Class : FullReorthogonalizedLanczosAlgorithm
FullReorthogonalizedLanczosAlgorithm :
prototype: FullReorthogonalizedLanczosAlgorithm (AbstractArchitecture* , int , int maxIter = , bool strongConvergence = false )
description:default constructor
input parameters:
architecture = architecture to use for matrix operations
nbrEigenvalue = number of wanted eigenvalues
maxIter = an approximation of maximal number of iteration
strongConvergence = flag indicating if the convergence test has to be done on the latest wanted eigenvalue (false) or all the wanted eigenvalue (true)
FullReorthogonalizedLanczosAlgorithm :
prototype: FullReorthogonalizedLanczosAlgorithm (const FullReorthogonalizedLanczosAlgorithm& algorithm )
description:copy constructor
input parameters:
algorithm = algorithm from which new one will be created
~FullReorthogonalizedLanczosAlgorithm :
prototype: ~FullReorthogonalizedLanczosAlgorithm ()
description:destructor
InitializeLanczosAlgorithm :
prototype: void InitializeLanczosAlgorithm ()
description:initialize Lanczos algorithm with a random vector
InitializeLanczosAlgorithm :
prototype: void InitializeLanczosAlgorithm (const Vector& vector )
description:initialize Lanczos algorithm with a given vector
input parameters:
vector = reference to the vector used as first step vector
GetGroundState :
prototype: Vector& GetGroundState ()
description:get last produced vector
input parameters:
return value: reference on last produced vector
GetEigenstates :
prototype: Vector* GetEigenstates (int nbrEigenstates )
description:get the n first eigenstates
input parameters:
nbrEigenstates = number of needed eigenstates
return value: array containing the eigenstates
RunLanczosAlgorithm :
prototype: void RunLanczosAlgorithm (int nbrIter )
description:run current Lanczos algorithm (continue from previous results if Lanczos algorithm has already been run)
input parameters:
nbrIter = number of iteration to do
TestConvergence :
prototype: bool TestConvergence ()
description:test if convergence has been reached
input parameters:
return value: true if convergence has been reached
Back to index.