TLDL Class Reference

computes LDL factorisation \( A = LDL^T \) or \( A = LDL^H \) More...

#include <mat_fac.hh>

Public Member Functions
void	factorise (TMatrix *A, const TTruncAcc &acc) const
auto	eval_matrix (TMatrix *A, const matform_t matform) const -> std::unique_ptr< TFacMatrix >
auto	inv_matrix (TMatrix *A, const matform_t matform) const -> std::unique_ptr< TFacInvMatrix >
void	split (const TMatrix *A, TMatrix *&L, TMatrix *&D) const
std::pair< TMatrix *, TMatrix *>	split (TMatrix *A) const
size_t	pm_steps (const TMatrix *A) const
	return number of factorisation steps for A for progress meter

Public Attributes
	__pad0__: _options( opts ) {} ~TLDL () {} void set_options ( const fac_options_t opts ) { _options = opts
	standard constructor with optional LDL settings

Detailed Description

This class computes the LDL factorisation \(A = LDL^T\) ( \(A = LDL^H\)) of a symmetric (hermitian) matrix \(A\) with lower, unit triangular matrix \(L\) and diagonal matrix \(D\).

The factorisation may be either point wise, i.e. a real LU factorisation, or block wise in which case, dense diagonal matrix blocks are inverted.

Support for multiple threads is available, although the expectable speedup is limited, e.g. best suited for at most 4 threads.

Member Function Documentation

eval_matrix()

auto eval_matrix	(	TMatrix *	A,
		const matform_t	matform
	)		const -> std::unique_ptr< TFacMatrix >

return suitable representation for evaluating factors L and U

Parameters

A	LDL factors to be represented
matform	format of original, non-factorised matrix

factorise()

void factorise	(	TMatrix *	A,
		const TTruncAcc &	acc
	)		const

compute LDL factorisation of given matrix

Parameters

A	on input matrix to factorise; on output factors L and D
acc	accuracy of factorisation

inv_matrix()

auto inv_matrix	(	TMatrix *	A,
		const matform_t	matform
	)		const -> std::unique_ptr< TFacInvMatrix >

return suitable inverse representation of factors L and U

Parameters

A	LDL factors to be represented
matform	format of original, non-factorised matrix

split() [1/2]

void split	(	const TMatrix *	A,
		TMatrix *&	L,
		TMatrix *&	D
	)		const

split given matrix A into individual factors L and D

Parameters

A	joined LDL factors
L	matrix pointer to store factor L
D	matrix pointer to store factor D
eval_type	determines block-wise or point-wise evaluation

split() [2/2]

std::pair< TMatrix *, TMatrix * > split

(

TMatrix *

A

)

const

split matrix A into individual factors L and D; A is destroyed and L/D will contain the former submatrices of A