Main Page | Modules | Class List | File List | Class Members | File Members

Threshold Elements of a General Size Matrix
[General Size Matrices]


Functions

Gan_SquMatrixgan_diagmat_diag_thres (Gan_SquMatrix *A, unsigned null_space, double thres, int *indexptr, double **elptr)
 Threshold small diagonal elements of a diagonal matrix.
Gan_SquMatrix_fgan_diagmatf_diag_thres (Gan_SquMatrix_f *A, unsigned null_space, float thres, int *indexptr, float **elptr)
 Threshold small diagonal elements of a diagonal matrix.

Function Documentation

Gan_SquMatrix* gan_diagmat_diag_thres Gan_SquMatrix A,
unsigned  null_space,
double  thres,
int *  indexptr,
double **  elptr
 

Threshold small diagonal elements of a diagonal matrix.

Parameters:
A Diagonal matrix
null_space Size of null-space in A to allow for
thres Threshold on elements of A
indexptr Pointer to index of last unaffected element of A
elptr Pointer to last unaffected element of A
Diagonal elements below the threshold thres times the first are set to zero. In addition any redundant diagonal elements (indicated by the value of null_space) are also set to zero. This can be used to eliminate conditioning problems when using the SVD algorithm gan_mat_svd() for least-squares. The diagonal elements must be in monotonically decreasing order from the top-left corner of the matrix.

The array x of state vectors specifies any redundant parameters in each block.

The contents of indexptr and elptr, if not NULL, are set respectively to the index of and a pointer to the last diagonal element not set to the high value.

Gan_SquMatrix_f* gan_diagmatf_diag_thres Gan_SquMatrix_f A,
unsigned  null_space,
float  thres,
int *  indexptr,
float **  elptr
 

Threshold small diagonal elements of a diagonal matrix.

Parameters:
A Diagonal matrix
null_space Size of null-space in A to allow for
thres Threshold on elements of A
indexptr Pointer to index of last unaffected element of A
elptr Pointer to last unaffected element of A
Diagonal elements below the threshold thres times the first are set to zero. In addition any redundant diagonal elements (indicated by the value of null_space) are also set to zero. This can be used to eliminate conditioning problems when using the SVD algorithm gan_mat_svd() for least-squares. The diagonal elements must be in monotonically decreasing order from the top-left corner of the matrix.

The array x of state vectors specifies any redundant parameters in each block.

The contents of indexptr and elptr, if not NULL, are set respectively to the index of and a pointer to the last diagonal element not set to the high value.


Generated on Fri Mar 17 12:44:58 2006 by  doxygen 1.3.9.1