her2k.h File Reference

Routines for hermitian rank-2k updates, a specialized form of Matrix*Matrix multiplication. More...

#include <myramath/utility/Number.h>

Include dependency graph for her2k.h:

Go to the source code of this file.

Classes
class	myra::Matrix< Number >
	Tabulates an IxJ matrix. Allows random access, has column major layout to be compatible with BLAS/LAPACK. More...
class	myra::MatrixRange< Number >
	Represents a mutable MatrixRange. More...
class	myra::CMatrixRange< Number >
	Represents a const MatrixRange. More...
class	myra::LowerMatrix< Number >
	Stores a lower triangular matrix in rectangular packed format. More...
class	myra::LowerMatrixRange< Number >
	Represents a mutable LowerMatrixRange. More...

Functions
void	myra::her2k_inplace (const MatrixRange< NumberS > &C, char uplo, const CMatrixRange< NumberS > &A, const CMatrixRange< NumberS > &B, char op='N', NumberS alpha=1, NumberS beta=0)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), touching only the specified triangle of C.
void	myra::her2k_inplace (const MatrixRange< NumberD > &C, char uplo, const CMatrixRange< NumberD > &A, const CMatrixRange< NumberD > &B, char op, NumberD alpha, NumberD beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), touching only the specified triangle of C.
void	myra::her2k_inplace (const MatrixRange< NumberC > &C, char uplo, const CMatrixRange< NumberC > &A, const CMatrixRange< NumberC > &B, char op, NumberC alpha, NumberS beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), touching only the specified triangle of C.
void	myra::her2k_inplace (const MatrixRange< NumberZ > &C, char uplo, const CMatrixRange< NumberZ > &A, const CMatrixRange< NumberZ > &B, char op, NumberZ alpha, NumberD beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), touching only the specified triangle of C.
void	myra::her2k_inplace (const LowerMatrixRange< NumberS > &C, const CMatrixRange< NumberS > &A, const CMatrixRange< NumberS > &B, char op='N', NumberS alpha=1, NumberS beta=0)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A))
void	myra::her2k_inplace (const LowerMatrixRange< NumberD > &C, const CMatrixRange< NumberD > &A, const CMatrixRange< NumberD > &B, char op, NumberD alpha, NumberD beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A))
void	myra::her2k_inplace (const LowerMatrixRange< NumberC > &C, const CMatrixRange< NumberC > &A, const CMatrixRange< NumberC > &B, char op, NumberC alpha, NumberS beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A))
void	myra::her2k_inplace (const LowerMatrixRange< NumberZ > &C, const CMatrixRange< NumberZ > &A, const CMatrixRange< NumberZ > &B, char op, NumberZ alpha, NumberD beta)
	Updates C = beta*C + alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A))
LowerMatrix< NumberS >	myra::her2k (const CMatrixRange< NumberS > &A, const CMatrixRange< NumberS > &B, char op='N', NumberS alpha=1)
	Returns alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), alpha is defaulted to 1.
LowerMatrix< NumberD >	myra::her2k (const CMatrixRange< NumberD > &A, const CMatrixRange< NumberD > &B, char op, NumberD alpha)
	Returns alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), alpha is defaulted to 1.
LowerMatrix< NumberC >	myra::her2k (const CMatrixRange< NumberC > &A, const CMatrixRange< NumberC > &B, char op, NumberC alpha)
	Returns alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), alpha is defaulted to 1.
LowerMatrix< NumberZ >	myra::her2k (const CMatrixRange< NumberZ > &A, const CMatrixRange< NumberZ > &B, char op, NumberZ alpha)
	Returns alpha*op(A)*hermitian(op(B)) + conj(alpha)*op(B)*hermitian(op(A)), alpha is defaulted to 1.

Detailed Description

Routines for hermitian rank-2k updates, a specialized form of Matrix*Matrix multiplication.

Four possibilies for op, each with a different effect:

'N' -> C = beta*C + alpha*A*hermitian(B) + conj(alpha)*B*hermitian(A) 'H' -> C = beta*C + alpha*hermitian(A)*B + conj(alpha)*hermitian(B)*A 'T' -> C = beta*C + alpha*transpose(A)*conjugate(B) + conj(alpha)*transpose(B)*conjugate(A) 'C' -> C = beta*C + alpha*conjugate(A)*transpose(B) + conj(alpha)*conjugate(B)*transpose(A)