fe::ext::BCGThreaded< MATRIX, VECTOR > Class Template Reference

solve Ax=b for x More...

#include <BCGThreaded.h>

Public Member Functions
void	solve (VECTOR &x, const MATRIX &A, const VECTOR &b)
void	setThreshold (F64 threshold)

Private Member Functions
void	solve (U32 thread)
void	solve (U32 thread, VECTOR &x, const MATRIX &A, const VECTOR &b)

Private Attributes
sp< Gang< Worker, I32 > >	m_spGang
const MATRIX *	m_A
VECTOR *	m_x
const VECTOR *	m_b
VECTOR	r
VECTOR	r_1
VECTOR	r_2
VECTOR	rb
VECTOR	rb_1
VECTOR	rb_2
VECTOR	p
VECTOR	p_1
VECTOR	pb
VECTOR	pb_1
VECTOR	temp [2]
VECTOR	Ap
F64	m_threshold
F64	m_dot_r_1
F64	m_alpha
F64	m_beta
U32	m_N
BWORD	m_break

Detailed Description

template<typename MATRIX, typename VECTOR>
class fe::ext::BCGThreaded< MATRIX, VECTOR >

solve Ax=b for x

Uses Biconjugate-Gradient. The matrix must be positive-definite, but not necessarily symmetric. For symmetric matrices, a regular Conjugate-Gradient can be about twice as fast.

The arguments are templated, so any argument types should work, given that they have the appropriate methods and operators.

---

The documentation for this class was generated from the following file:

• BCGThreaded.h