imstk::ProjectedGaussSeidelSolver< Scalar > Class Template Reference

Solves a linear system using the projected gauss seidel method. Only good for diagonally dominant systems, must have elements on diagonals though. The initial guess (start) is always zero, convergence value may be specified with epsilon, relaxation decreases the step size (useful when may rows exist in A) More...

#include <imstkProjectedGaussSeidelSolver.h>

Public Member Functions
void	setA (Eigen::SparseMatrix< Scalar > *A)
void	setMaxIterations (const unsigned int maxIterations)
	Set the maximum number of iterations.
void	setRelaxation (const Scalar relaxation)
	Similar to step size can be used to avoid overshooting the solution.
void	setEpsilon (const Scalar epsilon)
	Stops when energy=(x_i+1-x_i).norm() < epsilon, when the solution isn't changing anymore.
const double	getEnergy () const
	Energy is defined as energy=(x_i+1-x_i).norm()
Eigen::Matrix< Scalar, -1, 1 > &	solve (const Eigen::Matrix< Scalar, -1, 1 > &b, const Eigen::Matrix< Scalar, -1, 2 > &cu)

Detailed Description

template<typename Scalar>
class imstk::ProjectedGaussSeidelSolver< Scalar >

Solves a linear system using the projected gauss seidel method. Only good for diagonally dominant systems, must have elements on diagonals though. The initial guess (start) is always zero, convergence value may be specified with epsilon, relaxation decreases the step size (useful when may rows exist in A)

Definition at line 18 of file imstkRigidBodyModel2.h.

---

The documentation for this class was generated from the following files:

• Source/DynamicalModels/ObjectModels/imstkRigidBodyModel2.h
• Source/Solvers/imstkProjectedGaussSeidelSolver.h