Eigendecomposition of a General Size Matrix
[General Size Matrices]

Classes
struct	EigenStuff
struct	EigenStuff
struct	Gan_SymMatEigenStruct
	Structure for eigendecomposition of an accumulated symmetric matrix. More...
Functions
Gan_Bool	gan_symmat_eigen (Gan_SquMatrix *A, Gan_SquMatrix *W, Gan_Matrix *Z, Gan_Bool sort, double *work, unsigned long work_size)
	Computes eigenvalues and eigenvectors of a symmetric matrix.
Gan_Bool	gan_symmatf_eigen (Gan_SquMatrix_f *A, Gan_SquMatrix_f *W, Gan_Matrix_f *Z, Gan_Bool sort, float *work, unsigned long work_size)
	Computes eigenvalues and eigenvectors of a symmetric matrix.
Gan_SymMatEigenStruct *	gan_symeigen_form_data (Gan_SymMatEigenStruct *sme, long max_size, double *work_data, unsigned long work_data_size)
	Forms a symmetric matrix eigendecomposition structure.
void	gan_symeigen_free (Gan_SymMatEigenStruct *sme)
	Frees a symmetric matrix eigendecomposition structure.
Gan_Bool	gan_symeigen_reset (Gan_SymMatEigenStruct *sme, long size)
	Initialises an accumulated outer product matrix to zero.
Gan_Bool	gan_symeigen_increment (Gan_SymMatEigenStruct *sme, double scale,...)
	Increments an accumulated outer product matrix.
Gan_Bool	gan_symeigen_solve (Gan_SymMatEigenStruct *sme)
	Eigendecomposes a matrix accumulated by summing outer products.
Gan_SymMatEigenStruct *	gan_symeigen_form (Gan_SymMatEigenStruct *sme, long max_size)
	Macro: Forms a symmetric matrix eigendecomposition structure.
Gan_SymMatEigenStruct *	gan_symeigen_alloc (long max_size)
	Macro: Allocates a symmetric matrix eigendecomposition structure.

---

Function Documentation

Gan_SymMatEigenStruct* gan_symeigen_alloc (  long  max_size  )

Macro: Allocates a symmetric matrix eigendecomposition structure. Parameters: max_size  The envisaged largest size of matrix Returns: Pointer to the allocated structure, or NULL on failure. Allocates a structure with initial values to compute the eigendecomposition of a sum of vector outer products. Macro call to gan_symeigen_form_data(). See also: gan_symeigen_form_data(), gan_symeigen_form().

Gan_SymMatEigenStruct* gan_symeigen_form (  Gan_SymMatEigenStruct *  sme, long  max_size )

Macro: Forms a symmetric matrix eigendecomposition structure. Parameters: sme  A pointer to a structure or NULL max_size  The envisaged largest size of matrix Returns: Pointer to the formed structure, or NULL on failure. Allocates/fills a structure with initial values to compute the eigendecomposition of a sum of vector outer products. Macro call to gan_symeigen_form_data(). See also: gan_symeigen_form_data(), gan_symeigen_alloc().

Gan_SymMatEigenStruct * gan_symeigen_form_data (  Gan_SymMatEigenStruct *  sme, long  max_size, double *  work_data, unsigned long  work_data_size )

Forms a symmetric matrix eigendecomposition structure. Parameters: sme  A pointer to a structure or NULL max_size  The envisaged largest size of matrix work_data  Workspace array for CLAPACK eigendecomposition work_data_size  Size of work_data array Returns: Pointer to structure, or NULL on failure. Allocates/fills a structure with initial values to compute the eigendecomposition of a sum of vector outer products. See also: gan_symeigen_form(), gan_symeigen_alloc().

void gan_symeigen_free (  Gan_SymMatEigenStruct *  sme  )

Frees a symmetric matrix eigendecomposition structure. Parameters: sme  A pointer to a structure Returns: No value. Frees a structure created to compute the eigenvalues and eigenvectors of a symmetric matrix accumulated as a sum of vector outer products. See also: gan_symeigen_alloc().

Gan_Bool gan_symeigen_increment (  Gan_SymMatEigenStruct *  sme, double  scale,   ... )

Increments an accumulated outer product matrix. Parameters: sme  A pointer to a structure scale  A scaling for the outer product ...  A variable argument list of vector components Returns: GAN_TRUE on success, GAN_FALSE on failure. Increments an accumulated outer product matrix stored inside the sme structure with the outer product of a vector passed into the function in a variable argument list. The outer product is scaled by the given scaling factor scale before the matrix is incremented (equivalent to scaling the vector by the square-root of scale). See also: gan_symeigen_alloc().

Gan_Bool gan_symeigen_reset (  Gan_SymMatEigenStruct *  sme, long  size )

Initialises an accumulated outer product matrix to zero. Parameters: sme  A pointer to a structure size  The size of the matrix Returns: GAN_TRUE on success, GAN_FALSE on failure. Initialises an accumulated outer product matrix to zero. The size of the matrix is also set. See also: gan_symeigen_alloc().

Gan_Bool gan_symeigen_solve (  Gan_SymMatEigenStruct *  sme  )

Eigendecomposes a matrix accumulated by summing outer products. Parameters: sme  A pointer to a structure Returns: GAN_TRUE on success, GAN_FALSE on failure. Solves for the eigenvalues and eigenvectors of a matrix accumulated within the sme structure as a sum of vector outer products. See also: gan_symeigen_alloc(), gan_symeigen_increment().

Gan_Bool gan_symmat_eigen (  Gan_SquMatrix *  A, Gan_SquMatrix *  W, Gan_Matrix *  Z, Gan_Bool  sort, double *  work, unsigned long  work_size )

Computes eigenvalues and eigenvectors of a symmetric matrix. Parameters: A  Input matrix, overwritten on output W  Diagonal matrix of eigenvalues Z  Matrix of eigenvectors or NULL sort  Whether to sort eigenvalues and eigenvectors work  Workspace array work_size  size of workspace Returns: GAN_TRUE on success, or GAN_FALSE on failure. If Z is non-NULL it is filled with the eigenvectors of symmetric matrix A. The eigenvalues of A are written into W. If work is not NULL and LAPACK is being used, the size of the work array (work_size) must be at least 3 times the size of A. If sort is passed as GAN_TRUE, the eigenvalues are sorted into ascending order.

Gan_Bool gan_symmatf_eigen (  Gan_SquMatrix_f *  A, Gan_SquMatrix_f *  W, Gan_Matrix_f *  Z, Gan_Bool  sort, float *  work, unsigned long  work_size )

Computes eigenvalues and eigenvectors of a symmetric matrix. Parameters: A  Input matrix, overwritten on output W  Diagonal matrix of eigenvalues Z  Matrix of eigenvectors or NULL sort  Whether to sort eigenvalues and eigenvectors work  Workspace array work_size  size of workspace Returns: GAN_TRUE on success, or GAN_FALSE on failure. If Z is non-NULL it is filled with the eigenvectors of symmetric matrix A. The eigenvalues of A are written into W. If work is not NULL and LAPACK is being used, the size of the work array (work_size) must be at least 3 times the size of A. If sort is passed as GAN_TRUE, the eigenvalues are sorted into ascending order.

---