SparseGenMatrix::SparseGenMatrix( int rows, int cols, int nnz,
int krowM[], int jcolM[], double M[],
int deleteElts)
: m_Mt(NULL)
{
//cout << "SparseGenMatrix1 " << rows << " " << cols << " " << nnz << endl;
mStorage = SparseStorageHandle( new SparseStorage( rows, cols,nnz,
krowM, jcolM, M,
deleteElts) );
}
//public Ma57Solver::Ma57Solver( SparseSymMatrix * sgm, const int numOfNegEigVal_in ):this(sgm){};
void Ma57Solver::SetUpMa57Solver( SparseSymMatrix * sgm )
{
irowM = 0;
jcolM = 0;
fact = 0;
ifact = 0;
keep = 0;
iworkn = NULL; dworkn = NULL;
niworkn = ndworkn = 0;
ipessimism = 1.05;
rpessimism = 1.05;
FNAME(ma57id)( cntl, icntl );
icntl[1-1] = 0; // don't print warning messages
icntl[2-1] = 0; // no Warning messages
icntl[4-1] = 1; // no statistics messages
icntl[5-1] = 0; // no Print messages.
icntl[6-1] = gMA57_Ordering; // 4 use Metis; 5 automatic choice(MA47 or Metis); 3 min
// degree ordering as in MA27; 2 use MC47;
icntl[7-1] = 1; /* Pivoting strategy. */
icntl[9-1] = 10; // up to 10 steps of iterative refinement
icntl[11-1] = 16;
icntl[12-1] = 16;
icntl[15-1] = 1;
icntl[16-1] = 0;
setValTreatAsZero(1.e-20);
// set initial value of Threshold parameter
setPivotTol(gHSL_PivotLV);
// set the required precision for each linear system solve
kPrecision = 1.e-9;
mStorage = SparseStorageHandle( sgm->getStorage() );
n = mStorage->n;
if (gOuterSolve >=3 && separateHandDiag==1){
nnz = mStorage->krowM[n] + n;
M = new double[nnz];
memcpy( &M[0], &mStorage->M[0], mStorage->krowM[n] * sizeof( double ) );
memcpy( &M[mStorage->krowM[n]], &mStorage->additiveDiag[0], n * sizeof( double ) );
}else {
nnz = mStorage->krowM[n];
M = mStorage->M;
}
}
Ma57Solver::Ma57Solver( SparseSymMatrix * sgm )
{
irowM = 0;
jcolM = 0;
fact = 0;
ifact = 0;
keep = 0;
iworkn = NULL; dworkn = NULL;
niworkn = ndworkn = 0;
ipessimism = 1.4;
rpessimism = 1.4;
FNAME(ma57id)( cntl, icntl );
//icntl[1] = -1; // don't print warning messages
icntl[8] = 10; // up to 10 steps of iterative refinement
icntl[5] = 5; // 4 use Metis; 5 automatic choice(MA47 or Metis); 3 min
// degree ordering as in MA27; 2 use MC47;
// set initial value of "Treat As Zero" parameter
kTreatAsZero = 1.e-10; this->setTreatAsZero();
// set initial value of Threshold parameter
kThresholdPivoting = 1.e-5; this->setThresholdPivoting();
// set the largest value of ThresholdPivoting parameter we are
// willing to tolerate.
kThresholdPivotingMax = 1.e-1;
// set the increase factor for ThresholdPivoting parameter
kThresholdPivotingFactor = 10.0;
// set the required precision for each linear system solve
kPrecision = 1.e-9;
mStorage = SparseStorageHandle( sgm->getStorage() );
n = mStorage->n;
M = mStorage->M;
nnz = mStorage->krowM[n];
//!
ma27=new Ma27Solver(sgm);
sgm_=sgm;
}
Ma57Solver::Ma57Solver( SparseSymMatrix * sgm )
{
irowM = 0;
jcolM = 0;
fact = 0;
ifact = 0;
keep = 0;
ipessimism = 1.2;
rpessimism = 1.2;
ma57id_( cntl, icntl );
icntl[1] = -1; // don't print warning messages
icntl[8] = 10; // up to 10 steps of iterative refinement
// set initial value of "Treat As Zero" parameter
kTreatAsZero = 1.e-10;
this->setTreatAsZero();
// set initial value of Threshold parameter
kThresholdPivoting = 1.e-8;
this->setThresholdPivoting();
// set the largest value of ThresholdPivoting parameter we are
// willing to tolerate.
kThresholdPivotingMax = 1.e-2;
// set the increase factor for ThresholdPivoting parameter
kThresholdPivotingFactor = 10.0;
// set the required precision for each linear system solve
kPrecision = 1.e-7;
mStorage = SparseStorageHandle( sgm->getStorage() );
n = mStorage->n;
M = mStorage->M;
nnz = mStorage->krowM[n];
}
SparseGenMatrix::SparseGenMatrix( int rows, int cols, int nnz )
: m_Mt(NULL)
{
mStorage = SparseStorageHandle( new SparseStorage( rows, cols, nnz ) );
}
