fp);
}
}
}
}
// output to terminal
// printf("\n");
// for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac){
// printf("\tdirac = %02d\n", dirac);
// for(int offset = 0; offset <= max_mom_squared; offset++){
// for(int p = 0; p <= max_mom_squared; p++){
// if((p + offset) <= max_mom_squared){
// for(int p_u = p_min; p_u < p_max; ++p_u){
// if((rewr.mom_squared[p_u] == p) && ((p + offset) <= max_mom_squared)){
// for(int p_d = p_min; p_d < p_max; ++p_d){
// if(rewr.mom_squared[p_d] == (p + offset)){
// //printf(
// // "\t t\tRe(C4_1_con)\tIm(C4_1_con)\n\t----------------------------------\n");
//// for(int t1 = 0; t1 < Lt; ++t1){
//// printf("\t%02d\t%.5e\t%.5e\n", t1, real(C4_mes[p_u][p_d][dirac][dirac][t1]),
//// imag(C4_mes[p_u][p_d][dirac][dirac][t1]));
//// }
// printf("\n");
// printf("p_u = %02d\tp_d = %02d\n", p_u, p_d);
// }
// }
// }
// }
// }
// printf("\n");
// }
// }
// printf("\n");
// }
time = clock() - time;
printf("\t\tSUCCESS - %.1f seconds\n", ((float) time)/CLOCKS_PER_SEC);
#endif // 4-point contraction 1
// *************************************************************************
// FOUR PT CONTRACTION 2 ***************************************************
// *************************************************************************
#if 0 // 4-point contraction 2
// setting the correlation function to zero
std::cout << "\n\tcomputing the connected contribution of C4_2:\n";
time = clock();
for(int p_u = 0; p_u < number_of_momenta; ++p_u)
for(int p_d = 0; p_d < number_of_momenta; ++p_d)
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u)
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d)
for(int t1 = 0; t1 < Lt; ++t1)
C4_mes[p_u][p_d][dirac_u][dirac_d][t1] =
std::complex<double>(0.0, 0.0);
for(int t_source = 0; t_source < Lt; ++t_source){
for(int t_sink = 0; t_sink < Lt - 1; ++t_sink){
int t_source_1 = (t_source + 1) % Lt;
int t_sink_1 = (t_sink + 1) % Lt;
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
// complete diagramm
// every quark line must have its own random vec
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd3 = rnd1 + 1; rnd3 < number_of_rnd_vec; ++rnd3){
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; ++rnd2){
if((rnd2 != rnd1) && (rnd2 != rnd3)){
for(int rnd4 = rnd2 + 1; rnd4 < number_of_rnd_vec; ++rnd4){
if((rnd4 != rnd1) && (rnd4 != rnd3)){
C4_mes[p_u][p_d][dirac_u][dirac_d]
[abs((t_sink - t_source - Lt) % Lt)] +=
(Corr[p_u][number_of_momenta - p_d - 1]
[dirac_u][dirac_d][0][0][t_source_1][t_sink]
[rnd1][rnd3]) *
(Corr[number_of_momenta - p_u - 1][p_d]
[dirac_u][dirac_d][0][0][t_source][t_sink_1]
[rnd2][rnd4]);
}
}
}
}
}
}
}
}
}
}
}
}
}
// Normalization of 4pt-function. Accounts for all rnd-number combinations
for(int t = 0; t < Lt; ++t){
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_u = p_min; p_u < p_max; ++p_u) {
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
C4_mes[p_u][p_d][dirac_u][dirac_d][t] /= norm1;
}
}
}
}
}
}
// output to binary file
// see output to binary file for C2.
// write into folders with suffix "_unsuppressed". These only include
// correlators of the diagonal matrix elements of the GEVP for which
// the three-momentum remains unchanged for both quarks. Because the
// quarks have to be back-to-back, for the offdiagonal elements this
// cannot occur. The suppression can be interpreted as Zweig-suppressed
// gluon exchange
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p = 0; p <= max_mom_squared; p++){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d_unsuppressed/"
"C4_2_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d, p, p,
displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p){
fwrite((double*) &(C4_mes[p_u][p_u][dirac_u][dirac_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
fclose(fp);
}
}
}
// to build a GEVP, the correlators are written into a seperate folder
// for every dirac structure, momentum, (entry of the GEVP matrix).
// displacement is not supported at the moment
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p1 = 0; p1 <= max_mom_squared; p1++){
for(int p2 = p1; p2 <= max_mom_squared; p2++){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d/"
"C4_2_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d,
p1, p2, displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = p_min; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p1){
for(int p_d = p_min; p_d < p_max; ++p_d){
if(mom_squared[p_d] == p2){
fwrite((double*) &(C4_mes[p_u][p_d][dirac_u][dirac_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
}
}
fclose(fp);
}
}
}
}
// output to terminal
// printf("\n");
// for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac){
// printf("\tdirac = %02d\n", dirac);
// for(int offset = 0; offset <= max_mom_squared; offset++){
// for(int p = 0; p <= max_mom_squared; p++){
// if((p + offset) <= max_mom_squared){
// for(int p_u = p_min; p_u < p_max; ++p_u){
// if((rewr.mom_squared[p_u] == p) && ((p + offset) <= max_mom_squared)){
// for(int p_d = p_min; p_d < p_max; ++p_d){
// if(rewr.mom_squared[p_d] == (p + offset)){
// //printf(
// // "\t t\tRe(C4_2_con)\tIm(C4_2_con)\n\t----------------------------------\n");
//// for(int t1 = 0; t1 < Lt; ++t1){
//// printf("\t%02d\t%.5e\t%.5e\n", t1, real(C4_mes[p][p][dirac][dirac][t1]),
//// imag(C4_mes[p][p][dirac][dirac][t1]));
//// }
// printf("\n");
// printf("p_u = %02d\tp_d = %02d\n", p_u, p_d);
// }
// }
// }
// }
// }
// printf("\n");
// }
// }
// printf("\n");
// }
time = clock() - time;
printf("\t\tSUCCESS - %.1f seconds\n", ((float) time)/CLOCKS_PER_SEC);
#endif // 4-point contraction 2
// *************************************************************************
// FOUR PT CONTRACTION 3 ***************************************************
// *************************************************************************
// TODO: check dirac indices. maybe dirac(t_source) and dirac(t_sink) have
// to be equal or there may be four different structures rather than u- and
// d-quark always having the same dirac structure
// doesn't matter as long as all used dirac structures are equal
#if 0 // 4-point contraction 3
std::cout << "\n\tcomputing the connected contribution of C4_3:\n";
time = clock();
// setting the correlation function to zero
for(int p_u = 0; p_u < number_of_momenta; ++p_u)
for(int p_d = 0; p_d < number_of_momenta; ++p_d)
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u)
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d)
for(int t1 = 0; t1 < Lt; ++t1)
C4_mes[p_u][p_d][dirac_u][dirac_d][t1] =
std::complex<double>(0.0, 0.0);
for(int t_source = 0; t_source < Lt; ++t_source){
for(int t_sink = 0; t_sink < Lt; ++t_sink){
int t_source_1 = (t_source + 1) % Lt;
int t_sink_1 = (t_sink + 1) % Lt;
// initialize basic->contraction[]
// p_u = number_of_momenta/2 and the break; statement arrange
// for one-to-all calculation in momentum space. (only one source
// momentum is used. the first five are {(0,0,0), (0,0,1),
// (0,1,-1), (1,-1,-1), (0,0,2)}
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int p = 0; p <= max_mom_squared; p++){
for(int p_u = number_of_momenta/2; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p){
basic->init_operator_u(0, t_source, t_sink, rewr, 'b', p_u, 0);
basic->init_operator_u(1, t_source_1, t_sink_1, rewr, 'b',
number_of_momenta - p_u - 1, 0);
break;
}
}
}
}
// initialize basic->contraction_dagger[]
// build all momenta for sinks
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_d = p_min; p_d < p_max; ++p_d){
basic->init_operator_d(0, t_source_1, t_sink, rewr, 'b', p_d, 0);
basic->init_operator_d(1, t_source, t_sink_1, rewr, 'b',
number_of_momenta - p_d - 1, 0);
}
}
// build 4pt-function C4_mes for pi^+pi^+ Equivalent two just summing
// up the four-trace with same time difference between source and sink
// (all to all) for every dirac structure, momentum
// displacement not supported at the moment
// to build the trace with four matrices, build combinations
// X = D_d^-1(t_sink | t_source + 1)
// Gamma D_u^-1(t_source + 1 | t_sink + 1) Gamma
// Y = D_d^-1(t_sink + 1| t_source)
// Gamma D_u^-1(t_source| t_sink) Gamma
// these have dimension
// (4 * quarks[0].number_of_dilution_E) x (4 *
// quarks[0].number_of_dilution_E)
// thus the multiplication in this order is fastest
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int p = 0; p <= max_mom_squared; p++){
for(int p_u = number_of_momenta / 2; p_u < p_max; ++p_u) {
if(mom_squared[p_u] == p){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p_d = p_min; p_d < p_max; ++p_d) {
if(mom_squared[p_u] <= mom_squared[p_d]){
// initialisation of X. rnd loops and if-statements rule
// forbidden randomvector combinations (to improve
// statistical error never use the same randomvector
// for different indices
basic->get_operator_g5(op_2, 0, dirac_min + dirac_d,
number_of_momenta - p_d - 1);
basic->get_operator_charged(op_3, 1, t_sink_1, &rewr,
dirac_min + dirac_u, number_of_momenta - p_u - 1);
// second u quark: t_source_1 -> t_sink_1
for(int rnd3 = 1; rnd3 < number_of_rnd_vec; ++rnd3){
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; ++rnd2){
if(rnd2 != rnd3){
// first d quark: t_sink_1 -> t_source
for(int rnd4 = rnd2 + 1; rnd4 < number_of_rnd_vec; ++rnd4){
if(rnd4 != rnd3){
X[rnd3][rnd2][rnd4] = op_2[rnd3] *
op_3[rnd2][rnd4] ;
}
}
}
}
}
// initialisation of Y. see initialisation of X
basic->get_operator_g5(op_4, 1, dirac_min + dirac_d, p_d);
basic->get_operator_charged(op_1, 0, t_sink, rewr,
dirac_min + dirac_u, p_u);
// first u quark: t_source -> t_sink
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd3 = rnd1 + 1; rnd3 < number_of_rnd_vec; ++rnd3){
// second d quark: t_sink -> t_source_1
for(int rnd4 = 1; rnd4 < number_of_rnd_vec; ++rnd4){
if((rnd4 != rnd1) && (rnd4 != rnd3)){
Y[rnd4][rnd1][rnd3] = op_4[rnd4] *
op_1[rnd1][rnd3];
}
}
}
}
// complete diagramm. combine X and Y to four-trace
// C4_mes = tr(D_u^-1(t_source| t_sink) Gamma
// D_d^-1(t_sink | t_source + 1) Gamma
// D_u^-1(t_source + 1 | t_sink + 1) Gamma
// D_d^-1(t_sink + 1| t_source) Gamma)
// every quark line must have its own random vec
for(int rnd1 = 0; rnd1 < number_of_rnd_vec; ++rnd1){
for(int rnd3 = rnd1 + 1; rnd3 < number_of_rnd_vec; ++rnd3){
for(int rnd2 = 0; rnd2 < number_of_rnd_vec; ++rnd2){
if((rnd2 != rnd1) && (rnd2 != rnd3)){
for(int rnd4 = rnd2 + 1; rnd4 < number_of_rnd_vec; ++rnd4){
if((rnd4 != rnd1) && (rnd4 != rnd3)){
C4_mes[p_u][p_d][dirac_u][dirac_d]
[abs((t_sink - t_source - Lt) % Lt)] +=
((X[rnd3][rnd2][rnd4] *
Y[rnd4][rnd1][rnd3]).trace());
}
}
}
}
}
}
}
}
}
break;
}
}
}
}
}
}
// Normalization of 4pt-function. Accounts for all rnd-number combinations
for(int p1 = 0; p1 < number_of_momenta; ++p1)
for(int p2 = 0; p2 < number_of_momenta; ++p2)
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u)
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d)
for(int t = 0; t < Lt; ++t)
C4_mes[p1][p2][dirac_u][dirac_d][t] /= norm1;
// output to binary file
// see output to binary file for C2.
// for the C4_3 diagram the four propagators are connected in the same
// trace. Thus there are no gluon lines which could be cut to create a
// disconnected diagrams and thus no Zweig suppression.
// To build a GEVP, the correlators are written into a seperate folder
// for every dirac structure, momentum, (entry of the GEVP matrix).
// displacement is not supported at the moment
for(int dirac_u = 0; dirac_u < number_of_dirac; ++dirac_u){
for(int dirac_d = 0; dirac_d < number_of_dirac; ++dirac_d){
for(int p1 = 0; p1 <= max_mom_squared; p1++){
for(int p2 = p1; p2 <= max_mom_squared; p2++){
sprintf(outfile,
"%s/dirac_%02d_%02d_p_%01d_%01d_displ_%01d_%01d/"
"C4_3_conf%04d.dat",
outpath.c_str(), dirac_min + dirac_u, dirac_min + dirac_d,
p1, p2, displ_min, displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int p_u = number_of_momenta / 2; p_u < p_max; ++p_u){
if(mom_squared[p_u] == p1){
for(int p_d = p_min; p_d < p_max; ++p_d){
if(mom_squared[p_d] == p2){
fwrite((double*) &(C4_mes[p_u][p_d][dirac_u][dirac_d][0]),
sizeof(double), 2 * Lt, fp);
}
}
break;
}
}
fclose(fp);
}
}
}
}
#if 0
sprintf(outfile,
"%s/dirac_%02d_%02d_p_0_%01d_displ_%01d_%01d/C4_3_conf%04d.dat",
outpath.c_str(), dirac_min, dirac_max, 0, displ_min,
displ_max, config_i);
if((fp = fopen(outfile, "wb")) == NULL)
std::cout << "fail to open outputfile" << std::endl;
for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac)
fwrite((double*) C4_mes[number_of_momenta/2]
[number_of_momenta/2][dirac][dirac], sizeof(double), 2 * Lt, fp);
fclose(fp);
#endif
// output to terminal
// printf("\n");
// for(int dirac = dirac_min; dirac < dirac_max + 1; ++dirac){
// printf("\tdirac = %02d\n", dirac);
// for(int p = p_min; p < p_max; ++p) {
// printf("\tmomentum = %02d\n", p);
// //printf(
// // "\t t\tRe(C4_3_con)\tIm(C4_3_con)\n\t----------------------------------\n");
// for(int t1 = 0; t1 < Lt; ++t1){
// printf("\t%02d\t%.5e\t%.5e\n", t1, real(C4_mes[p][p][dirac][dirac][t1]),
// imag(C4_mes[p][p][dirac][dirac][t1]));
// }
// printf("\n");
// }
// printf("\n");
// }
time = clock() - time;
printf("\t\tSUCCESS - %.1f seconds\n", ((float) time)/CLOCKS_PER_SEC);
#endif // 4-point contraction 3
// *************************************************************************
// FOUR PT CONTRACTION 4 ***************************************************
// *************************************************************************
// identical to FOUR PT CONTRACTION 3
} // loop over configs ends here
// TODO: freeing all memory!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
delete rewr;
delete basic;
}
int main (int argc, char *argv[])
{
if ((argc < 3) || (argc > 4)) usage (argv[0]);
std::string mapfile = argv[1];
std::string quad_list_prefix = argv[2];
Healpix_Map<double> map;
read_Healpix_map_from_fits (mapfile, map);
bool have_mask = false;
Healpix_Map<double> mask;
if (argc == 4) {
read_Healpix_map_from_fits (argv[3], mask);
have_mask = true;
}
// Figure out how many bins there are by trying to open files.
std::vector<std::string> quad_list_files
= Npoint_Functions::get_range_file_list(quad_list_prefix, 0, 180);
std::vector<double> bin_list(quad_list_files.size());
std::vector<double> Corr(quad_list_files.size());
#pragma omp parallel shared(Corr, bin_list, quad_list_files)
{
Npoint_Functions::Quadrilateral_List_File<int> qlf;
#pragma omp for schedule(dynamic,2)
for (size_t k=0; k < quad_list_files.size(); ++k) {
if (! qlf.initialize (quad_list_files[k])) {
std::cerr << "Error initializing quadrilateral list from "
<< quad_list_files[k] << std::endl;
std::exit(1);
}
if (static_cast<size_t>(map.Nside()) != qlf.Nside()) {
std::cerr << "Map has Nside = " << map.Nside()
<< " but quad list has Nside = " << qlf.Nside()
<< "\nGiving up!\n";
std::exit(1);
}
if (map.Scheme() != qlf.Scheme()) map.swap_scheme();
if (have_mask) {
if (static_cast<size_t>(mask.Nside()) != qlf.Nside()) {
std::cerr << "Mask and quadrilateral lists do not have"
<< " the same Nside: " << mask.Nside()
<< " != " << qlf.Nside() << std::endl;
std::exit(1);
}
if (mask.Scheme() != qlf.Scheme()) mask.swap_scheme();
}
#pragma omp critical
{
std::cerr
#ifdef OMP
<< omp_get_thread_num() << " "
#endif
<< k << std::endl;
}
bin_list[k] = qlf.bin_value();
if (have_mask) {
Corr[k] = calculate_masked_fourpoint_function (map, mask, qlf);
} else {
Corr[k] = calculate_fourpoint_function (map, qlf);
}
}
}
for (size_t k=0; k < bin_list.size(); ++k) {
// Same format as spice
std::cout << bin_list[k]*M_PI/180 << " "
<< cos(bin_list[k]*M_PI/180) << " "
<< Corr[k] << std::endl;
}
return 0;
}
/**
* @brief Apply the whole algorithm of K-SVD
*
* @param img_noisy : pointer to an allocated array containing
* the original noisy image;
* @param img_denoised : pointer to an allocated array which
* will contain the final denoised image;
* @param patches : matrix containing all patches including in
* img_noisy;
* @param dictionary : initial random dictionary, which will be
* updated in each iteration of the algo;
* @param sigma : noise value;
* @param N1 : size of patches (N1 x N1);
* @param N2 : number of atoms in the dictionary;
* @param N_iter : number of iteration;
* @param gamma : value used in the correction matrix in the
* case of color image;
* @param C : coefficient used for the stopping criteria of
* the ORMP;
* @param width : width of both images;
* @param height : height of both images;
* @param chnls : number of channels of both images;
* @param doReconstruction : if true, do the reconstruction of
* the final denoised image from patches
* (only in the case of the acceleration
* trick).
*
* @return none.
**/
void ksvd_process(matD_t &patches,
matD_t &dictionary,
matD_t &gamma,
const unsigned N1, // size of features (i.e. 324)
const unsigned N2, // size of the dictionary (i.e. 1000)
const unsigned N_iter, // i.e. 40
const double C)
{
//! Declarations
const unsigned N1_2 = N1;
const double corr = 0; //(sqrtl(1.0l + gamma) - 1.0l) / ((double) N1_2);
const unsigned chnls = 1;
const double eps = ((double) (N1_2)) * C * C;
const unsigned h_p = patches[0].size();
const unsigned w_p = patches.size();
//! Mat & Vec initializations
matD_t dict_ormp (N2 , vecD_t(h_p, 0.0l));
matD_t patches_ormp(w_p, vecD_t(h_p, 0.0l));
matD_t tmp (h_p, vecD_t(N2, 0.0l));
vecD_t normCol (N2);
matD_t Corr (h_p, vecD_t(h_p, 0.0l));
vecD_t U (h_p);
vecD_t V;
matD_t E (w_p, vecD_t(h_p));
//! Vector for ORMP
matD_t ormp_val (w_p, vecD_t ());
matU_t ormp_ind (w_p, vecU_t ());
matD_t res_ormp (N2, vecD_t (w_p));
matU_t omega_table (N2, vecU_t ());
vecU_t omega_size_table(N2, 0);
matD_t alpha (N2, vecD_t ()); // this is a function parameter
//! To avoid reallocation of memory
for (unsigned k = 0; k < w_p; k++)
{
ormp_val[k].reserve(N2);
ormp_ind[k].reserve(N2);
}
for (matU_t::iterator it = omega_table.begin(); it < omega_table.end(); it++)
it->reserve(w_p);
V.reserve(w_p);
//! Correcting matrix
for (unsigned i = 0; i < h_p; i++)
Corr[i][i] = 1.0l;
for (unsigned c = 0; c < 1; c++)
{
matD_t::iterator it_Corr = Corr.begin() + N1_2 * c;
for (unsigned i = 0; i < N1_2; i++, it_Corr++)
{
iterD_t it = it_Corr->begin() + N1_2 * c;
for (unsigned j = 0; j < N1_2; j++, it++)
(*it) += corr;
}
}
#pragma omp parallel for
for (int j = 0; j < w_p; j++)
{
for (unsigned c = 0; c < chnls; c++)
{
iterD_t it_ormp = patches_ormp[j].begin() + c * N1_2;
iterD_t it = patches[j].begin() + c * N1_2;
for (unsigned i = 0; i < N1_2; i++, it++, it_ormp++)
{
double val = 0.0l;
iterD_t it_tmp = patches[j].begin() + c * N1_2;
for (unsigned k = 0; k < N1_2; k++, it_tmp++)
val += corr * (*it_tmp);
(*it_ormp) = val + (*it);
}
}
}
//! Big loop
for (unsigned iter = 0; iter < N_iter; iter++)
{
std::cout << "Step " << iter + 1 << ":" << std::endl;
std::cout << " - Sparse coding" << std::endl;
for (unsigned i = 0; i < h_p; i++)
{
iterD_t it_tmp = tmp[i].begin();
for (unsigned j = 0; j < N2; j++, it_tmp++)
{
double val = 0.0l;
iterD_t it_corr_i = Corr[i].begin();
iterD_t it_dict_j = dictionary[j].begin();
for (unsigned k = 0; k < h_p; k++, it_corr_i++, it_dict_j++)
val += (*it_corr_i) * (*it_dict_j);
(*it_tmp) = val * val;
}
}
iterD_t it_normCol = normCol.begin();
for (unsigned j = 0; j < N2; j++, it_normCol++)
{
double val = 0.0l;
for (unsigned i = 0; i < h_p; i++)
val += tmp[i][j];
(*it_normCol) = 1.0l / sqrtl(val);
}
for (unsigned i = 0; i < h_p; i++)
{
iterD_t it_normCol_j = normCol.begin();
for (unsigned j = 0; j < N2; j++, it_normCol_j++)
{
double val = 0.0l;
iterD_t it_corr_i = Corr[i].begin();
iterD_t it_dict_j = dictionary[j].begin();
for (unsigned k = 0; k < h_p; k++, it_corr_i++, it_dict_j++)
val += (*it_corr_i) * (*it_dict_j);
dict_ormp[j][i] = val * (*it_normCol_j);
}
}
//! ORMP process
std::cout << " - ORMP process" << std::endl;
ormp_process(patches_ormp, dict_ormp, ormp_ind, ormp_val, N2, eps);
for (unsigned i = 0; i < w_p; i++)
{
iterU_t it_ind = ormp_ind[i].begin();
iterD_t it_val = ormp_val[i].begin();
const unsigned size = ormp_val[i].size();
for (unsigned j = 0; j < size; j++, it_ind++, it_val++)
(*it_val) *= normCol[*it_ind];
}
//! Residus
for (unsigned i = 0; i < N2; i++)
{
omega_size_table[i] = 0;
omega_table[i].clear();
alpha[i].clear();
for (iterD_t it = res_ormp[i].begin(); it < res_ormp[i].end(); it++)
*it = 0.0l;
}
for (unsigned i = 0; i < w_p; i++)
{
iterU_t it_ind = ormp_ind[i].begin();
iterD_t it_val = ormp_val[i].begin();
for (unsigned j = 0; j < ormp_val[i].size(); j++, it_ind++, it_val++)
{
omega_table[*it_ind].push_back(i);
omega_size_table[*it_ind]++;
alpha[*it_ind].push_back(*it_val);
res_ormp[*it_ind][i] = *it_val;
}
}
//! Dictionary update
std::cout << " - Dictionary update" << std::endl;
for (unsigned l = 0; l < N2; l++)
{
//! Initializations
const unsigned omega_size = omega_size_table[l];
iterD_t it_dict_l = dictionary[l].begin();
iterD_t it_alpha_l = alpha[l].begin();
iterU_t it_omega_l = omega_table[l].begin();
U.assign(U.size(), 0.0l);
if (omega_size > 0)
{
iterD_t it_a = it_alpha_l;
iterU_t it_o = it_omega_l;
for (unsigned j = 0; j < omega_size; j++, it_a++, it_o++)
{
iterD_t it_d = it_dict_l;
iterD_t it_e = E[j].begin();
iterD_t it_p = patches[*it_o].begin();
for (unsigned i = 0; i < h_p; i++, it_d++, it_e++, it_p++)
(*it_e) = (*it_p) + (*it_d) * (*it_a);
}
matD_t::iterator it_res = res_ormp.begin();
for (unsigned k = 0; k < N2; k++, it_res++)
{
iterU_t it_o = it_omega_l;
iterD_t it_dict_k = dictionary[k].begin();
for (unsigned j = 0; j < omega_size; j++, it_o++)
{
const double val = (*it_res)[*it_o];
if (fabs(val) > 0.0l)
{
iterD_t it_d = it_dict_k;
iterD_t it_e = E[j].begin();
for (unsigned i = 0; i < h_p; i++, it_d++, it_e++)
(*it_e) -= (*it_d) * val;
}
}
}
//! SVD truncated
V.resize(omega_size);
double S = svd_trunc(E, U, V);
dictionary[l] = U;
it_a = it_alpha_l;
iterD_t it_v = V.begin();
it_o = it_omega_l;
for (unsigned j = 0; j < omega_size; j++, it_a++, it_v++, it_o++)
res_ormp[l][*it_o] = (*it_a) = (*it_v) * S;
}
}
std::cout << " - done." << std::endl;
}
// USE omega_table, omega_size_table, and alpha information
// above to build the gamma matrix
// the size of the gamma matrix should be (sizeofdict)x(numofpatches)
for(unsigned i = 0; i < N2; i++)
{
for(unsigned j = 0; j < omega_size_table[i]; j++)
{
unsigned pI = omega_table[i].at(j);
float alphaV = alpha[i].at(j);
gamma[pI].at(i) = alphaV;
}
}
}
int main (int argc, char *argv[])
{
if ((argc < 5) || (argc > 6)) usage (argv[0]);
std::string quad_list_prefix = argv[1];
std::string alm_dir = argv[2];
size_t Nstart, Nend;
if (! Npoint_Functions::from_string (argv[3], Nstart)) {
std::cerr << "Could not parse Nstart\n";
usage (argv[0]);
}
if (! Npoint_Functions::from_string (argv[4], Nend)) {
std::cerr << "Could not parse Nend\n";
usage (argv[0]);
}
bool have_mask = false;
Healpix_Map<double> mask;
if (argc == 6) {
read_Healpix_map_from_fits (argv[5], mask);
have_mask = true;
}
// Figure out how many bins there are by trying to open files.
std::vector<std::string> quad_list_files
= Npoint_Functions::get_range_file_list(quad_list_prefix, 0, 400);
if (quad_list_files.size() == 0) {
std::cerr << "No quad list files found!\n";
usage (argv[0]);
}
int Lmax;
std::vector<Healpix_Map<double> > maps (Nend-Nstart);
// Make maps
{
Npoint_Functions::Quadrilateral_List_File<int> qlf;
qlf.initialize (quad_list_files[0]);
if (have_mask) {
if (static_cast<size_t>(mask.Nside()) != qlf.Nside()) {
std::cerr << "Mask and quadrilateral lists do not have"
<< " the same Nside: " << mask.Nside()
<< " != " << qlf.Nside() << std::endl;
std::exit(1);
}
if (mask.Scheme() != qlf.Scheme()) mask.swap_scheme();
}
Lmax = std::min(200UL, 4*qlf.Nside()+1);
//#pragma omp parallel shared(qlf, maps)
{
Alm<xcomplex<double> > alm (Lmax, Lmax);
//#pragma omp for schedule(static)
for (size_t k=0; k < maps.size(); ++k) {
read_Alm_from_fits (dirtree::filename(alm_dir, "alm_T_", ".fits",
k+Nstart),
alm, Lmax, Lmax);
maps[k].SetNside (qlf.Nside(), RING);
alm2map (alm, maps[k]);
if (maps[k].Scheme() != qlf.Scheme()) maps[k].swap_scheme();
}
}
}
std::vector<double> bin_list(quad_list_files.size());
/* We will generate this by bin for each map so make the bin number the
* first index. */
std::vector<std::vector<double> > Corr(quad_list_files.size());
#pragma omp parallel shared(Corr, bin_list, quad_list_files, maps, mask)
{
Npoint_Functions::Quadrilateral_List_File<int> qlf;
#pragma omp for schedule(dynamic,2)
for (size_t k=0; k < quad_list_files.size(); ++k) {
if (! qlf.initialize (quad_list_files[k])) {
std::cerr << "Error initializing quadrilateral list from "
<< quad_list_files[k] << std::endl;
std::exit(1);
}
bin_list[k] = qlf.bin_value();
if (have_mask) {
Npoint_Functions::calculate_masked_fourpoint_function_list
(maps, mask, qlf, Corr[k]);
} else {
Npoint_Functions::calculate_fourpoint_function_list
(maps, qlf, Corr[k]);
}
}
}
std::cout << "# LCDM four point function from " << quad_list_prefix
<< std::endl;
std::cout << "# First line is bin values, rest are the four point function.\n";
for (size_t k=0; k < bin_list.size(); ++k) {
std::cout << bin_list[k] << " ";
}
std::cout << std::endl;
for (size_t j=0; j < maps.size(); ++j) {
for (size_t k=0; k < bin_list.size(); ++k) {
std::cout << Corr[k][j] << " ";
}
std::cout << std::endl;
}
return 0;
}
