Description

Generalized iterative eigenvalue solver implementing Krylov-Schur shift-and-invert method for real symmetric matrices.

Features:

generalized eigenvalue problem
symmetric real sparse matrices
shift-and-invert with real shift

#include <ChSymGenEigenvalueSolver.h>

Inheritance diagram for chrono::modal::ChSymGenEigenvalueSolverKrylovSchur:

[legend]

Collaboration diagram for chrono::modal::ChSymGenEigenvalueSolverKrylovSchur:

[legend]

Public Member Functions
virtual int	Solve (const ChSparseMatrix &A, const ChSparseMatrix &B, ChMatrixDynamic< ScalarType > &eigvects, ChVectorDynamic< ScalarType > &eigvals, int num_modes, ScalarType shift) const override
	Solve the generalized eigenvalue problem Aeigvects = Beigvects*diag(eigvals) A and B are expected to be symmetric and real 'eigvects' will be resized to [A.rows() x num_modes] 'eigvals' will be resized to [num_modes] the number of converged eigenvalues is returned.

Public Member Functions inherited from chrono::modal::ChSymGenEigenvalueSolver
virtual int	Solve (const ChSparseMatrix &A, const ChSparseMatrix &B, ChMatrixDynamic< ScalarType > &eigvects, ChVectorDynamic< ScalarType > &eigvals, int num_modes, ScalarType shift) const =0
	Solve the generalized eigenvalue problem Aeigvects = Beigvects*diag(eigvals) A and B are expected to be symmetric and real 'eigvects' will be resized to [A.rows() x num_modes] 'eigvals' will be resized to [num_modes] the number of converged eigenvalues is returned.

Public Member Functions inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
double	GetTimeMatrixAssembly () const
	Get cumulative time for matrix assembly.

double	GetTimeEigenSetup () const
	Get cumulative time for the eigensolver setup.

double	GetTimeEigenSolver () const
	Get cumulative time for the eigensolver solution.

double	GetTimeSolutionPostProcessing () const
	Get cumulative time for post-solver solution postprocessing.

Additional Inherited Members
Public Types inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
using	ScalarType = double

Static Public Member Functions inherited from chrono::modal::ChSymGenEigenvalueSolver
static void	GetNaturalFrequencies (const ChVectorDynamic< ScalarType > &eigvals, ChVectorDynamic< double > &freq)
	Retrieve the natural frequencies from an eigenvalues array.

static double	GetNaturalFrequency (ScalarType eigval)
	Retrieve the natural frequencies from an eigenvalue.

static ScalarType	GetOptimalShift (double freq)
	Get the optimal shift corresponding to a given frequency.

Static Public Member Functions inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
static void	SortRitzPairs (ChVectorDynamic< ScalarType > &eigvals, ChMatrixDynamic< ScalarType > &eigvects, std::function< bool(const ChVectorDynamic< ScalarType > &, int, int)> ordering_fun=[](const ChVectorDynamic< std::complex< double >> &eigv, int a, int b) { return std::abs(eigv(a))< std::abs(eigv(b));})
	Sort the eigenvalues and eigenvectors in the order specified by the ordering function in-place.

static double	GetMaxResidual (const ChSparseMatrix &A, const ChSparseMatrix &B, const ChMatrixDynamic< ScalarType > &eigvects, const ChVectorDynamic< ScalarType > &eigvals)

static double	GetMaxResidual (const ChSparseMatrix &K, const ChSparseMatrix &M, const ChSparseMatrix &Cq, const ChMatrixDynamic< ScalarType > &eigvects, const ChVectorDynamic< ScalarType > &eigvals)

static double	GetMaxResidual (const ChSparseMatrix &K, const ChSparseMatrix &R, const ChSparseMatrix &M, const ChSparseMatrix &Cq, const ChMatrixDynamic< ScalarType > &eigvects, const ChVectorDynamic< ScalarType > &eigvals)

static double	BuildUndampedSystem (const ChSparseMatrix &M, const ChSparseMatrix &K, const ChSparseMatrix &Cq, ChSparseMatrix &A, ChSparseMatrix &B, bool scaleCq)
	Build the quadratic eigenvalue problem matrix for the undamped case. More...

static double	BuildDampedSystem (const ChSparseMatrix &M, const ChSparseMatrix &R, const ChSparseMatrix &K, const ChSparseMatrix &Cq, ChSparseMatrix &A, ChSparseMatrix &B, bool scaleCq)
	Build the quadratic eigenvalue problem matrix for the damped case. More...

static void	InsertUniqueRitzPairs (const ChVectorDynamic< ScalarType > &eigvals_source, const ChMatrixDynamic< ScalarType > &eigvects_source, ChVectorDynamic< ScalarType > &eigvals_total, ChMatrixDynamic< ScalarType > &eigvects_total, std::function< double(ScalarType)> freq_from_eigval_fun, int &found_eigs, int num_eigvals_source, double equal_freq_tolerance=1e-4)
	Add unique ritz pairs to a wider set of ritz pairs. More...

Public Attributes inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
double	tolerance
	tolerance for the iterative solver.

int	max_iterations
	upper limit for the number of iterations. If too low might not converge.

bool	verbose
	turn to true to see some diagnostic.

bool	sort_ritz_pairs
	sort the eigenvalues based on the smallest absolute value

Static Protected Member Functions inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
static Eigen::PermutationMatrix< Eigen::Dynamic, Eigen::Dynamic >	GetPermutationMatrix (int num_elements, std::function< bool(int, int)> ordering_fun)
	Returns the permutation matrix to sort the given elements based on the ordering function.

Protected Attributes inherited from chrono::modal::ChGeneralizedEigenvalueSolver< double >
ChTimer	m_timer_matrix_assembly
	timer for matrix assembly

ChTimer	m_timer_eigen_setup
	timer for eigensolver setup

ChTimer	m_timer_eigen_solver
	timer for eigensolver solution

ChTimer	m_timer_solution_postprocessing
	timer for conversion of eigensolver solution

const int	m_min_subspace_size

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

/builds/uwsbel/chrono/src/chrono_modal/ChSymGenEigenvalueSolver.h
/builds/uwsbel/chrono/src/chrono_modal/ChSymGenEigenvalueSolver.cpp

Description

Public Member Functions

Additional Inherited Members