//==============================================================================
int Ifpack2_Amesos::
ApplyInverse(const Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& X, Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& Y) const
{
if (IsComputed() == false)
IFPACK2_CHK_ERR(-1);
if (X.NumVectors() != Y.NumVectors())
IFPACK2_CHK_ERR(-1); // wrong input
Time_->ResetStartTime();
// AztecOO gives X and Y pointing to the same memory location,
// need to create an auxiliary vector, Xcopy
Teuchos::RCP<const Tpetra_MultiVector> Xcopy;
if (X.Pointers()[0] == Y.Pointers()[0])
Xcopy = Teuchos::rcp( new Tpetra_MultiVector(X) );
else
Xcopy = Teuchos::rcp( &X, false );
Problem_->SetLHS(&Y);
Problem_->SetRHS((Tpetra_MultiVector*)Xcopy.get());
IFPACK2_CHK_ERR(Solver_->Solve());
++NumApplyInverse_;
ApplyInverseTime_ += Time_->ElapsedTime();
return(0);
}
//=============================================================================
// This function finds X such that LDU Y = X or U(trans) D L(trans) Y = X for multiple RHS
int Ifpack2_ILU::ApplyInverse(const Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& X,
Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& Y) const
{
#ifdef ENABLE_IFPACK2_ILU_TEUCHOS_TIMERS
TEUCHOS_FUNC_TIME_MONITOR("Ifpack2_ILU::ApplyInverse");
#endif
if (!IsComputed())
IFPACK2_CHK_ERR(-3);
if (X.NumVectors() != Y.NumVectors())
IFPACK2_CHK_ERR(-2);
Time_.ResetStartTime();
// AztecOO gives X and Y pointing to the same memory location,
// need to create an auxiliary vector, Xcopy
Teuchos::RCP< const Tpetra_MultiVector > Xcopy;
if (X.Pointers()[0] == Y.Pointers()[0])
Xcopy = Teuchos::rcp( new Tpetra_MultiVector(X) );
else
Xcopy = Teuchos::rcp( &X, false );
IFPACK2_CHK_ERR(Solve(Ifpack2_ILU::UseTranspose(), *Xcopy, Y));
// approx is the number of nonzeros in L and U
ApplyInverseFlops_ += X.NumVectors() * 4 *
(L_->NumGlobalNonzeros() + U_->NumGlobalNonzeros());
++NumApplyInverse_;
ApplyInverseTime_ += Time_.ElapsedTime();
return(0);
}
/*
//old Apply (no communication)
int Ifpack2_NodeFilter::Apply(const Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& X,
Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& Y) const
{
// skip expensive checks, I suppose input data are ok
Y.PutScalar(0.0);
int NumVectors = Y.NumVectors();
double** X_ptr;
double** Y_ptr;
X.ExtractView(&X_ptr);
Y.ExtractView(&Y_ptr);
for (int i = 0 ; i < NumRows_ ; ++i) {
int Nnz;
int ierr = Matrix_->ExtractMyRowCopy(i,MaxNumEntriesA_,Nnz,&Values_[0],
&Indices_[0]);
IFPACK2_CHK_ERR(ierr);
for (int j = 0 ; j < Nnz ; ++j) {
if (Indices_[j] < NumRows_ ) {
for (int k = 0 ; k < NumVectors ; ++k)
Y_ptr[k][i] += Values_[j] * X_ptr[k][Indices_[j]];
}
}
}
return(0);
}
*/
int Ifpack2_NodeFilter::Apply(const Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& X, Tpetra_MultiVector<Scalar,LocalOrdinal,GlobalOrdinal,Node>& Y) const {
//
// This function forms the product Y = A * X.
//
int NumEntries;
int NumVectors = X.NumVectors();
if (NumVectors!=Y.NumVectors()) {
EPETRA_CHK_ERR(-1); // Need same number of vectors in each MV
}
// Make sure Import and Export Vectors are compatible
UpdateImportVector(NumVectors);
UpdateExportVector(NumVectors);
double ** Xp = (double**) X.Pointers();
double ** Yp = (double**) Y.Pointers();
// If we have a non-trivial importer, we must import elements that are permuted or are on other processors
if (Importer()!=0) {
EPETRA_CHK_ERR(ImportVector_->Import(X, *Importer(), Insert));
Xp = (double**)ImportVector_->Pointers();
}
// If we have a non-trivial exporter, we must export elements that are permuted or belong to other processors
if (Exporter()!=0) {
Yp = (double**)ExportVector_->Pointers();
}
// Do actual computation
assert(ovA_!=0);
int *MyRows;
double *MyValues;
int *MyIndices;
if(Acrs_){
// A rows - CrsMatrix Case
IFPACK2_CHK_ERR(Acrs_->ExtractCrsDataPointers(MyRows,MyIndices,MyValues));
//special case NumVectors==1
if (NumVectors==1) {
for(int i=0;i<NumMyRowsA_;i++) {
int LocRow=Ar_LIDMap_[i];
double sum = 0.0;
for(int j = MyRows[i]; j < MyRows[i+1]; j++)
sum += MyValues[j]*Xp[0][Ac_LIDMap_[MyIndices[j]]];
Yp[0][LocRow] = sum;
}
}
else {
for(int i=0;i<NumMyRowsA_;i++) {
int LocRow=Ar_LIDMap_[i];
for (int k=0; k<NumVectors; k++) {
double sum = 0.0;
for(int j = MyRows[i]; j < MyRows[i+1]; j++)
sum += MyValues[j]*Xp[k][Ac_LIDMap_[MyIndices[j]]];
Yp[k][LocRow] = sum;
}
}
}
}
else{
// A rows - RowMatrix Case
MyValues=&Values_[0];
MyIndices=&Indices_[0];
for(int i=0;i<NumMyRowsA_;i++) {
ovA_->A().ExtractMyRowCopy(i,MaxNumEntries_,NumEntries,MyValues,MyIndices);
int LocRow=Ar_LIDMap_[i];
for (int k=0; k<NumVectors; k++) {
double sum = 0.0;
for(int j = 0; j < NumEntries; j++)
sum += MyValues[j]*Xp[k][Ac_LIDMap_[MyIndices[j]]];
Yp[k][LocRow] = sum;
}
}
}
// B rows, always CrsMatrix
IFPACK2_CHK_ERR(ovA_->B().ExtractCrsDataPointers(MyRows,MyIndices,MyValues));
//special case NumVectors==1
if (NumVectors==1) {
for(int i=0;i<NumMyRowsB_;i++) {
int LocRow=Br_LIDMap_[i];
double sum = 0.0;
for(int j = MyRows[i]; j < MyRows[i+1]; j++)
sum += MyValues[j]*Xp[0][Bc_LIDMap_[MyIndices[j]]];
Yp[0][LocRow] = sum;
}
} else {
for(int i=0;i<NumMyRowsB_;i++) {
int LocRow=Br_LIDMap_[i];
for (int k=0; k<NumVectors; k++) { //FIXME optimization, check for NumVectors=1
double sum = 0.0;
for(int j = MyRows[i]; j < MyRows[i+1]; j++)
sum += MyValues[j]*Xp[k][Bc_LIDMap_[MyIndices[j]]];
Yp[k][LocRow] = sum;
}
}
}
if (Exporter()!=0) {
Y.PutScalar(0.0); // Make sure target is zero
Y.Export(*ExportVector_, *Exporter(), Add); // Fill Y with Values from export vector
}
// Handle case of rangemap being a local replicated map
if (!OperatorRangeMap().DistributedGlobal() && Comm().NumProc()>1) EPETRA_CHK_ERR(Y.Reduce());
return(0);
} //Apply