bool BehaviorNode::CheckPreconditions(const Agent* pAgent, bool bIsAlive) const
{
Precondition::EPhase phase = bIsAlive ? Precondition::E_UPDATE : Precondition::E_ENTER;
//satisfied if there is no preconditions
if (this->m_preconditions.size() == 0)
{
return true;
}
if (this->m_both_precond == 0)
{
if (phase == Precondition::E_ENTER && this->m_enter_precond == 0)
{
return true;
}
if (phase == Precondition::E_UPDATE && this->m_update_precond == 0)
{
return true;
}
}
bool firstValidPrecond = true;
bool lastCombineValue = false;
for (uint32_t i = 0; i < this->m_preconditions.size(); ++i)
{
Precondition* pPrecond = (Precondition*)this->m_preconditions[i];
if (pPrecond != NULL)
{
Precondition::EPhase ph = pPrecond->GetPhase();
if (ph == Precondition::E_BOTH || ph == phase)
{
bool taskBoolean = pPrecond->Evaluate((Agent*)pAgent);
CombineResults(firstValidPrecond, lastCombineValue, pPrecond, taskBoolean);
}
}
}
return lastCombineValue;
}
// The CubeObject parameter is used for 2 purposes: 1) to retrieve
// cube's alt list, and 2) to compute cube's word cost for the
// tesseract result. The call to CubeObject::WordCost() modifies
// the object's alternate list, so previous state will be lost.
float TesseractCubeCombiner::CombineResults(WERD_RES *tess_res,
CubeObject *cube_obj) {
// If no combiner is loaded or the cube object is undefined,
// tesseract wins with probability 1.0
if (combiner_net_ == NULL || cube_obj == NULL) {
tprintf("Cube WARNING (TesseractCubeCombiner::CombineResults): "
"Cube objects not initialized; defaulting to Tesseract\n");
return 1.0;
}
// Retrieve the alternate list from the CubeObject's current state.
// If the alt list empty, tesseract wins with probability 1.0
WordAltList *cube_alt_list = cube_obj->AlternateList();
if (cube_alt_list == NULL)
cube_alt_list = cube_obj->RecognizeWord();
if (cube_alt_list == NULL || cube_alt_list->AltCount() <= 0) {
tprintf("Cube WARNING (TesseractCubeCombiner::CombineResults): "
"Cube returned no results; defaulting to Tesseract\n");
return 1.0;
}
return CombineResults(tess_res, cube_obj, cube_alt_list);
}
int main(int argc, char *argv[])
{
//vector<rPowers> PowerTable;
int Omega, NumShortTerms, Ordering, IsTriplet, ShPower;
double Alpha, Beta, Gamma, Mu, Lambda1, Lambda2, Lambda3, Kappa;
double **PhiHPhi, **PhiPhi;
vector <double> B(1, 0.0), ARow(1, 0.0), ShortTerms;
double SLS = 0.0, SLC = 0.0;
QuadPoints q;
int sf, l;
double r2Cusp, r3Cusp;
int Node = 0, TotalNodes = 1;
ifstream ParameterFile, FileShortRange;
ofstream OutFile;
//int NodeStart, NodeEnd;
time_t TimeStart, TimeEnd;
#ifdef USE_MPI
char ProcessorName[MPI_MAX_PROCESSOR_NAME];
MPI_Status MpiStatus;
int MpiError, ProcNameLen;
#endif
//@TODO: Will MPI time be different?
TimeStart = time(NULL);
//omp_set_num_threads(1);
// Initialize global constants.
PI = 3.1415926535897932384626433832795029L;
if (argc < 5) {
cerr << "Not enough parameters on the command line." << endl;
cerr << "Usage: Scattering kappa parameterfile.txt shortrangefile.psh results.txt" << endl;
cout << endl;
exit(1);
}
cout << setprecision(18);
cout.setf(ios::showpoint);
// MPI initialization
//@TODO: Check MpiError.
#ifdef USE_MPI
MpiError = MPI_Init(&argc, &argv); // All MPI programs start with MPI_Init; all 'N' processes exist thereafter.
MpiError = MPI_Comm_size(MPI_COMM_WORLD, &TotalNodes); // Find out how big the SPMD world is
MpiError = MPI_Comm_rank(MPI_COMM_WORLD, &Node); // and what this process's rank is.
MpiError = MPI_Get_processor_name(ProcessorName, &ProcNameLen);
string LogName;
if (argc > 5)
LogName = argv[5];
else
LogName = "test.log";
//MpiError = MPI_File_open(MPI_COMM_WORLD, (char*)LogName.c_str(), MPI_MODE_WRONLY | MPI_MODE_CREATE, MPI_INFO_NULL, &MpiLog);
//cout << "Node " << Node << " of " << TotalNodes << " on " << ProcessorName << endl;
// The following code here just shows what the nodes are and how many threads each has.
stringstream ThreadStringStream;
int ThreadStrLen;
int MaxThreads = omp_get_max_threads();
ThreadStringStream << MaxThreads << " threads on Node " << Node << " of " << TotalNodes << " on " << ProcessorName;
if (Node == 0) {
cout << endl << ThreadStringStream.str() << endl;
for (int i = 1; i < TotalNodes; i++) {
MpiError = MPI_Recv(&ThreadStrLen, 1, MPI_INT, i, 0, MPI_COMM_WORLD, &MpiStatus);
char *StringStream = new char[ThreadStrLen+1];
MpiError = MPI_Recv(StringStream, ThreadStrLen, MPI_CHAR, i, 0, MPI_COMM_WORLD, &MpiStatus);
StringStream[ThreadStrLen] = NULL;
cout << StringStream << endl;
}
cout << endl;
}
else {
ThreadStrLen = ThreadStringStream.str().length();
const char *ThreadString = ThreadStringStream.str().c_str();
MpiError = MPI_Send(&ThreadStrLen, 1, MPI_INT, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send((void*)ThreadString, ThreadStrLen, MPI_CHAR, 0, 0, MPI_COMM_WORLD);
}
#else
Node = 0;
#endif
if (Node == 0) {
ParameterFile.open(argv[2]);
OutFile.open(argv[4]);
if (!ParameterFile.is_open()) {
cout << "Could not open parameter file...exiting." << endl;
FinishMPI();
exit(3);
}
if (!OutFile.is_open()) {
cout << "Could not open output file...exiting." << endl;
FinishMPI();
exit(4);
}
OutFile << setprecision(18);
OutFile.setf(ios::showpoint);
ShowDateTime(OutFile);
ReadParamFile(ParameterFile, q, Mu, ShPower, Lambda1, Lambda2, Lambda3, r2Cusp, r3Cusp);
// Read in short-range short-range elements. These have already been calculated by the PsHBound program.
// We are reading in only the binary versions (they were originally text files).
//
FileShortRange.open(argv[3], ios::in | ios::binary);
if (FileShortRange.fail()) {
cerr << "Unable to open file " << argv[3] << " for reading." << endl;
FinishMPI();
exit(2);
}
// Create short-range short-range terms
//
// Read in omega and nonlinear parameters.
if (!ReadShortHeader(FileShortRange, Omega, IsTriplet, Ordering, NumShortTerms, Alpha, Beta, Gamma, l)) {
cerr << argv[3] << " is not a valid Ps-H short-range file...exiting." << endl;
FinishMPI();
exit(3);
}
// Calculate number of terms for a given omega. Do we need to compare to the one in the short-range file above?
NumShortTerms = CalcPowerTableSize(Omega);
Kappa = atof(argv[1]);
if (IsTriplet == 0) sf = 1;
else if (IsTriplet == 1) sf = -1;
else {
cout << "IsTriplet parameter incorrect in " << argv[3] << endl;
exit(6);
}
WriteHeader(OutFile, l, IsTriplet);
cout << "Omega: " << Omega << endl;
cout << "Number of terms: " << NumShortTerms << endl;
cout << "Alpha: " << Alpha << " Beta: " << Beta << " Gamma: " << Gamma << endl;
cout << "Mu: " << Mu << endl;
cout << "Shielding power: " << ShPower << endl;
cout << "Kappa: " << Kappa << endl;
cout << "Lambda: " << Lambda1 << " " << Lambda2 << " " << Lambda3 << endl;
cout << endl << "Number of quadrature points" << endl;
cout << "Long-long: " << q.LongLong_r1 << " " << q.LongLong_r2Leg << " " << q.LongLong_r2Lag << " " << q.LongLong_r3Leg << " " << q.LongLong_r3Lag << " " << q.LongLong_r12 << " " << q.LongLong_r13 << " " << q.LongLong_phi23 << endl;
cout << "Long-long 2/r23 term: " << q.LongLongr23_r1 << " " << q.LongLongr23_r2Leg << " " << q.LongLongr23_r2Lag << " " << q.LongLongr23_r3Leg << " " << q.LongLongr23_r3Lag << " " << q.LongLongr23_phi12 << " " << q.LongLongr23_r13 << " " << q.LongLongr23_r23 << endl;
cout << "Short-long with qi = 0: " << q.ShortLong_r1 << " " << q.ShortLong_r2Leg << " " << q.ShortLong_r2Lag << " " << q.ShortLong_r3Leg << " " << q.ShortLong_r3Lag << " " << q.ShortLong_r12 << " " << q.ShortLong_r13 << " " << q.ShortLong_phi23 << endl;
cout << "Short-long 2/r23 with qi = 0: " << q.ShortLongr23_r1 << " " << q.ShortLongr23_r2Leg << " " << q.ShortLongr23_r2Lag << " " << q.ShortLongr23_r3Leg << " " << q.ShortLongr23_r3Lag << " " << q.ShortLongr23_r12 << " " << q.ShortLongr23_phi13 << " " << q.ShortLongr23_r23 << endl;
cout << "Short-long (full) with qi > 0: " << q.ShortLongQiGt0_r1 << " " << q.ShortLongQiGt0_r2Leg << " " << q.ShortLongQiGt0_r2Lag << " " << q.ShortLongQiGt0_r3Leg << " " << q.ShortLongQiGt0_r3Lag << " " << q.ShortLongQiGt0_r12 << " " << q.ShortLongQiGt0_r13 << " " << q.ShortLongQiGt0_phi23 << endl;
cout << endl;
cout << "Cusp parameters" << endl;
cout << r2Cusp << " " << r3Cusp << endl;
cout << endl;
if (NumShortTerms > 0) {
// Allocate memory for the overlap matrix and point PhiPhiP to rows of PhiPhi so we
// can access it like a 2D array but have it in contiguous memory for LAPACK.
PhiPhi = new double*[NumShortTerms*2];
PhiPhi[0] = new double[NumShortTerms*NumShortTerms*4];
for (int i = 1; i < NumShortTerms*2; i++) {
PhiPhi[i] = PhiPhi[i-1] + NumShortTerms*2;
}
// Read in the <phi|phi> matrix elements.
FileShortRange.read((char*)PhiPhi[0], NumShortTerms*NumShortTerms*4*sizeof(double));
// Allocate memory for the <phi|H|phi> matrix and point PhiHPhiP to rows of PhiHPhi so we
// can access it like a 2D array but have it in contiguous memory for LAPACK.
PhiHPhi = new double*[NumShortTerms*2];
PhiHPhi[0] = new double[NumShortTerms*NumShortTerms*4];
for (int i = 1; i < NumShortTerms*2; i++) {
PhiHPhi[i] = PhiHPhi[i-1] + NumShortTerms*2;
}
// Read in the <phi|H|phi> matrix elements.
FileShortRange.read((char*)PhiHPhi[0], NumShortTerms*NumShortTerms*4*sizeof(double));
// Allocate matrix of short-range - short-range terms.
ShortTerms.resize(NumShortTerms*NumShortTerms*4);
//@TODO: Remove next line.
//memset(ShortTerms, 0, NumShortTerms*NumShortTerms*4*sizeof(double)); // Initialize to all 0.
}
}
//@TODO: Check results of MpiError.
#ifdef USE_MPI
MpiError = MPI_Bcast(&Omega, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&NumShortTerms, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Ordering, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&IsTriplet, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Mu, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&ShPower, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Kappa, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Lambda1, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Lambda2, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Lambda3, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Alpha, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Beta, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&Gamma, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&r2Cusp, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&r3Cusp, 1, MPI_DOUBLE, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&sf, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&l, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&ShPower, 1, MPI_INT, 0, MPI_COMM_WORLD);
MpiError = MPI_Bcast(&q, sizeof(q), MPI_UNSIGNED_CHAR, 0, MPI_COMM_WORLD);
#endif
ARow.resize(NumShortTerms*2+1);
B.resize(NumShortTerms*2+1);
//@TODO: Remove next line.
//memset(ARow, 0, (NumShortTerms+1)*sizeof(double)); // Initialize to all 0.
//memset(B, 0, (NumShortTerms+1)*sizeof(double)); // Initialize to all 0.
vector <rPowers> PowerTableQi0, PowerTableQiGt0;
vector <vector <rPowers> > PowerTableQi0Array(TotalNodes), PowerTableQiGt0Array(TotalNodes);
vector <double> AResultsQi0, BResultsQi0, AResultsQiGt0, BResultsQiGt0;
vector <double> AResultsQi0Final, BResultsQi0Final, AResultsQiGt0Final, BResultsQiGt0Final;
int NumTerms, NumTermsQi0, NumTermsQiGt0;
NumTerms = CalcPowerTableSize(Omega);
if (Node == 0) {
NumTermsQi0 = CalcPowerTableSizeQi0(Omega, Ordering, 0, NumShortTerms);
NumTermsQiGt0 = CalcPowerTableSizeQiGt0(Omega, Ordering, 0, NumShortTerms);
// The *2 comes from the 2 types of symmetry
AResultsQi0.resize(NumTermsQi0*2, 0.0);
AResultsQiGt0.resize(NumTermsQiGt0*2, 0.0);
BResultsQi0.resize(NumTermsQi0*2, 0.0);
BResultsQiGt0.resize(NumTermsQiGt0*2, 0.0);
AResultsQi0Final.resize(NumTermsQi0*2, 0.0);
AResultsQiGt0Final.resize(NumTermsQiGt0*2, 0.0);
BResultsQi0Final.resize(NumTermsQi0*2, 0.0);
BResultsQiGt0Final.resize(NumTermsQiGt0*2, 0.0);
PowerTableQi0.resize(NumTermsQi0*2, rPowers(Alpha, Beta, Gamma));
PowerTableQiGt0.resize(NumTermsQiGt0*2, rPowers(Alpha, Beta, Gamma));
GenOmegaPowerTableQi0(Omega, l, Ordering, PowerTableQi0, 0, NumShortTerms-1);
GenOmegaPowerTableQiGt0(Omega, l, Ordering, PowerTableQiGt0, 0, NumShortTerms-1);
}
#ifdef USE_MPI
double NumTermsQi0Proc, NumTermsQiGt0Proc;
vector <int> NumTermsQi0Array(TotalNodes), NumTermsQiGt0Array(TotalNodes);
MpiError = MPI_Barrier(MPI_COMM_WORLD);
// Tell all processes what terms they should be evaluating.
if (Node == 0) {
NumTermsQi0Proc = (double)NumTermsQi0 / (double)TotalNodes; //@TODO: Need the typecast?
NumTermsQiGt0Proc = (double)NumTermsQiGt0 / (double)TotalNodes; //@TODO: Need the typecast?
int Qi0Pos = (int)NumTermsQi0Proc, QiGt0Pos = (int)NumTermsQiGt0Proc;
for (int i = 1; i < TotalNodes; i++) {
//@TODO: Do we need to save these values in an array?
NumTermsQi0Array[i] = int(NumTermsQi0Proc * (i+1)) - int(NumTermsQi0Proc * i);
NumTermsQiGt0Array[i] = int(NumTermsQiGt0Proc * (i+1)) - int(NumTermsQiGt0Proc * i);
MpiError = MPI_Send(&NumTermsQi0Array[i], 1, MPI_INT, i, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&NumTermsQiGt0Array[i], 1, MPI_INT, i, 0, MPI_COMM_WORLD);
cout << "Node " << i << ": " << NumTermsQi0Array[i] << " " << NumTermsQiGt0Array[i] << endl;
PowerTableQi0Array[i].resize(NumTermsQi0Array[i]);
PowerTableQiGt0Array[i].resize(NumTermsQiGt0Array[i]);
for (int j = 0; j < NumTermsQi0Array[i]; j++, Qi0Pos++) {
PowerTableQi0Array[i][j] = PowerTableQi0[Qi0Pos];
//cout << i << ": " << PowerTableQi0[Qi0Pos].ki + PowerTableQi0[Qi0Pos].li + PowerTableQi0[Qi0Pos].mi + PowerTableQi0[Qi0Pos].ni + PowerTableQi0[Qi0Pos].pi + PowerTableQi0[Qi0Pos].qi << " - " <<
// PowerTableQi0[Qi0Pos].ki << " " << PowerTableQi0[Qi0Pos].li << " " << PowerTableQi0[Qi0Pos].mi << " " << PowerTableQi0[Qi0Pos].ni << " " << PowerTableQi0[Qi0Pos].pi << " " << PowerTableQi0[Qi0Pos].qi << endl;
}
#ifdef VERBOSE
cout << endl;
#endif
for (int j = 0; j < NumTermsQiGt0Array[i]; j++, QiGt0Pos++) {
PowerTableQiGt0Array[i][j] = PowerTableQiGt0[QiGt0Pos];
#ifdef VERBOSE
cout << i << ": " << PowerTableQiGt0[QiGt0Pos].ki + PowerTableQiGt0[QiGt0Pos].li + PowerTableQiGt0[QiGt0Pos].mi + PowerTableQiGt0[QiGt0Pos].ni + PowerTableQiGt0[QiGt0Pos].pi + PowerTableQiGt0[QiGt0Pos].qi << " - " <<
PowerTableQiGt0[QiGt0Pos].ki << " " << PowerTableQiGt0[QiGt0Pos].li << " " << PowerTableQiGt0[QiGt0Pos].mi << " " << PowerTableQiGt0[QiGt0Pos].ni << " " << PowerTableQiGt0[QiGt0Pos].pi << " " << PowerTableQiGt0[QiGt0Pos].qi << endl;
#endif//VERBOSE
}
//@TODO: How safe is this?
MpiError = MPI_Send(&PowerTableQi0Array[i][0], sizeof(rPowers)*NumTermsQi0Array[i], MPI_BYTE, i, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&PowerTableQiGt0Array[i][0], sizeof(rPowers)*NumTermsQiGt0Array[i], MPI_BYTE, i, 0, MPI_COMM_WORLD);
}
NumTermsQi0Array[0] = int(NumTermsQi0Proc);
NumTermsQiGt0Array[0] = int(NumTermsQiGt0Proc);
NumTermsQi0 = NumTermsQi0Array[0];
NumTermsQiGt0 = NumTermsQiGt0Array[0];
//AResultsQi0.resize(NumTermsQi0*2, 0.0);
//AResultsQiGt0.resize(NumTermsQiGt0*2, 0.0);
//BResultsQi0.resize(NumTermsQi0*2, 0.0);
//BResultsQiGt0.resize(NumTermsQiGt0*2, 0.0);
//@TODO: Temporary?
int NumTermsQi0Temp = CalcPowerTableSizeQi0(Omega, Ordering, 0, NumShortTerms);
for (int i = 0; i < NumTermsQi0; i++) {
PowerTableQi0[NumTermsQi0+i].ki = PowerTableQi0[NumTermsQi0Temp+i].ki;
PowerTableQi0[NumTermsQi0+i].li = PowerTableQi0[NumTermsQi0Temp+i].li;
PowerTableQi0[NumTermsQi0+i].mi = PowerTableQi0[NumTermsQi0Temp+i].mi;
PowerTableQi0[NumTermsQi0+i].ni = PowerTableQi0[NumTermsQi0Temp+i].ni;
PowerTableQi0[NumTermsQi0+i].pi = PowerTableQi0[NumTermsQi0Temp+i].pi;
PowerTableQi0[NumTermsQi0+i].qi = PowerTableQi0[NumTermsQi0Temp+i].qi;
PowerTableQi0[NumTermsQi0+i].alpha = PowerTableQi0[NumTermsQi0Temp+i].alpha;
PowerTableQi0[NumTermsQi0+i].beta = PowerTableQi0[NumTermsQi0Temp+i].beta;
PowerTableQi0[NumTermsQi0+i].gamma = PowerTableQi0[NumTermsQi0Temp+i].gamma;
}
int NumTermsQiGt0Temp = CalcPowerTableSizeQiGt0(Omega, Ordering, 0, NumShortTerms);
for (int i = 0; i < NumTermsQiGt0; i++) {
PowerTableQiGt0[NumTermsQiGt0+i].ki = PowerTableQiGt0[NumTermsQiGt0Temp+i].ki;
PowerTableQiGt0[NumTermsQiGt0+i].li = PowerTableQiGt0[NumTermsQiGt0Temp+i].li;
PowerTableQiGt0[NumTermsQiGt0+i].mi = PowerTableQiGt0[NumTermsQiGt0Temp+i].mi;
PowerTableQiGt0[NumTermsQiGt0+i].ni = PowerTableQiGt0[NumTermsQiGt0Temp+i].ni;
PowerTableQiGt0[NumTermsQiGt0+i].pi = PowerTableQiGt0[NumTermsQiGt0Temp+i].pi;
PowerTableQiGt0[NumTermsQiGt0+i].qi = PowerTableQiGt0[NumTermsQiGt0Temp+i].qi;
PowerTableQiGt0[NumTermsQiGt0+i].alpha = PowerTableQiGt0[NumTermsQiGt0Temp+i].alpha;
PowerTableQiGt0[NumTermsQiGt0+i].beta = PowerTableQiGt0[NumTermsQiGt0Temp+i].beta;
PowerTableQiGt0[NumTermsQiGt0+i].gamma = PowerTableQiGt0[NumTermsQiGt0Temp+i].gamma;
}
cout << "Node " << 0 << ": " << NumTermsQi0Array[0] << " " << NumTermsQiGt0Array[0] << endl;
for (int i = 0; i < NumTermsQiGt0; i++) {
//cout << 0 << ": " << PowerTableQi0[i].ki + PowerTableQi0[i].li + PowerTableQi0[i].mi + PowerTableQi0[i].ni + PowerTableQi0[i].pi + PowerTableQi0[i].qi << " - " <<
// PowerTableQi0[i].ki << " " << PowerTableQi0[i].li << " " << PowerTableQi0[i].mi << " " << PowerTableQi0[i].ni << " " << PowerTableQi0[i].pi << " " << PowerTableQi0[i].qi << endl;
#ifdef VERBOSE
cout << 0 << ": " << PowerTableQiGt0[i].ki + PowerTableQiGt0[i].li + PowerTableQiGt0[i].mi + PowerTableQiGt0[i].ni + PowerTableQiGt0[i].pi + PowerTableQiGt0[i].qi << " - " <<
PowerTableQiGt0[i].ki << " " << PowerTableQiGt0[i].li << " " << PowerTableQiGt0[i].mi << " " << PowerTableQiGt0[i].ni << " " << PowerTableQiGt0[i].pi << " " << PowerTableQiGt0[i].qi << endl;
#endif//VERBOSE
}
cout << endl;
}
else {
MpiError = MPI_Recv(&NumTermsQi0, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&NumTermsQiGt0, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
PowerTableQi0.resize(NumTermsQi0*2);
PowerTableQiGt0.resize(NumTermsQiGt0*2);
AResultsQi0.resize(NumTermsQi0*2);
AResultsQiGt0.resize(NumTermsQiGt0*2);
BResultsQi0.resize(NumTermsQi0*2);
BResultsQiGt0.resize(NumTermsQiGt0*2);
MpiError = MPI_Recv(&PowerTableQi0[0], sizeof(rPowers)*NumTermsQi0, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&PowerTableQiGt0[0], sizeof(rPowers)*NumTermsQiGt0, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
// Create phi2 terms
for (int i = 0; i < NumTermsQi0; i++) {
PowerTableQi0[NumTermsQi0+i].ki = PowerTableQi0[i].ki - l;
PowerTableQi0[NumTermsQi0+i].li = PowerTableQi0[i].li + l;
PowerTableQi0[NumTermsQi0+i].mi = PowerTableQi0[i].mi;
PowerTableQi0[NumTermsQi0+i].ni = PowerTableQi0[i].ni;
PowerTableQi0[NumTermsQi0+i].pi = PowerTableQi0[i].pi;
PowerTableQi0[NumTermsQi0+i].qi = PowerTableQi0[i].qi;
PowerTableQi0[NumTermsQi0+i].alpha = PowerTableQi0[i].alpha;
PowerTableQi0[NumTermsQi0+i].beta = PowerTableQi0[i].beta;
PowerTableQi0[NumTermsQi0+i].gamma = PowerTableQi0[i].gamma;
}
for (int i = 0; i < NumTermsQiGt0; i++) {
PowerTableQiGt0[NumTermsQiGt0+i].ki = PowerTableQiGt0[i].ki - l;
PowerTableQiGt0[NumTermsQiGt0+i].li = PowerTableQiGt0[i].li + l;
PowerTableQiGt0[NumTermsQiGt0+i].mi = PowerTableQiGt0[i].mi;
PowerTableQiGt0[NumTermsQiGt0+i].ni = PowerTableQiGt0[i].ni;
PowerTableQiGt0[NumTermsQiGt0+i].pi = PowerTableQiGt0[i].pi;
PowerTableQiGt0[NumTermsQiGt0+i].qi = PowerTableQiGt0[i].qi;
PowerTableQiGt0[NumTermsQiGt0+i].alpha = PowerTableQiGt0[i].alpha;
PowerTableQiGt0[NumTermsQiGt0+i].beta = PowerTableQiGt0[i].beta;
PowerTableQiGt0[NumTermsQiGt0+i].gamma = PowerTableQiGt0[i].gamma;
}
}
//cout << "Node " << Node << ": " << NodeStart << " " << NodeEnd << endl;
MpiError = MPI_Barrier(MPI_COMM_WORLD);
#endif
//#ifdef USE_MPI
// MpiError = MPI_Barrier(MPI_COMM_WORLD);
//
// // Tell all processes what terms they should be evaluating.
// if (Node == 0) {
// NumTermsProc = (double)(NumShortTerms+1) / (double)TotalNodes; //@TODO: Need the typecast?
// for (int i = 1; i < TotalNodes; i++) {
// NodeStart = NumTermsProc * i;
// NodeEnd = NumTermsProc * (i+1) - 1;
// MpiError = MPI_Send(&NodeStart, 1, MPI_INT, i, 0, MPI_COMM_WORLD);
// MpiError = MPI_Send(&NodeEnd, 1, MPI_INT, i, 0, MPI_COMM_WORLD);
// }
// // This is the set of values for the root node to take.
// NodeStart = 0;
// NodeEnd = NumTermsProc * (Node+1) - 1;
// }
// else {
// MpiError = MPI_Recv(&NodeStart, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
// MpiError = MPI_Recv(&NodeEnd, 1, MPI_INT, 0, 0, MPI_COMM_WORLD, &MpiStatus);
// }
// MpiError = MPI_Barrier(MPI_COMM_WORLD);
// //@TODO: To clean this up some, we could just output this in the loop above for node 0.
// cout << "Node " << Node << ": " << NodeStart << " " << NodeEnd << endl;
//#else
// NumTermsProc = (double)(NumShortTerms+1);
// NodeStart = 0;
// NodeEnd = NumTermsProc * (Node+1) - 1;
//#endif//USE_MPI
//TimeStart = time(NULL);
//CalcARowAndBVector(Node, NumTerms, Omega, PowerTable, AResults, ARow, BResults, B, SLS, q, r2Cusp, r3Cusp, Alpha, Beta, Gamma, Kappa, Mu, sf);
CalcARowAndBVector(Node, NumTermsQi0, NumTermsQiGt0, Omega, PowerTableQi0, PowerTableQiGt0, AResultsQi0, AResultsQiGt0, ARow, BResultsQi0, BResultsQiGt0, B, SLS, SLC, l, q, r2Cusp, r3Cusp, Alpha, Beta, Gamma, Kappa, Mu, Lambda1, Lambda2, Lambda3, ShPower, sf);
//TimeEnd = time(NULL);
//cout << "Time elapsed: " << difftime(TimeEnd, TimeStart) << endl;
//OutFile << "Time elapsed: " << difftime(TimeEnd, TimeStart) << endl;
#ifdef USE_MPI
if (Node == 0) {
//@TODO: Temporary
int NumTermsQi0Temp = CalcPowerTableSizeQi0(Omega, Ordering, 0, NumShortTerms);
int NumTermsQiGt0Temp = CalcPowerTableSizeQiGt0(Omega, Ordering, 0, NumShortTerms);
for (int i = 0; i < NumTermsQi0; i++) {
AResultsQi0Final[i] = AResultsQi0[i];
AResultsQi0Final[NumTermsQi0Temp+i] = AResultsQi0[NumTermsQi0+i];
BResultsQi0Final[i] = BResultsQi0[i];
BResultsQi0Final[NumTermsQi0Temp+i] = BResultsQi0[NumTermsQi0+i];
}
for (int i = 0; i < NumTermsQiGt0; i++) {
AResultsQiGt0Final[i] = AResultsQiGt0[i];
AResultsQiGt0Final[NumTermsQiGt0Temp+i] = AResultsQiGt0[NumTermsQiGt0+i];
BResultsQiGt0Final[i] = BResultsQiGt0[i];
BResultsQiGt0Final[NumTermsQiGt0Temp+i] = BResultsQiGt0[NumTermsQiGt0+i];
}
cout << " Node " << Node << " (" << ProcessorName << ") finished computation at " << ShowTime() << endl;
// ResultsQi0 and ResultsQiGt0 already have process 0's results.
// Gather the results from the rest of the processes.
int nQi0 = NumTermsQi0Array[0], nQiGt0 = NumTermsQiGt0Array[0];
// IMPORTANT NOTE! i++ must be at the end, or else it increments before nQi0 and nQiGt0.
for (int i = 1; i < TotalNodes; nQi0 += NumTermsQi0Array[i], nQiGt0 += NumTermsQiGt0Array[i], i++) {
// Phi1 terms
MpiError = MPI_Recv(&AResultsQi0Final[nQi0], NumTermsQi0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&BResultsQi0Final[nQi0], NumTermsQi0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&AResultsQiGt0Final[nQiGt0], NumTermsQiGt0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&BResultsQiGt0Final[nQiGt0], NumTermsQiGt0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
// Phi2 terms
MpiError = MPI_Recv(&AResultsQi0Final[NumTermsQi0Temp+nQi0], NumTermsQi0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&BResultsQi0Final[NumTermsQi0Temp+nQi0], NumTermsQi0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&AResultsQiGt0Final[NumTermsQiGt0Temp+nQiGt0], NumTermsQiGt0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
MpiError = MPI_Recv(&BResultsQiGt0Final[NumTermsQiGt0Temp+nQiGt0], NumTermsQiGt0Array[i], MPI_DOUBLE, i, 0, MPI_COMM_WORLD, &MpiStatus);
}
}
else {
cout << " Node " << Node << " (" << ProcessorName << ") finished computation at " << ShowTime() << endl;
// Phi1 terms
MpiError = MPI_Send(&AResultsQi0[0], NumTermsQi0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&BResultsQi0[0], NumTermsQi0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&AResultsQiGt0[0], NumTermsQiGt0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&BResultsQiGt0[0], NumTermsQiGt0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
// Phi2 terms
MpiError = MPI_Send(&AResultsQi0[NumTermsQi0], NumTermsQi0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&BResultsQi0[NumTermsQi0], NumTermsQi0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&AResultsQiGt0[NumTermsQiGt0], NumTermsQiGt0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
MpiError = MPI_Send(&BResultsQiGt0[NumTermsQiGt0], NumTermsQiGt0, MPI_DOUBLE, 0, 0, MPI_COMM_WORLD);
}
#else
AResultsQi0Final = AResultsQi0;
BResultsQi0Final = BResultsQi0;
AResultsQiGt0Final = AResultsQiGt0;
BResultsQiGt0Final = BResultsQiGt0;
#endif
if (Node == 0) {
//@TODO: Put directly into A and B
vector <double> AResults(NumShortTerms*2), BResults(NumShortTerms*2);
CombineResults(Omega, Ordering, AResultsQi0Final, AResultsQiGt0Final, AResults, 0, NumShortTerms);
CombineResults(Omega, Ordering, BResultsQi0Final, BResultsQiGt0Final, BResults, 0, NumShortTerms);
for (int i = 0; i < NumShortTerms*2; i++) {
ARow[i+1] = AResults[i];
}
for (int i = 0; i < NumShortTerms*2; i++) {
B[i+1] = BResults[i];
}
int Multiplier;
if (l == 0) Multiplier = 1; // S-wave only has a single symmetry
else Multiplier = 2;
cout << endl << endl << "A matrix row" << endl;
for (int i = 0; i < NumShortTerms*Multiplier+1; i++) {
cout << i << " " << ARow[i] << endl;
}
cout << endl << "B vector" << endl;
for (int i = 0; i < NumShortTerms*Multiplier+1; i++) {
cout << i << " " << B[i] << endl;
}
cout << endl;
// Construct the rest of the matrix A with the short-range - short-range terms.
for (int i = 0; i < NumShortTerms*2; i++) {
for (int j = 0; j < NumShortTerms*2; j++) {
ShortTerms[i*NumShortTerms*2+j] = PhiHPhi[i][j] - 0.5*Kappa*Kappa * PhiPhi[i][j] + 1.5*PhiPhi[i][j];
}
}
}
if (Node == 0) {
//@TODO: Move further up.
// Output results to file
if (Ordering == 0)
OutFile << "Using Denton's ordering" << endl;
else
OutFile << "Using Peter Van Reeth's ordering" << endl;
OutFile << "Omega: " << Omega << endl;
OutFile << "Number of terms: " << NumShortTerms << endl;
OutFile << "Alpha: " << Alpha << " Beta: " << Beta << " Gamma: " << Gamma << endl;
OutFile << "Mu: " << Mu << endl;
OutFile << "Shielding power: " << ShPower << endl;
OutFile << "Kappa: " << Kappa << endl;
OutFile << "Lambda: " << Lambda1 << " " << Lambda2 << " " << Lambda3 << " " << endl;
OutFile << endl << "Number of quadrature points" << endl;
OutFile << "Long-long: " << q.LongLong_r1 << " " << q.LongLong_r2Leg << " " << q.LongLong_r2Lag << " " << q.LongLong_r3Leg << " " << q.LongLong_r3Lag << " " << q.LongLong_r12 << " " << q.LongLong_r13 << " " << q.LongLong_phi23 << endl;
OutFile << "Long-long 2/r23 term: " << q.LongLongr23_r1 << " " << q.LongLongr23_r2Leg << " " << q.LongLongr23_r2Lag << " " << q.LongLongr23_r3Leg << " " << q.LongLongr23_r3Lag << " " << q.LongLongr23_phi12 << " " << q.LongLongr23_r13 << " " << q.LongLongr23_r23 << endl;
OutFile << "Short-long with qi = 0: " << q.ShortLong_r1 << " " << q.ShortLong_r2Leg << " " << q.ShortLong_r2Lag << " " << q.ShortLong_r3Leg << " " << q.ShortLong_r3Lag << " " << q.ShortLong_r12 << " " << q.ShortLong_r13 << " " << q.ShortLong_phi23 << endl;
OutFile << "Short-long 2/r23 with qi = 0: " << q.ShortLongr23_r1 << " " << q.ShortLongr23_r2Leg << " " << q.ShortLongr23_r2Lag << " " << q.ShortLongr23_r3Leg << " " << q.ShortLongr23_r3Lag << " " << q.ShortLongr23_r12 << " " << q.ShortLongr23_phi13 << " " << q.ShortLongr23_r23 << endl;
OutFile << "Short-long (full) with qi > 0: " << q.ShortLongQiGt0_r1 << " " << q.ShortLongQiGt0_r2Leg << " " << q.ShortLongQiGt0_r2Lag << " " << q.ShortLongQiGt0_r3Leg << " " << q.ShortLongQiGt0_r3Lag << " " << q.ShortLongQiGt0_r12 << " " << q.ShortLongQiGt0_r13 << " " << q.ShortLongQiGt0_phi23 << endl;
OutFile << endl;
OutFile << "Cusp parameters" << endl;
OutFile << r2Cusp << " " << r3Cusp << endl;
OutFile << endl;
int Multiplier;
if (l == 0) Multiplier = 1; // S-wave only has a single symmetry
else Multiplier = 2;
OutFile << "A matrix row" << endl;
for (int i = 0; i < NumShortTerms*Multiplier+1; i++) {
OutFile << i << " " << ARow[i] << endl;
}
OutFile << endl << "B vector" << endl;
for (int i = 0; i < NumShortTerms*Multiplier+1; i++) {
OutFile << i << " " << B[i] << endl;
}
//cout << "SLS Term: " << SLS << endl;
//cout << "SLC Term: " << SLC << endl;
//cout << "SLC - CLS: " << SLC - B[0] << endl << endl;
cout << endl << "SLS Term" << endl << SLS << endl;
cout << endl << "SLC Term" << endl << SLC << endl;
cout << "SLC - CLS: " << SLC - B[0] << endl << endl;
OutFile << endl << "SLS Term" << endl << SLS << endl;
OutFile << endl << "SLC Term" << endl << SLC << endl;
OutFile << "SLC - CLS: " << SLC - B[0] << endl << endl;
double KohnPhase, InvKohnPhase, ComplexKohnPhase;
KohnPhase = Kohn(NumShortTerms, ARow, B, ShortTerms, SLS);
InvKohnPhase = InverseKohn(NumShortTerms, ARow, B, ShortTerms, SLS);
ComplexKohnPhase = ComplexKohnT(NumShortTerms, ARow, B, ShortTerms, SLS);
cout << "Kohn phase shift: " << KohnPhase << endl;
cout << "Inverse Kohn phase shift: " << InvKohnPhase << endl;
cout << "Complex Kohn phase shift: " << ComplexKohnPhase << endl << endl;
OutFile << "Kohn phase shift: " << KohnPhase << endl;
OutFile << "Inverse Kohn phase shift: " << InvKohnPhase << endl;
OutFile << "Complex (T-matrix) Kohn phase shift: " << ComplexKohnPhase << endl << endl;
double CrossSection = 4.0 * PI * sin(KohnPhase) * sin(KohnPhase); // (2l+1) = 1 with l = 0
cout << "Kohn partial wave cross section: " << CrossSection << endl;
OutFile << "Kohn Partial wave cross section: " << CrossSection << endl;
CrossSection = 4.0 * PI * sin(InvKohnPhase) * sin(InvKohnPhase);
cout << "Inverse Kohn partial wave cross section: " << CrossSection << endl;
OutFile << "Inverse Kohn partial wave cross section: " << CrossSection << endl;
CrossSection = 4.0 * PI * sin(ComplexKohnPhase) * sin(ComplexKohnPhase);
cout << "Complex (T-matrix) Kohn partial wave cross section: " << CrossSection << endl << endl;
OutFile << "Complex (T-matrix) Kohn partial wave cross section: " << CrossSection << endl << endl;
}
// Cleanup
if (Node == 0) {
TimeEnd = time(NULL);
cout << "Time elapsed: " << difftime(TimeEnd, TimeStart) << endl;
OutFile << "Time elapsed: " << difftime(TimeEnd, TimeStart) << endl;
ParameterFile.close();
OutFile.close();
FileShortRange.close();
if (NumShortTerms > 0) {
delete [] PhiHPhi[0];
delete [] PhiHPhi;
delete [] PhiPhi[0];
delete [] PhiPhi;
}
}
cout << endl;
fflush(stdout);
#ifdef USE_MPI
//MpiError = MPI_File_close(&MpiLog);
MpiError = MPI_Finalize();
#endif
return 0;
}