LinearSolvableIndefinite< Vector > Struct Template Reference

#include <LinearSolver.h>

Public Member Functions
virtual void	hessian (Vector &Av, const Vector &v)=0
	Linear operator A applied to the vector v.
virtual void	precondition (Vector &v)=0
	Apply precondioner to vector v. Required for CG algorithm.
virtual void	Ksqrt (Vector &v)=0
	Apply square root of preconditioner to v. Required for MINRES algorithm.
virtual double	sync (double x) const
	Override to synchronize scalars over MPI processes (if the same minimization is happening in sync over many processes)
int	solve (const Vector &b, const SolverParams &params)
int	CG (const Vector &b, const SolverParams &params)
int	MINRES (const Vector &b, const SolverParams &params)

Public Attributes
Vector	state
	Solution to the system A.state+b = 0.

Detailed Description

template<typename Vector>
struct LinearSolvableIndefinite< Vector >

Interface for symmetric (possibly indefinite) linear systems of the form Ax+b = 0. Choose between conjugate gradient or MINRES algorithms. Used in variational perturbation solver. CG implementation more memory-efficient compared to LinearSolvable. MINRES uses more memory but converges smoothly.

Template Parameters

Vector

Same requirements as the Vector for Minimizable

Member Function Documentation

CG()

template<typename Vector >

int LinearSolvableIndefinite< Vector >::CG	(	const Vector &	b,
		const SolverParams &	params
	)

Solve Ax+b=0 using linear conjugate gradients.

Returns

the number of iterations taken to achieve target tolerance

MINRES()

template<typename Vector >

int LinearSolvableIndefinite< Vector >::MINRES	(	const Vector &	b,
		const SolverParams &	params
	)

Solve Ax+b=0 using MINRES algorithm.

Returns

the number of iterations taken to achieve target tolerance

solve()

template<typename Vector >

int LinearSolvableIndefinite< Vector >::solve	(	const Vector &	b,
		const SolverParams &	params
	)

Solve the linear system Ax+b=0 using algorithm specified by params

Returns

the number of iterations taken to achieve target tolerance

---

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

• perturb/LinearSolver.h