Gauss-Seidel based solver implementation. More...
#include <gs_solver.hpp>
Public Member Functions | |
| GSSolver (double w=1.66, double eps=1.0e-6, uint32_t imax=10000, double newton_Reps=1.0e-5, double newton_dXeps=1.0e-6, uint32_t newton_imax=10) | |
| Constructor. | |
| ~GSSolver () | |
| Destructor. | |
| virtual void | solve (const Problem &p, Vector &X) |
| Solve problem p defined in geometry g. Initial guess and solution are in vector X. | |
| virtual void | reset (void) |
| Reset solver. | |
| void | set_newton_imax (uint32_t newton_imax) |
| Sets maximum iteration count for Newton-Raphson steps. | |
| void | set_newton_residual_eps (double newton_Reps) |
| Sets the accuracy request for Newton-Raphson residual. | |
| void | set_newton_step_eps (double newton_dXeps) |
| Sets the accuracy request for Newton-Raphson step size. | |
Static Public Member Functions | |
| static bool | gauss_seidel (const CRowMatrix &mat, const Vector &rhs, Vector &sol, uint32_t &imax, double &eps, double w) |
| Direct interface to gauss_seidel matrix solver. | |
Detailed Description
Gauss-Seidel based solver implementation.
An implementation of virtual class Solver using Gauss-Seidel/Successive over relaxation method for solving matrix equation. Linear and nonlinear solvers. Nonlinear solver not implemented.
The Gauss-Seidel solver has special limitations for the problem type because of convergence issues. The problem has to have first order neumann edges and the edge smoothing must be switched off.
Constructor & Destructor Documentation
| GSSolver::GSSolver | ( | double | w = 1.66, |
| double | eps = 1.0e-6, |
||
| uint32_t | imax = 10000, |
||
| double | newton_Reps = 1.0e-5, |
||
| double | newton_dXeps = 1.0e-6, |
||
| uint32_t | newton_imax = 10 |
||
| ) |
Constructor.
| GSSolver::~GSSolver | ( | ) | [inline] |
Destructor.
Member Function Documentation
| static bool GSSolver::gauss_seidel | ( | const CRowMatrix & | mat, |
| const Vector & | rhs, | ||
| Vector & | sol, | ||
| uint32_t & | imax, | ||
| double & | eps, | ||
| double | w | ||
| ) | [static] |
Direct interface to gauss_seidel matrix solver.
Solves matrix problem mat * sol = rhs. The matrix is sorted to be on the ascending order as required by the GAUSS_SEIDEL library.
| virtual void GSSolver::reset | ( | void | ) | [virtual] |
Reset solver.
This is a signal from the problem that the problem has changed and internal caches (if they exist) in the solver should be resetted.
Implements Solver.
| void GSSolver::set_newton_imax | ( | uint32_t | newton_imax | ) | [inline] |
Sets maximum iteration count for Newton-Raphson steps.
| void GSSolver::set_newton_residual_eps | ( | double | newton_Reps | ) | [inline] |
Sets the accuracy request for Newton-Raphson residual.
| void GSSolver::set_newton_step_eps | ( | double | newton_dXeps | ) | [inline] |
Sets the accuracy request for Newton-Raphson step size.
Solve problem p defined in geometry g. Initial guess and solution are in vector X.
The documentation for this class was generated from the following file:
