Nonoverlapping parallel CG solver with Jacobi preconditioner.
More...
#include <dune/pdelab/backend/istl/novlpistlsolverbackend.hh>
|
| ISTLBackend_NOVLP_CG_Jacobi (const GFS &gfs_, unsigned maxiter_=5000, int verbose_=1) |
| make a linear solver object More...
|
|
template<class V > |
V::ElementType | norm (const V &v) const |
| compute global norm of a vector More...
|
|
template<class M , class V , class W > |
void | apply (M &A, V &z, W &r, typename Dune::template FieldTraits< typename V::ElementType >::real_type reduction) |
| solve the given linear system More...
|
|
const LinearSolverResult< double > & | result () const |
| Return access to result data. More...
|
|
template<class GFS>
class Dune::PDELab::ISTLBackend_NOVLP_CG_Jacobi< GFS >
Nonoverlapping parallel CG solver with Jacobi preconditioner.
◆ ISTLBackend_NOVLP_CG_Jacobi()
make a linear solver object
- Parameters
-
gfs_ | A grid function space |
maxiter_ | Maximum number of iterations to do. |
verbose_ | Verbosity level, directly handed to the CGSolver. |
◆ apply()
template<class GFS >
template<class M , class V , class W >
solve the given linear system
- Parameters
-
A | The matrix to solve. Should be a matrix from one of PDELabs ISTL backends (only ISTLBCRSMatrixBackend at the moment). |
z | The solution vector to be computed |
r | Right hand side |
reduction | to be achieved |
Solve the linear system A*z=r such that norm(A*z0-r)/norm(A*z-r) < reduction where z0 is the initial value of z.
◆ norm()
template<class GFS >
template<class V >
compute global norm of a vector
- Parameters
-
v | The vector to compute the norm of. Should be an inconsistent vector (i.e. the entries corresponding a DoF on the border should only contain the summand of this process). |
◆ result()
Return access to result data.
The documentation for this class was generated from the following file: