dla.h File Reference

(2014-04-30 15:38:09 -0400, (build fd32605))

#include "defines.h"

Go to the source code of this file.

Namespaces
	af

Enumerations
enum	af_solve_t {   af_solve_none = 0, af_solve_posdef = 1, af_solve_nonposdef = 2, af_solve_gaussian = 3,   af_solve_pseudo = 4, af_solve_ctrans = 256, af_solve_trans = 512, af_solve_uppertri = 1024,   af_solve_lowertri = 2048, af_solve_tridiag = 4096, af_solve_blkdiag = 8192 }

Functions
array	lu (const array &in)
	LU factorization (packed). More...
void	lu (array &lower, array &upper, const array &in)
	LU factorization. More...
void	lu (array &lower, array &upper, array &pivot, const array &in)
	LU factorization (with pivoting). More...
array	qr (const array &in)
	QR factorization (packed). More...
void	qr (array &q, array &r, const array &in)
	QR factorization. More...
void	qr (array &q, array &r, array &tau, const array &in)
	QR factorization with tau. More...
array	cholesky (const array &X, bool is_upper=true)
	Cholesky decomposition ("Y^T * Y == X"). More...
array	cholesky (unsigned &info, const array &X, bool is_upper=true)
	Cholesky decomposition ("Y^T * Y == X"). More...
array	hessenberg (const array &in)
	Hessenberg matrix form. More...
void	hessenberg (array &h, array &q, const array &in)
	Hessenberg matrix h with unitary permutation matrix q. More...
array	eigen (const array &in, bool is_diag=false)
	Compute eigenvalues. More...
void	eigen (array &values, array &vectors, const array &in)
	Eigenvalues and eigenvectors. More...
array	svd (const array &in, bool is_diag=false)
	Singular values. More...
void	svd (array &s, array &u, array &v, const array &in)
	Singular values with unitary bases: in = u * s * v. More...
array	inverse (const array &in)
	Matrix inversion. More...
array	matpow (const array &base, double exponent)
	Matrix power. More...
unsigned	rank (const array &in, double tolerance=1e-5)
	Rank of matrix. More...
template<typename T >
T	det (const array &in)
	Matrix determinant. More...
array	solve (const array &A, const array &b, af_solve_t options=af_solve_none)
	Solve linear system. More...
Device pointer interface: lu Decomposition
See also Linear Algebra Parameters [out] d_piv Array of the pivot indices. Size: min(m, n) [out] d_U The upper triangular matrix. Size: size(min(m,n), n) [in,out] d_L If d_U isn't NULL, d_L (Size: (m, n)) contains lower triangular matrix. If d_U is NULL, d_L (First m x min(m,n) elements) contain the packed version of lu decomposition. [in] m number of rows in the input [in] n number of columns in the input [in] batch Number of tiles of input being handled.
afError	af_lu_S (int *d_piv, float *d_U, float *d_L, unsigned m, unsigned n, unsigned batch)
	lu Decomposition on single precision data. No DLA license required. More...
afError	af_lu_C (int *d_piv, cfloat *d_U, cfloat *d_L, unsigned m, unsigned n, unsigned batch)
	lu Decomposition on single precision, complex data. DLA license required. More...
afError	af_lu_D (int *d_piv, double *d_U, double *d_L, unsigned m, unsigned n, unsigned batch)
	lu Decomposition on double precision data. DLA license required. More...
afError	af_lu_Z (int *d_piv, cdouble *d_U, cdouble *d_L, unsigned m, unsigned n, unsigned batch)
	lu Decomposition on double precision, complex data. DLA license required. More...
Device pointer interface: qr decomposition
See also Linear Algebra Parameters [out] d_tau Vector on device containing scalar factors. Size: min(m, n). [out] d_R The upper triangular matrix. Size: (k, n) [in,out] d_Q If d_R isn't NULL, d_Q contains orthogonal matrix. Size: (m, k) if d_R is NULL, d_Q contains the packed version of qr decomposition. Size: (m, n) [in] m number of rows in the input [in] n number of columns in the input [in] k number of columns to be calculated in the orthogonal matrix. [in] batch Number of tiles of input being handled.
afError	af_qr_S (float *d_tau, float *d_R, float *d_Q, unsigned m, unsigned n, unsigned k, unsigned batch)
	qr decomposition on single precision data. More...
afError	af_qr_C (cfloat *d_tau, cfloat *d_R, cfloat *d_Q, unsigned m, unsigned n, unsigned k, unsigned batch)
	qr decomposition on single precision, complex data. DLA license required. More...
afError	af_qr_D (double *d_tau, double *d_R, double *d_Q, unsigned m, unsigned n, unsigned k, unsigned batch)
	qr decomposition on double precision data. DLA license required. More...
afError	af_qr_Z (cdouble *d_tau, cdouble *d_R, cdouble *d_Q, unsigned m, unsigned n, unsigned k, unsigned batch)
	qr decomposition on double precision, complex data. DLA license required. More...
Device pointer interface: Cholesky decomposition
See also Linear Algebra Parameters [out] d_R Matrix of Size: (n, n). [in] n Width of the input matrix d_A. [in] d_A Input square matrix of width n. [in] is_upper Flag to specify the format out the required output. d_R is upper triangular if is_upper is true. d_R is lower triangular if is_upper is false. [in] batch Number of tiles of input being handled.
afError	af_cholesky_S (float *d_R, unsigned *info, unsigned n, const float *d_A, bool is_upper, unsigned batch)
	Cholesky decomposition, single precision data. No DLA license required. More...
afError	af_cholesky_C (cfloat *d_R, unsigned *info, unsigned n, const cfloat *d_A, bool is_upper, unsigned batch)
	Cholesky decomposition, single precision complex data. DLA license required. More...
afError	af_cholesky_D (double *d_R, unsigned *info, unsigned n, const double *d_A, bool is_upper, unsigned batch)
	Cholesky decomposition, double precision data. DLA license required. More...
afError	af_cholesky_Z (cdouble *d_R, unsigned *info, unsigned n, const cdouble *d_A, bool is_upper, unsigned batch)
	Cholesky decomposition, double precision, complex data. DLA license required. More...
Device pointer interface: Hessenberg Matrix
See also Linear Algebra Parameters [out] d_H Hessenberg matrix of input d_A [out] d_Q Unitary matrix such that d_A = d_Q * d_H * d_Q' [in] n Width of the input matrix d_A. [in] d_A Input square matrix of width n. [in] batch Number of tiles of input being handled.
afError	af_hessenberg_S (float *d_H, float *d_Q, unsigned n, const float *d_A, unsigned batch)
	Find the hessenberg matrix on single precision data. More...
afError	af_hessenberg_C (cfloat *d_H, cfloat *d_Q, unsigned n, const cfloat *d_A, unsigned batch)
	Find the hessenberg matrix on single precision complex data. More...
afError	af_hessenberg_D (double *d_H, double *d_Q, unsigned n, const double *d_A, unsigned batch)
	Find the hessenberg matrix on double precision data. More...
afError	af_hessenberg_Z (cdouble *d_H, cdouble *d_Q, unsigned n, const cdouble *d_A, unsigned batch)
	Find the hessenberg matrix on double precision complex data. More...
Device pointer interface: Eigen vectors and Eigen values
See also Linear Algebra Parameters [out] d_Val The output containing the eigen values of d_A. For real inputs, this is allocated internally based on the resulting complexity of output and is_imag indicates complexity, e.g. real-valued hermitial matrices produce real-valued eigen values while non-hermitian produce complex-valued eigen values. [out] d_Vec The output containing the eigen vectors of d_A, or NULL if this is to be ignored (faster). For real inputs, this is allocated internally based on complexity of output and is_imag indicates complexity. For complex inputs, caller allocates. Pass NULL to avoid computing eigen vectors (faster). [out] is_imag The output format specifier for real inputs. If is_imag is true, the outputs are stored as cfloat or (cdouble). If is_imag is false, the outputs are stored as float (or double). [in] n Width of the input matrix d_A. [in] d_A Input square matrix of width n. [in] is_diag Specifier of d_Val data structure. If is_diag is true, d_Val is a diagonal matrix. if is_diag is false, d_Val is a vector. is_diag can not be false if d_Vec is not NULL [in] batch Number of tiles of input being handled.
afError	af_eigen_S (void **d_Val, void **d_Vec, bool *is_imag, unsigned n, const float *d_A, bool is_diag, unsigned batch)
	Eigen value decomposition of single precision input. More...
afError	af_eigen_D (void **d_Val, void **d_Vec, bool *is_imag, unsigned n, const double *d_A, bool is_diag, unsigned batch)
	Eigen value decomposition of single precision complex input. More...
afError	af_eigen_C (void **d_Val, cfloat *d_Vec, bool *is_imag, unsigned n, const cfloat *d_A, bool is_diag, unsigned batch)
	Eigen value decomposition of double precision input. More...
afError	af_eigen_Z (void **d_Val, cdouble *d_Vec, bool *is_imag, unsigned n, const cdouble *d_A, bool is_diag, unsigned batch)
	Eigen value decomposition of double precision complex input. More...
Device pointer interface: Singular value decomposition
See also Linear Algebra Parameters [out] d_S The output containing the singular of the input. if is_diag is false, d_S is a vector. if is_diag is true , d_S is a diagonal matrix. d_A = d_U * d_S * d_V' [out] d_U Left unitary Matrix [out] d_V Right unitary Matrix [in] jobU Can be one of 'A', 'O', 'S', 'N' similar to lapack [in] jobV Can be one of 'A', 'O', 'S', 'N' similar to lapack [in] m Number of rows in the input. [in] n Number of columns in the input [in] d_A The input matrix. [in] m_ Number of rows required in the output. [in] n_ Number of columns required in the output. [in] is_diag Data structure specifier of input d_S. is_diag can not be false if d_U or d_V are not NULL [in] batch Number of tiles of input being handled.
afError	af_svd_S (float *d_S, float *d_U, float *d_V, char jobU, char jobV, unsigned m, unsigned n, const float *d_A, unsigned m_, unsigned n_, bool is_diag, unsigned batch)
	Singular value decomposition on single precision input DLA license not required. More...
afError	af_svd_C (float *d_S, cfloat *d_U, cfloat *d_V, char jobU, char jobV, unsigned m, unsigned n, const cfloat *d_A, unsigned m_, unsigned n_, bool is_diag, unsigned batch)
	Singular value decomposition on single precision complex input DLA license required. More...
afError	af_svd_D (double *d_S, double *d_U, double *d_V, char jobU, char jobV, unsigned m, unsigned n, const double *d_A, unsigned m_, unsigned n_, bool is_diag, unsigned batch)
	Singular value decomposition on double precision input DLA license not required. More...
afError	af_svd_Z (double *d_S, cdouble *d_U, cdouble *d_V, char jobU, char jobV, unsigned m, unsigned n, const cdouble *d_A, unsigned m_, unsigned n_, bool is_diag, unsigned batch)
	Singular value decomposition on double precision complex input DLA license not required. More...
Device pointer interface: Matrix inversion
See also Linear Algebra Parameters [out] d_out The inverted matrix of d_in. [in] n Width of the input matrix. [in,out] d_A The input matrix. Inversion done in-place if d_out is NULL [in] batch Number of tiles of input being handled.
afError	af_inverse_S (float *out, unsigned n, float *d_in, unsigned batch)
	Inversion of single precision matrix. DLA license required. More...
afError	af_inverse_C (cfloat *out, unsigned n, cfloat *d_in, unsigned batch)
	Inversion of single precision complex matrix. DLA license required. More...
afError	af_inverse_D (double *out, unsigned n, double *d_in, unsigned batch)
	Inversion of double precision matrix. DLA license required. More...
afError	af_inverse_Z (cdouble *out, unsigned n, cdouble *d_in, unsigned batch)
	Inversion of double precision complex matrix. DLA license required. More...
Device pointer interface: Matrix determinant
Parameters [out] res The determinant of the input matrix d_X [in] n Width of the input matrix. [in,out] d_X The input matrix. [in] inplace Input data in d_X is destroyed if true. [in] batch Number of tiles of input being handled.
afError	af_det_S (float *res, unsigned n, float *d_X, bool inplace, unsigned batch)
	Determinant of single precision matrix. DLA license not required. More...
afError	af_det_C (cfloat *res, unsigned n, cfloat *d_X, bool inplace, unsigned batch)
	Determinant of single precision complex matrix. DLA license required. More...
afError	af_det_D (double *res, unsigned n, double *d_X, bool inplace, unsigned batch)
	Determinant of double precision matrix. DLA license required. More...
afError	af_det_Z (cdouble *res, unsigned n, cdouble *d_X, bool inplace, unsigned batch)
	Determinant of double precision complex matrix. DLA license required. More...
Device pointer interface: Matrix power
See also Linear Algebra Parameters [out] d_out d_out is the output containing pow(d_in, power) [in] n Width of the input matrix. [in] d_in The input matrix. [in] power The exponent the input has to be raised to. [in] batch Number of tiles of input being handled. [in] is_cplx To signal if the output is real or complex
afError	af_matrixPower_S (void **d_out, unsigned n, const float *d_in, float power, unsigned batch, bool *is_cplx)
	Matrix power for single precision matrix. More...
afError	af_matrixPower_C (void **d_out, unsigned n, const cfloat *d_in, float power, unsigned batch, bool *is_cplx)
	Matrix power for single precision, complex matrix. DLA license required. More...
afError	af_matrixPower_D (void **d_out, unsigned n, const double *d_in, double power, unsigned batch, bool *is_cplx)
	Matrix power for double precision matrix. DLA license required. More...
afError	af_matrixPower_Z (void **d_out, unsigned n, const cdouble *d_in, double power, unsigned batch, bool *is_cplx)
	Matrix power for double precision, complex matrix. DLA license required. More...
Device pointer interface: Solving linear systems.
See also Linear Algebra Parameters [out] d_X Solution to the equation d_A * d_X = d_B. [in] m Number of rows in d_A. [in] n Number of columns in d_A, Number of rows in d_B. [in] d_A The co-efficient matrix. [in] batch_A The number of tiles of d_A being computed. [in] k Number of columns in d_B. [in] d_B The residual matrix. [in] batch_B The number of tiles of d_B being computed. [in] opts Give specific information about the system. opts combination (sum) of: 0 for no information (solves using LU/CHOL for m == n, QR for m != n) 1 if d_A is positive definite 2 if d_A is not positive definite 3 Use Gaussian elimination (fast, cannot be combined with other options) 4 Use pseudo inverse (fast, cannot be combined with other options) 256 to do conjugate transpose on d_A before solving 512 to do transpose on d_A before solving 1024 if d_A is Upper triangular 2048 if d_A is Lower triangular
afError	af_linearSolve_SS (float *d_X, unsigned m, unsigned n, const float *d_A, unsigned batch_A, unsigned k, const float *d_B, unsigned batch_B, unsigned opts)
	Solve a single precision system. DLA not required for opts = 0, 3. More...
afError	af_linearSolve_CC (cfloat *d_X, unsigned m, unsigned n, const cfloat *d_A, unsigned batch_A, unsigned k, const cfloat *d_B, unsigned batch_B, unsigned opts)
	Solve a single precision complex system. DLA not required for opts = 3. More...
afError	af_linearSolve_DD (double *d_X, unsigned m, unsigned n, const double *d_A, unsigned batch_A, unsigned k, const double *d_B, unsigned batch_B, unsigned opts)
	Solve a double precision system. DLA not required for opts = 3. More...
afError	af_linearSolve_ZZ (cdouble *d_X, unsigned m, unsigned n, const cdouble *d_A, unsigned batch_A, unsigned k, const cdouble *d_B, unsigned batch_B, unsigned opts)
	Solve a double precision complex system. DLA not required for opts = 3. More...
Get original pivot indices
d_piv returned by af_lu_* contain indices from their updated locations. af_piv_final gives the original locations of each of the pivots. See also Linear Algebra Parameters [out] d_out Original pivot locations. [in] m Number of rows in original input. [in] k Number of columns original input. [in] d_piv Contains updated pivot indices. [in] batch The number of input tiles.
afError	af_piv_final_I (int *d_out, unsigned m, unsigned k, const int *d_piv, unsigned batch)
afError	af_piv_final_U (unsigned *d_out, unsigned m, unsigned k, const int *d_piv, unsigned batch)
afError	af_piv_final_S (float *d_out, unsigned m, unsigned k, const int *d_piv, unsigned batch)
afError	af_piv_final_D (double *d_out, unsigned m, unsigned k, const int *d_piv, unsigned batch)

Function Documentation

afError af_piv_final_I	(	int *	d_out,
		unsigned	m,
		unsigned	k,
		const int *	d_piv,
		unsigned	batch
	)

afError af_piv_final_U	(	unsigned *	d_out,
		unsigned	m,
		unsigned	k,
		const int *	d_piv,
		unsigned	batch
	)

afError af_piv_final_S	(	float *	d_out,
		unsigned	m,
		unsigned	k,
		const int *	d_piv,
		unsigned	batch
	)

afError af_piv_final_D	(	double *	d_out,
		unsigned	m,
		unsigned	k,
		const int *	d_piv,
		unsigned	batch
	)