SG++

This class solves an (lhs + lambda*I) * alpha = b system of linear equations after the offline and online phases are done. More...

#include <DBMatDMSOrthoAdapt.hpp>

## Public Member Functions

Constructor. More...

void solve (sgpp::base::DataMatrix &T_inv, sgpp::base::DataMatrix &Q, sgpp::base::DataMatrix &B, sgpp::base::DataVector &b, sgpp::base::DataVector &alpha)
Solves the system after the corresponding offline and online objects are done with decomposing and adaptivity, resp. More...

Public Member Functions inherited from sgpp::datadriven::DBMatDecompMatrixSolver
DBMatDecompMatrixSolver ()

Public Member Functions inherited from sgpp::solver::SGSolver
double getEpsilon ()
gets the the epsilon, that is used in the SGSolver More...

size_t getNumberIterations ()
function that returns the number of needed solve steps More...

double getResiduum ()
function the returns the residuum (current or final), error of the solver More...

void setEpsilon (double eps)
resets the epsilon, that is used in the SGSolver More...

void setMaxIterations (size_t nIterations)
resets the number of maximum iterations More...

SGSolver (size_t nMaximumIterations, double epsilon)
Std-Constructor. More...

virtual ~SGSolver ()
Std-Destructor. More...

Protected Attributes inherited from sgpp::solver::SGSolver
double myEpsilon
epsilon needed in the, e.g. final error in the iterative solver, or a timestep More...

size_t nIterations
Number of Iterations needed for the solve. More...

size_t nMaxIterations
Number of maximum iterations for cg. More...

double residuum
residuum More...

## Detailed Description

This class solves an (lhs + lambda*I) * alpha = b system of linear equations after the offline and online phases are done.

default

Constructor.

## Member Function Documentation

 void sgpp::datadriven::DBMatDMSOrthoAdapt::solve ( sgpp::base::DataMatrix & T_inv, sgpp::base::DataMatrix & Q, sgpp::base::DataMatrix & B, sgpp::base::DataVector & b, sgpp::base::DataVector & alpha )

Solves the system after the corresponding offline and online objects are done with decomposing and adaptivity, resp.

The computation done: alpha = Q*T_inv*Q^t*b + B*b

Parameters
 T_inv Inverse of a tridiagonal matrix Q Orthogonal matrix, part of hessenberg_decomp of the lhs matrix B Storage of the online objects refined/coarsened points b The right side of the system alpha The solution vector of the system, computed values go there