AMRSolver Class Reference

Class which manages grid hierarchy and composite elliptic solution. More...

#include <AMRSolver.H>

Collaboration diagram for AMRSolver:

[legend]List of all members.

Public Member Functions
	AMRSolver ()
	Creates a AMRSolver whose internal state is undefined.
	AMRSolver (const Vector< DisjointBoxLayout > &a_gridsLevel, const Vector< Box > &a_domainLevel, const Vector< Real > &a_dxLevel, const Vector< int > &a_refRatio, int a_numLevels, int a_lBase, const LevelOp *const a_opin, int a_ncomp=1, bool a_limitCoarsening=false)
	Creates a fully-defined AMRSolver.
	AMRSolver (const Vector< DisjointBoxLayout > &a_gridsLevel, const Vector< ProblemDomain > &a_domainLevel, const Vector< Real > &a_dxLevel, const Vector< int > &a_refRatio, int a_numLevels, int a_lBase, const LevelOp *const a_opin, int a_ncomp=1, bool a_limitCoarsening=false)
	Creates a fully-defined AMRSolver.
void	define (const Vector< DisjointBoxLayout > &a_gridsLevel, const Vector< Box > &a_domainLevel, const Vector< Real > &a_dxLevel, const Vector< int > &a_refRatio, int a_numLevels, int a_lBase, const LevelOp *const a_opin, int a_ncomp=1, bool a_limitCoarsening=false)
	Defines AMRSolver object's internal state.
void	define (const Vector< DisjointBoxLayout > &a_gridsLevel, const Vector< ProblemDomain > &a_domainLevel, const Vector< Real > &a_dxLevel, const Vector< int > &a_refRatio, int a_numLevels, int a_lBase, const LevelOp *const a_opin, int a_ncomp=1, bool a_limitCoarsening=false)
	Defines AMRSolver object's internal state.
	~AMRSolver ()
	Destructor.
bool	isDefined () const
	Returns true if full define function has been called.
void	setNumSmoothUp (int a_numSmoothUp)
	Set number of multigrid smoothings on way up v-cycle. Default is 4.
void	setNumSmoothDown (int a_numSmoothDown)
	Set number of multigrid smoothings on way down v-cycle; Default is 4.
void	setTolerance (Real a_tolerance)
	Set tolerance of iterative solution. Default is 1.0e-10.
void	setOperatorTolerance (Real a_operatorTolerance)
	Set "operator tolerance" of iterative solution.
void	setMaxIter (int a_maxIter)
	Set max number of iterations. Default is 42.
void	setMinIter (int a_minIter)
	Set min number of iterations.
void	setNumVCyclesBottom (int a_numVCycleBottom)
	Set the number of V-cycles performed at the bottom of the AMR v-cycle.
void	setConvergenceMetric (Real a_metric, int a_comp)
	set the convergence metrics (if appropriate) for this problem
void	solveAMR (Vector< LevelData< FArrayBox > * > &a_phiLevel, const Vector< LevelData< FArrayBox > * > &a_rhsLevel, bool a_initializePhiToZero=true)
	Solves on hierarchy to tolerance m_tolerance.
void	AMRVCycleMG (Vector< LevelData< FArrayBox > * > &a_corrLevel, const Vector< LevelData< FArrayBox > * > &a_residLevel)
	Does one relaxation V-cycle using a MG solver.
Vector< Real >	computeResidualNorm (Vector< LevelData< FArrayBox > * > &a_phiLevel, const Vector< LevelData< FArrayBox > * > &a_rhsLevel, int a_normType)
	Calculate norm of multilevel residual on levels lBase to lmax.
Vector< Real >	computeResidualNorm (Vector< LevelData< FArrayBox > * > &a_residLevel, Vector< LevelData< FArrayBox > * > &a_phiLevel, const Vector< LevelData< FArrayBox > * > &a_rhsLevel, int a_normType)
	Calculate norm of multilevel residual on levels lBase to lmax.
void	computeAMRResidual (Vector< LevelData< FArrayBox > * > &a_phiLevel, const Vector< LevelData< FArrayBox > * > &a_rhsLevel, LevelData< FArrayBox > &a_res, int a_ilev)
	Calculate multilevel residual on level ilev.
void	applyAMROperator (Vector< LevelData< FArrayBox > * > &a_phiLevel, LevelData< FArrayBox > &a_LofPhi, int a_ilev)
	Calculate multilevel L(phi). includes refluxing and all that.
void	applyAMROperatorHphys (Vector< LevelData< FArrayBox > * > &a_phiLevel, LevelData< FArrayBox > &a_LofPhi, int a_ilev)
	Calculate multilevel L(phi) with homogeneous physical BCs.
void	setVerbose (bool a_verbose)
	set whether the solver does i/o while solving. default is true
Protected Member Functions
void	setDefaultValues ()
void	clear ()
Vector< Real >	computeNorm (const Vector< LevelData< FArrayBox > * > &a_phiVect, int a_normType)
Protected Attributes
Real	m_errorTolerance
Real	m_tolerance
Vector< Real >	m_convergenceMetrics
Real	m_operatorTolerance
int	m_numLevels
int	m_finestLevel
Vector< int >	m_refRatio
Vector< Real >	m_dxLevel
Vector< DisjointBoxLayout >	m_gridsLevel
Vector< ProblemDomain >	m_domainLevel
int	m_maxIter
int	m_minIter
int	m_numSmoothUp
int	m_numVCyclesBottom
int	m_numSmoothDown
Vector< AMRLevelMG * >	m_amrmgLevel
bool	m_isDefined
LevelSolver	m_levelSolver
int	m_lBase
int	m_ncomp
bool	m_verbose
Friends
class	AMRLevelMG

---

Detailed Description

Class which manages grid hierarchy and composite elliptic solution.

AMRSolver manages the AMR/multigrid solution to the elliptic equation on a multiple level grid that satisfies certain proper nesting conditions. It can be used either as a solver, or to apply the AMR / multigrid V-cycle as a preconditioner for some other iterative method, such as a Krylov method.

---

Constructor & Destructor Documentation

AMRSolver::AMRSolver (   )

Creates a AMRSolver whose internal state is undefined. Need to call define(...) function to use any of the functionality of the class

AMRSolver::AMRSolver (  const Vector< DisjointBoxLayout > &  a_gridsLevel, const Vector< Box > &  a_domainLevel, const Vector< Real > &  a_dxLevel, const Vector< int > &  a_refRatio, int  a_numLevels, int  a_lBase, const LevelOp *const  a_opin, int  a_ncomp = 1, bool  a_limitCoarsening = false )

Creates a fully-defined AMRSolver. Calls define function with identical arguments

AMRSolver::AMRSolver (  const Vector< DisjointBoxLayout > &  a_gridsLevel, const Vector< ProblemDomain > &  a_domainLevel, const Vector< Real > &  a_dxLevel, const Vector< int > &  a_refRatio, int  a_numLevels, int  a_lBase, const LevelOp *const  a_opin, int  a_ncomp = 1, bool  a_limitCoarsening = false )

Creates a fully-defined AMRSolver. Calls define function with identical arguments

AMRSolver::~AMRSolver (   )

Destructor.

---

Member Function Documentation

void AMRSolver::AMRVCycleMG (  Vector< LevelData< FArrayBox > * > &  a_corrLevel, const Vector< LevelData< FArrayBox > * > &  a_residLevel )

Does one relaxation V-cycle using a MG solver. Problem is assumed to already be in residual-correction form. Inputs: corrLevel - pointers to current guess at the correction values for levels lMin = max(lBase-1,0) ... lMax. Vector index corresponds to level number. residLevel - pointers to AMR residual for levels lMin ... lMax. Vector index corresponds to level number. Outputs: corrLevel - LevelData<FArrayBox>s pointed to for levels lMin,...,lMax are updated in place.

void AMRSolver::applyAMROperator (  Vector< LevelData< FArrayBox > * > &  a_phiLevel, LevelData< FArrayBox > &  a_LofPhi, int  a_ilev )

Calculate multilevel L(phi). includes refluxing and all that. Includes refluxing and all that. This is the three-level operator.

void AMRSolver::applyAMROperatorHphys (  Vector< LevelData< FArrayBox > * > &  a_phiLevel, LevelData< FArrayBox > &  a_LofPhi, int  a_ilev )

Calculate multilevel L(phi) with homogeneous physical BCs. Calculate multilevel L(phi) with homogeneous physical BCs (but with inhomogeneous C/F BCs). includes refluxing and all that This is the three-level operator.

void AMRSolver::clear (   )  [protected]

Clear internal storage

void AMRSolver::computeAMRResidual (  Vector< LevelData< FArrayBox > * > &  a_phiLevel, const Vector< LevelData< FArrayBox > * > &  a_rhsLevel, LevelData< FArrayBox > &  a_res, int  a_ilev )

Calculate multilevel residual on level ilev.

Vector<Real> AMRSolver::computeNorm (  const Vector< LevelData< FArrayBox > * > &  a_phiVect, int  a_normType )  [protected]

compute multilevel norm on levels lBase to lMax for all components in a_phiVect. Multilevel norm does not include covered regions.

Vector<Real> AMRSolver::computeResidualNorm (  Vector< LevelData< FArrayBox > * > &  a_residLevel, Vector< LevelData< FArrayBox > * > &  a_phiLevel, const Vector< LevelData< FArrayBox > * > &  a_rhsLevel, int  a_normType )

Calculate norm of multilevel residual on levels lBase to lmax. Does not include data covered by finer levels. Also returns residual in argument a_residLevel.

Vector<Real> AMRSolver::computeResidualNorm (  Vector< LevelData< FArrayBox > * > &  a_phiLevel, const Vector< LevelData< FArrayBox > * > &  a_rhsLevel, int  a_normType )

Calculate norm of multilevel residual on levels lBase to lmax. Does not include data covered by finer levels.

void AMRSolver::define (  const Vector< DisjointBoxLayout > &  a_gridsLevel, const Vector< ProblemDomain > &  a_domainLevel, const Vector< Real > &  a_dxLevel, const Vector< int > &  a_refRatio, int  a_numLevels, int  a_lBase, const LevelOp *const  a_opin, int  a_ncomp = 1, bool  a_limitCoarsening = false )

Defines AMRSolver object's internal state. Inputs: gridsLevel -- The grids at all levels. Each element in the vector is a level's worth of grids. gridsLevel[0] are grids for level 0 and so forth. Vector index corresponds to level number. domainLevel -- The domains at all levels. Each element in the vector is a level's domain. domainLevel[0] is the domain for level 0 and so forth. Vector index corresponds to level number. dxLevel -- The grid spacing at all levels. Each element in the vector is a level's $$. dxLevel[0] -- is $$ for level 0 and so forth. Vector index corresponds to level number. refRatio -- The refinement ratio between all levels. refRatio[0] is the refinement ratio between level 0 and level 1; Vector index corresponds to level number. numlevels -- The number of AMR levels in the calculation. The length of the Vector arguments has to be at least numlevels. lBase -- coarsest level on which solution is to be computed. This needs to be set at the time of definition, in order to build the bottom LevelSolver. a_opin -- The levelop to use in the solution. a_limitCoarsening: if true, only do multigrid coarsening down to next coarser AMR level (only coarsen by a_nRefCrse). If false, coarsen as far as possible. Only relevant when lBase > 0.

void AMRSolver::define (  const Vector< DisjointBoxLayout > &  a_gridsLevel, const Vector< Box > &  a_domainLevel, const Vector< Real > &  a_dxLevel, const Vector< int > &  a_refRatio, int  a_numLevels, int  a_lBase, const LevelOp *const  a_opin, int  a_ncomp = 1, bool  a_limitCoarsening = false )

Defines AMRSolver object's internal state. Inputs: gridsLevel -- The grids at all levels. Each element in the vector is a level's worth of grids. gridsLevel[0] are grids for level 0 and so forth. Vector index corresponds to level number. domainLevel -- The domains at all levels. Each element in the vector is a level's domain. domainLevel[0] is the domain for level 0 and so forth. Vector index corresponds to level number. dxLevel -- The grid spacing at all levels. Each element in the vector is a level's $$. dxLevel[0] -- is $$ for level 0 and so forth. Vector index corresponds to level number. refRatio -- The refinement ratio between all levels. refRatio[0] is the refinement ratio between level 0 and level 1; Vector index corresponds to level number. numlevels -- The number of AMR levels in the calculation. The length of the Vector arguments has to be at least numlevels. lBase -- coarsest level on which solution is to be computed. This needs to be set at the time of definition, in order to build the bottom LevelSolver. a_opin -- The levelop to use in the solution. a_limitCoarsening: if true, only do multigrid coarsening down to next coarser AMR level (only coarsen by a_nRefCrse). If false, coarsen as far as possible. Only relevant when lBase > 0.

bool AMRSolver::isDefined (   )  const

Returns true if full define function has been called.

void AMRSolver::setConvergenceMetric (  Real  a_metric, int  a_comp )

set the convergence metrics (if appropriate) for this problem sets metric against which we define convergence (for example, the norm of the RHS). Also passes a_metric through to the LevelMG (and eventually to a bottom solver.)

void AMRSolver::setDefaultValues (   )  [protected]

Set data to default values. not for external use

void AMRSolver::setMaxIter (  int  a_maxIter  )

Set max number of iterations. Default is 42.

void AMRSolver::setMinIter (  int  a_minIter  )

Set min number of iterations. Only relevant when residual is not decreasing fast enough to satisfy the "operator tolerance". Default is 4.

void AMRSolver::setNumSmoothDown (  int  a_numSmoothDown  )

Set number of multigrid smoothings on way down v-cycle; Default is 4.

void AMRSolver::setNumSmoothUp (  int  a_numSmoothUp  )

Set number of multigrid smoothings on way up v-cycle. Default is 4.

void AMRSolver::setNumVCyclesBottom (  int  a_numVCycleBottom  )

Set the number of V-cycles performed at the bottom of the AMR v-cycle. Set the number of v-cycles performed from the base level to the maximum coarsening of the base level during one overall v-cycle on the AMR hierarchy. Default is 1.

void AMRSolver::setOperatorTolerance (  Real  a_operatorTolerance  )

Set "operator tolerance" of iterative solution. Iteration will stop if (new_residual/old_residual) > 1-operatorTolerance (and at least minIter iterations have been performed). Default is 1.0e-5.

void AMRSolver::setTolerance (  Real  a_tolerance  )

Set tolerance of iterative solution. Default is 1.0e-10.

void AMRSolver::setVerbose (  bool  a_verbose  )

set whether the solver does i/o while solving. default is true

void AMRSolver::solveAMR (  Vector< LevelData< FArrayBox > * > &  a_phiLevel, const Vector< LevelData< FArrayBox > * > &  a_rhsLevel, bool  a_initializePhiToZero = true )

Solves on hierarchy to tolerance m_tolerance. Solves the elliptic equation over the hierarchy of levels lBase ... lMax where $lMax = numlevels-1$. If lBase > 0, then the data at level lBase - 1 is used to interpolate boundary conditions at boundary cells that are not adjacent to the domain boundary. Inputs: phiLevel - pointers to current guess at the solution values for levels (lMin = max(lBase-1,0)) ... lMax. Vector index corresponds to level number. rhsLevel - pointers to right-hand side for levels lmin ... lMax. Vector index corresponds to level number. Outputs: phiLevel - LevelData<FArrayBox>s pointed to for levels lMin,...,lMax are updated in place.

---

Friends And Related Function Documentation

friend class AMRLevelMG [friend]

---

Member Data Documentation

Vector<AMRLevelMG*> AMRSolver::m_amrmgLevel [protected]

Array of AMRLevelMGs containing data at each refinement level

Vector<Real> AMRSolver::m_convergenceMetrics [protected]

metrics used to judge convergence

Vector<ProblemDomain> AMRSolver::m_domainLevel [protected]

Domains.

Vector<Real> AMRSolver::m_dxLevel [protected]

Grid spacing.

Real AMRSolver::m_errorTolerance [protected]

Error tolerance for error estimation

int AMRSolver::m_finestLevel [protected]

Maximum number of working level is m_numLevels - 1

Vector<DisjointBoxLayout> AMRSolver::m_gridsLevel [protected]

Grids.

bool AMRSolver::m_isDefined [protected]

Has this solver been defined?

int AMRSolver::m_lBase [protected]

LevelSolver AMRSolver::m_levelSolver [protected]

int AMRSolver::m_maxIter [protected]

Maximum number of solver iterations

int AMRSolver::m_minIter [protected]

Minimum number of solver iterations before "hung converence" criteria is valid

int AMRSolver::m_ncomp [protected]

int AMRSolver::m_numLevels [protected]

Number of levels allowed

int AMRSolver::m_numSmoothDown [protected]

Number of smoother iterations on down cycle

int AMRSolver::m_numSmoothUp [protected]

Number of smoother iterations on up cycle

int AMRSolver::m_numVCyclesBottom [protected]

Number of VCycles for max iter of levelsolver

Real AMRSolver::m_operatorTolerance [protected]

Tolerance on maxOperator

Vector<int> AMRSolver::m_refRatio [protected]

Refinement ratio.

Real AMRSolver::m_tolerance [protected]

Solver tolerance

bool AMRSolver::m_verbose [protected]

default is true

---

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

• AMRSolver.H

---