KrylovBiCGStab< T > Class Template Reference

Preconditioned BiCGStab solver acting on an AMR hierarchy through AMRMultigridKrylovOp. More...

#include <CD_KrylovBiCGStab.H>

Inheritance diagram for KrylovBiCGStab< T >:

Public Member Functions
	KrylovBiCGStab () noexcept=default
	Default constructor. Call define() before use.
	~KrylovBiCGStab () noexcept
	Destructor. Frees the persistent work vectors.
void	define (AMRMultigridKrylovOp< T > *a_op, const Vector< LevelData< T > * > &a_template) noexcept
	Allocate the persistent work vectors. Call once per regrid.
void	undefine () noexcept
	Free the persistent work vectors.
bool	isDefined () const noexcept
	Whether define() has been called and the work vectors are allocated.
bool	solve (Vector< LevelData< T > * > &a_phi, const Vector< LevelData< T > * > &a_rhs) noexcept
	Solve a_op(a_phi) = a_rhs (homogeneous operator) with preconditioned BiCGStab.

Public Attributes
Real	m_eps = 1.E-10
	Relative residual tolerance, measured against m_residualScale.
int	m_maxIter = 50
	Maximum number of iterations.
int	m_numRestarts = 5
	Maximum number of restarts on breakdown.
int	m_verbosity = -1
	Verbosity; at >= 4 a terse one-line-per-iteration residual/rate history is printed.
int	m_exitStatus = -1
	Exit status after solve(): 1 = converged, 2 = breakdown, 3 = max iterations, -1 = not run.
Real	m_residualNorm = -1.0
	Final residual norm after solve() (in the adapter inner-product norm).
int	m_iterations = 0
	Number of iterations performed by the last solve().
Real	m_residualScale = 0.0
	Norm that both the printed residual and the convergence test are measured against.

Protected Attributes
AMRMultigridKrylovOp< T > *	m_op = nullptr
	Operator/adapter (not owned).
bool	m_isDefined = false
	Whether the work vectors are allocated.
Vector< LevelData< T > * >	m_r
	Residual / s.
Vector< LevelData< T > * >	m_rTilde
	Shadow residual.
Vector< LevelData< T > * >	m_e
	Accumulated correction (added to a_phi at the end).
Vector< LevelData< T > * >	m_p
	Search direction.
Vector< LevelData< T > * >	m_pTilde
	Preconditioned search direction.
Vector< LevelData< T > * >	m_sTilde
	Preconditioned s.
Vector< LevelData< T > * >	m_t
	A*s_tilde.
Vector< LevelData< T > * >	m_v
	A*p_tilde.

Detailed Description

template<class T>
class KrylovBiCGStab< T >

Preconditioned BiCGStab solver acting on an AMR hierarchy through AMRMultigridKrylovOp.

A chombo-discharge replacement for Chombo's BiCGStabSolver on the outer Krylov path, written for scale-out: the eight work vectors are allocated once in define() and reused across solves (no per-solve allocation), and the omega numerator/denominator inner products are fused into a single mDotProduct so the adapter issues one MPI reduction instead of two. The matrix-vector product and preconditioner are the homogeneous AMR operator and multigrid cycle supplied by the adapter; solve() updates the solution in place and honours a non-zero initial guess (warm start).

Note

T is the FAB type (EBCellFAB or MFCellFAB).

Member Function Documentation

define()

template<class T >

void KrylovBiCGStab< T >::define ( AMRMultigridKrylovOp< T > *  a_op, const Vector< LevelData< T > * > &  a_template  )

noexcept

Allocate the persistent work vectors. Call once per regrid.

Parameters

[in]	a_op	The (already define()d) operator/adapter to solve against
[in]	a_template	A hierarchy whose layout the work vectors mirror (e.g. the rhs)

isDefined()

template<class T >

bool KrylovBiCGStab< T >::isDefined ( ) const

noexcept

Whether define() has been called and the work vectors are allocated.

Returns

True if defined.

solve()

template<class T >

bool KrylovBiCGStab< T >::solve ( Vector< LevelData< T > * > &  a_phi, const Vector< LevelData< T > * > &  a_rhs  )

noexcept

Solve a_op(a_phi) = a_rhs (homogeneous operator) with preconditioned BiCGStab.

Parameters

[in,out]	a_phi	Solution (in: initial guess; out: solution). Warm starts are honoured.
[in]	a_rhs	Right-hand side.

Returns

True if converged (m_exitStatus == 1).

Member Data Documentation

m_residualScale

template<class T >

Real KrylovBiCGStab< T >::m_residualScale = 0.0

Norm that both the printed residual and the convergence test are measured against.

The driver sets this to the zero-residual ||r_zero|| = ||rhs - L(0)||, which is the scale the hosting solver itself uses to decide whether a solution has converged. Measuring the exit test on that same scale is what makes m_eps mean the same thing here as it does to the caller, and is also what lets a warm start pay for itself: scaling the tolerance by this solve's own initial residual instead would demand the same reduction factor no matter how good the initial guess was. A non-positive value falls back to this solve's initial residual norm, which is the right scale when there is no host to agree with.

---

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

• Source/Elliptic/CD_KrylovBiCGStab.H
• Source/Elliptic/CD_KrylovBiCGStabImplem.H