CdrGodunov Class Reference

Godunov implementation for advection. More...

#include <CD_CdrGodunov.H>

Inheritance diagram for CdrGodunov:

Collaboration diagram for CdrGodunov:

Public Member Functions
	CdrGodunov ()
	Constructor.
virtual	~CdrGodunov ()
	Destructor (does nothing)
virtual void	parseOptions () override
	Parse class options to put object in usable state.
virtual void	parseRuntimeOptions () override
	Parse runtime options.
virtual void	allocate () override
	Allocate internal data holders in class object.
virtual Real	computeAdvectionDt () override
	Compute the largest possible advective time step (for explicit methods).
Public Member Functions inherited from CdrMultigrid
	CdrMultigrid ()
	Constructor.
virtual	~CdrMultigrid ()
	Destructor.
virtual void	registerOperators () override
	Register operator.
virtual void	preRegrid (const int a_lbase, const int a_oldFinestLevel) override
	Perform pre-regrid operations.
virtual void	computeDivJ (EBAMRCellData &a_divJ, EBAMRCellData &a_phi, const Real a_extrapDt, const bool a_conservativeOnly, const bool a_ebFlux, const bool a_domainFlux) override final
	Compute div(J) explicitly, where J = nV - D*grad(n).
virtual void	computeDivF (EBAMRCellData &a_divF, EBAMRCellData &a_phi, const Real a_extrapDt, const bool a_conservativeOnly, const bool a_ebFlux, const bool a_domainFlux) override final
	Compute div(v*phi) explicitly.
virtual void	computeDivD (EBAMRCellData &a_divD, EBAMRCellData &a_phi, const bool a_conservativeOnly, const bool a_ebFlux, const bool a_domainFlux) override final
	Compute div(D*grad(phi)) explicitly.
virtual void	advanceEuler (EBAMRCellData &a_newPhi, const EBAMRCellData &a_oldPhi, const EBAMRCellData &a_source, const Real a_dt) override final
	Implicit diffusion Euler advance with source term.
virtual void	advanceCrankNicholson (EBAMRCellData &a_newPhi, const EBAMRCellData &a_oldPhi, const EBAMRCellData &a_source, const Real a_dt) override final
	Implicit diffusion Crank-Nicholson advance with source term.
Public Member Functions inherited from CdrSolver
	CdrSolver ()
	Default constructor.
	CdrSolver (const CdrSolver &a_other)=delete
	Disallowed copy constructor.
	CdrSolver (const CdrSolver &&a_other)=delete
	Disallowed move constructor.
CdrSolver &	operator= (const CdrSolver &a_other)=delete
	Disallowed assignment operator.
CdrSolver &	operator= (const CdrSolver &&a_other)=delete
	Disallowed move assignment operator.
virtual	~CdrSolver ()
	Constructor.
void	setDefaultDomainBC ()
	This sets default boundary conditions (wall type).
void	setDomainBcType (const CdrDomainBC::DomainSide a_domainSide, const CdrDomainBC::BcType a_bcType)
	Set domain bc type on domain side.
void	setDomainBcFunction (const CdrDomainBC::DomainSide a_domainSide, const CdrDomainBC::FluxFunction &a_fluxFunction)
	Set domain bc function on particular domain side.
virtual void	advanceEuler (EBAMRCellData &a_newPhi, const EBAMRCellData &a_oldPhi, const Real a_dt)
	Implicit diffusion Euler advance without source term.
virtual void	advanceCrankNicholson (EBAMRCellData &a_newPhi, const EBAMRCellData &a_oldPhi, const Real a_dt)
	Implicit diffusion Crank-Nicholson advance without source term.
virtual void	computeDivG (EBAMRCellData &a_divG, EBAMRFluxData &a_G, const EBAMRIVData &a_ebFlux, const bool a_conservativeOnly)
	Compute div(G) where G is a general face-centered flux on face centers and EB centers. This can involve mass redistribution.
virtual void	gwnDiffusionSource (EBAMRCellData &a_noiseSource, const EBAMRCellData &a_cellPhi)
	Compute a random gaussian white noise source term.
virtual void	averageVelocityToFaces ()
	Average velocities to faces.
virtual void	deallocate ()
	Deallocate internal storage.
virtual void	setRealm (const std::string &a_realm)
	Set the realm for this solver.
virtual RefCountedPtr< CdrSpecies > &	getSpecies () noexcept
	Get the CDR species.
virtual const RefCountedPtr< CdrSpecies > &	getSpecies () const noexcept
	Get the CDR species. Const version.
virtual void	setSpecies (const RefCountedPtr< CdrSpecies > &a_species)
	Set species.
virtual void	setComputationalGeometry (const RefCountedPtr< ComputationalGeometry > &a_computationalGeometry)
	Set computational geometry.
virtual void	setAmr (const RefCountedPtr< AmrMesh > &a_amr)
	Set the amr object.
virtual void	setPhase (phase::which_phase a_phase)
	Set phase.
virtual void	setVerbosity (const int a_verbosity)
	Set verbosity.
virtual void	setTime (const int a_step, const Real a_time, const Real a_dt)
	Set the time for this solver.
virtual void	setVelocity (const EBAMRCellData &a_velocity)
	Set velocity from data holder.
virtual void	setVelocity (const RealVect &a_velocity)
	Set constant velocity.
virtual void	setVelocity (const std::function< RealVect(const RealVect a_pos)> &a_velocity)
	Set velocity using a polymorphic function.
virtual void	setDiffusionCoefficient (const EBAMRFluxData &a_diffusionCoefficient, const EBAMRIVData &a_ebDiffusionCoefficient)
	Data-based version of setting diffusion coefficients (which are stored on faces)
virtual void	setDiffusionCoefficient (const Real a_diffusionCoefficient)
	Set diffusion coefficient to be constant everywhere (they are stored on faces)
virtual void	setDiffusionCoefficient (const std::function< Real(const RealVect a_position)> &a_diffCo)
	Polymorphic way of setting diffusion coefficients every.
virtual void	setSource (const EBAMRCellData &a_source)
	Data based version of setting source terms.
virtual void	setSource (const Real a_source)
	Set constant source terms everywhere.
virtual void	setSource (const std::function< Real(const RealVect a_position)> &a_source)
	Polymorphic way of setting source terms.
virtual void	setEbFlux (const EBAMRIVData &a_ebFlux)
	Data-based version of setting the EB flux.
virtual void	setEbFlux (const Real a_ebFlux)
	Set the eb flux to a constant.
virtual void	initialData ()
	Fill m_phi state with initial data from m_species.
virtual void	writePlotFile ()
	Write plot file.
virtual void	writePlotData (LevelData< EBCellFAB > &a_output, int &a_icomp, const std::string &a_outputRealm, const int a_level) const noexcept
	Write output data to a_output.
virtual void	regrid (const int a_lmin, const int a_oldFinestLevel, const int a_newFinestLevel)
	Regrid this solver.
virtual std::string	getRealm () const
	Get the realm where this solver is registered.
virtual std::string	getName () const
	Get solver name.
virtual Vector< std::string >	getPlotVariableNames () const
	Get output plot names.
virtual int	getNumberOfPlotVariables () const
	Get number of output fields.
virtual Real	computeDiffusionDt ()
	Compute the largest possible diffusive time step (for explicit methods)
virtual Real	computeAdvectionDiffusionDt ()
	Compute the largest possible diffusive time step (for explicit methods).
virtual Real	computeSourceDt (const Real a_max, const Real a_tolerance)
	Compute the largest possible source time step (for explicit methods.
virtual void	weightedUpwind (EBAMRCellData &a_weightedUpwindPhi, const int a_pow)
	Compute an upwind-weighted version of phi.
virtual Real	computeMass ()
	Compute the "physical mass" in m_phi.
virtual Real	computeMass (const EBAMRCellData &a_phi, const bool a_kappaScale=true)
	Compute the "physical mass" in the input argument.
virtual Real	computeCharge ()
	Compute the total charge in m_phi.
virtual bool	isDiffusive ()
	Return true if the solver is diffusive and false otherwise.
virtual bool	isMobile ()
	Return true if the solver is mobile and false otherwise.
virtual EBAMRCellData &	getPhi ()
	Get the cell-centered phi.
virtual EBAMRCellData &	getSource ()
	Get the source term.
virtual EBAMRCellData &	getCellCenteredVelocity ()
	Get the cell-centered velocity.
virtual EBAMRFluxData &	getFaceCenteredVelocity ()
	Get the face-centered velocities.
virtual EBAMRIVData &	getEbCenteredVelocity ()
	Get the eb-centered velocities.
virtual EBAMRCellData &	getCellCenteredDiffusionCoefficient ()
	Get the cell-centered diffusion coefficient.
virtual EBAMRFluxData &	getFaceCenteredDiffusionCoefficient ()
	Get the face-centered diffusion coefficient.
virtual EBAMRIVData &	getEbCenteredDiffusionCoefficient ()
	Get the EB-centered diffusion coefficient.
virtual EBAMRIVData &	getEbFlux ()
	Get the eb flux data holder.
virtual EBAMRIFData &	getDomainFlux ()
	Get the domain flux data holder.
virtual void	extrapolateAdvectiveFluxToEB () noexcept
	Extrapolate advective flux to EB.
virtual void	extrapolateAdvectiveFluxToEB (EBAMRIVData &a_ebFlux) const noexcept
	Extrapolate advective flux to EB.
virtual void	redistribute (EBAMRCellData &a_phi, const EBAMRIVData &a_delta) const noexcept
	Add data through redistribution into cell-centered holders.

Protected Member Functions
virtual void	parseSlopeLimiter ()
	Parses slope limiter options.
virtual void	parseExtrapolateSourceTerm ()
virtual void	advectToFaces (EBAMRFluxData &a_facePhi, const EBAMRCellData &a_cellPhi, const Real a_extrapDt) override
	Godunov face extrapolation method for advection.
Protected Member Functions inherited from CdrMultigrid
virtual void	setupDiffusionSolver ()
	Set up diffusion solver.
virtual void	setupHelmholtzFactory ()
	Setup the operator factory.
virtual void	setupMultigrid ()
	Setup multigrid.
virtual void	setMultigridSolverCoefficients ()
	Set the multigrid solver coefficients.
virtual void	resetAlphaAndBeta (const Real a_alpha, const Real a_beta)
	Reset alpha and beta-coefficients in the multigrid solvers.
virtual void	computeKappaLphi (EBAMRCellData &a_kappaLphi, const EBAMRCellData &a_phi)
	Compute kappa * L(phi) using the multigrid operator. This is different from computeDivD.
virtual void	parseMultigridSettings ()
	Parse solver settings for geometric multigrid.
Protected Member Functions inherited from CdrSolver
virtual void	averageVelocityToFaces (EBAMRFluxData &a_faceVelocity, const EBAMRCellData &a_cellVelocity)
	Average cell-centered velocities to face centers.
virtual void	computeAdvectionFlux (EBAMRFluxData &a_flux, const EBAMRFluxData &a_facePhi, const EBAMRFluxData &a_faceVelocity, const bool a_addDomainFlux=true)
	Set up face-centered advection flux.
virtual void	computeAdvectionFlux (LevelData< EBFluxFAB > &a_flux, const LevelData< EBFluxFAB > &a_facePhi, const LevelData< EBFluxFAB > &a_faceVelocity, const int a_lvl)
	Set up face-centered advection flux on a grid level.
virtual void	computeDiffusionFlux (EBAMRFluxData &a_flux, const EBAMRCellData &a_phi, const bool a_addDomainFlux)
	Compute the face-centered diffusion flux.
virtual void	computeDiffusionFlux (LevelData< EBFluxFAB > &a_flux, const LevelData< EBCellFAB > &a_phi, const int a_lvl)
	Compute the face-centered diffusion flux.
virtual void	computeAdvectionDiffusionFlux (EBAMRFluxData &a_flux, const EBAMRCellData &a_cellStates, const EBAMRFluxData &a_faceStates, const EBAMRFluxData &a_faceVelocities, const EBAMRFluxData &a_faceDiffCo, const bool a_addDomainFlux)
	Compute the full advection-diffusion flux. This assumes that the solver is mobile and diffusive.
virtual void	resetDomainFlux (EBAMRFluxData &a_flux)
	Set flux to zero on domain boundaries.
virtual void	fillDomainFlux (EBAMRFluxData &a_flux)
	Set domain in data holder. This sets the flux on the boundary to either zero or to m_domainFlux.
virtual void	fillDomainFlux (LevelData< EBFluxFAB > &a_flux, const int a_level)
	Set domain in data holder. This sets the flux on the boundary to either zero or to m_domainFlux.
virtual void	conservativeDivergenceNoKappaDivision (EBAMRCellData &a_conservativeDivergence, EBAMRFluxData &a_flux, const EBAMRIVData &a_ebFlux)
	Compute conservative divergence from fluxes.
virtual void	nonConservativeDivergence (EBAMRIVData &a_nonConservativeDivergence, const EBAMRCellData &a_divG)
	Compute the non-conservative divergence.
virtual void	hybridDivergence (EBAMRCellData &a_hybridDivergence, EBAMRIVData &a_massDifference, const EBAMRIVData &a_nonConservativeDivergence)
	Use the non-conservative divergence to make the conservative divergence hold the hybrid divergence.
virtual void	hybridDivergence (LevelData< EBCellFAB > &a_hybridDivergence, LevelData< BaseIVFAB< Real > > &a_massDifference, const LevelData< BaseIVFAB< Real > > &a_nonConservativeDivergence, const int a_lvl)
	Make the hybrid divergence. On the way in, a_hybridDivergence must hold the conservative divergence.
virtual void	conservativeDivergenceRegular (LevelData< EBCellFAB > &a_divJ, const LevelData< EBFluxFAB > &a_flux, const int a_lvl)
	Compute the conservative divergence over regular cells.
virtual void	interpolateFluxToFaceCentroids (EBAMRFluxData &a_flux)
	Interpolate flux to centroids.
virtual void	interpolateFluxToFaceCentroids (LevelData< EBFluxFAB > &a_flux, const int a_lvl)
	Interpolate flux to centroids.
virtual void	computeDivergenceIrregular (LevelData< EBCellFAB > &a_divG, const LevelData< EBFluxFAB > &a_centroidFluxes, const LevelData< BaseIVFAB< Real > > &a_ebFlux, const int a_lvl)
	Compute conservative divergence on irregular cells (not kappa divided)
virtual void	initialDataDistribution ()
	Fill initial data from a distribution function.
virtual void	initialDataParticles ()
	Fill initial data from particles.
virtual void	defineInterpolationStencils ()
	Define stencils for doing face-centered to face-centroid-centered states.
virtual void	parseDomainBc ()
	Parses domain BC options.
virtual void	parsePlotVariables ()
	Parses plot variables.
virtual void	parsePlotMode ()
	Parse plot mode.
virtual void	parseDivergenceComputation ()
	Parse the conservation.
virtual void	parseRegridSlopes ()
	Parse slope regrid.
virtual std::string	makeBcString (const int a_dir, const Side::LoHiSide a_side) const
	Shortcut for making a boundary condition string.
virtual void	fillGwn (EBAMRFluxData &a_noise, const Real a_sigma)
	Gaussian noise field.
virtual void	smoothHeavisideFaces (EBAMRFluxData &a_facePhi, const EBAMRCellData &a_cellPhi)
	Use Heaviside smoothing for computing face-centered states.
virtual void	writeData (LevelData< EBCellFAB > &a_output, int &a_comp, const EBAMRCellData &a_data, const std::string &a_outputRealm, const int a_level, const bool a_interpToCentroids, const bool a_interpGhost) const noexcept
	Write data to output. Convenience function.

Protected Attributes
Vector< RefCountedPtr< EBAdvectLevelIntegrator > >	m_levelAdvect
	Advection object – this implements the Graves/Trebotich discretization.
bool	m_limitSlopes
	If true, slopes are limited (always use limiting).
bool	m_extrapolateSourceTerm
	Turn on/off source terms when time-extrapolating.
Protected Attributes inherited from CdrMultigrid
EBHelmholtzOp::Smoother	m_smoother
	Relaxation type for gmg.
int	m_chebyOrder
	Chebyshev polynomial degree (only used when m_smoother is Smoother::Chebyshev)
Real	m_chebyEigRatio
	Chebyshev eigenvalue ratio lambda_max/lambda_min (only used when m_smoother is Smoother::Chebyshev)
int	m_rasInnerSweeps
	Number of inner frozen-ghost sweeps per restricted-additive-Schwarz iteration (only used when m_smoother is Smoother::RestrictedAdditiveSchwarz)
MultigridType	m_multigridType
	GMG multigrid type.
RefCountedPtr< AMRMultiGrid< LevelData< EBCellFAB > > >	m_multigridSolver
	Geometric multigrid solver.
RefCountedPtr< EBHelmholtzOpFactory >	m_helmholtzOpFactory
	Operator factory.
BiCGStabSolver< LevelData< EBCellFAB > >	m_bicgstab
	Conjugate gradient solver bottom MG level.
EBSimpleSolver	m_simpleSolver
	Simple solver.
GMRESSolver< LevelData< EBCellFAB > >	m_gmres
	GMRES solver.
EllipticSolverChain::Settings	m_krylovSettings
	Outer-Krylov settings (selects gmg/gmres/bicgstab and the krylov_* parameters)
EllipticSolverChain::FallbackPolicy	m_fallbackPolicy
	Fallback-chain latch policy (persists across regrids; re-probes GMG every krylov_retry_interval solves)
EllipticSolverChain::KrylovSolver< EBCellFAB >	m_krylov
	Outer-Krylov driver (V-cycle-preconditioned BiCGStab/GMRES with persistent work vectors)
bool	m_hasMultigridSolver
	Is multigrid solver defined or not?
bool	m_multigridRefluxFree
	Use reflux-free AMR operator in GMG.
EBAMRCellData	m_helmAcoef
	Storage for Helmholtz a-coefficient. Always 1.
EBAMRCellData	m_residual
	Storage for multigrid residual.
int	m_multigridVerbosity
	Verbosity for geometric multigrid.
int	m_multigridPreSmooth
	Number of smoothings before averaging.
int	m_multigridPostSmooth
	Number of smoothings before averaging.
int	m_multigridBottomSmooth
	Number of smoothing before bottom solver.
int	m_multigridMaxIterations
	Maximum number of iterations.
int	m_multigridMinIterations
	Minimum number of iterations.
BottomSolverType	m_bottomSolverType
	Bottom solver type.
int	m_numSmoothingsForSimpleSolver
	Number of smoothing for bottom solver.
int	m_minCellsBottom
	Set bottom drop depth.
Real	m_multigridExitTolerance
	Multigrid tolerance.
Real	m_multigridExitHang
	Multigrid hand.
Protected Attributes inherited from CdrSolver
RefCountedPtr< CdrSpecies >	m_species
	Species through which e.g. mobility/diffusion and initial conditions is passed.
RefCountedPtr< ComputationalGeometry >	m_computationalGeometry
	Computational geometry.
RefCountedPtr< AmrMesh >	m_amr
	AMR; needed for grid stuff.
Vector< RefCountedPtr< LayoutData< BaseIFFAB< FaceStencil > > > >	m_interpStencils [SpaceDim]
	Stencils for interpolating face-centered fluxes to face centroids.
phase::which_phase	m_phase
	Phase.
std::string	m_name
	Solver name.
std::string	m_className
	Class name.
std::string	m_realm
	Realm where this solver is registered.
CdrDomainBC	m_domainBC
	Domain BCs.
EBAMRCellData	m_phi
	Cell-centered data (i.e. the advected-diffused quantity)
EBAMRCellData	m_source
	Source term.
EBAMRCellData	m_cellVelocity
	Cell-centered velocities.
EBAMRFluxData	m_faceStates
	Holder for face centered states.
EBAMRFluxData	m_faceVelocity
	Face-centered velocities (only normal components)
EBAMRIVData	m_nonConservativeDivG
	Scratch storage for the non-conservative divergence.
EBAMRIVData	m_massDifference
	Scratch storage for the mass difference.
EBAMRIVData	m_ebZero
	Scratch storage for the EB flux.
EBAMRCellData	m_cachePhi
	Cached state vector for regrid.
EBAMRCellData	m_cacheSource
	Cached source term for regrids.
EBAMRIVData	m_ebVelocity
	EB-centered velocities.
EBAMRIVData	m_ebFlux
	Flux through the embedded boundary.
EBAMRIFData	m_domainFlux
	Domain flux.
EBAMRCellData	m_cellCenteredDiffusionCoefficient
	Diffusion coefficients on cell centers.
EBAMRFluxData	m_faceCenteredDiffusionCoefficient
	Diffusion coefficients on face centers.
EBAMRIVData	m_ebCenteredDiffusionCoefficient
	Diffusion coefficients on EB faces.
std::map< CdrDomainBC::DomainSide, CdrDomainBC::FluxFunction >	m_domainFluxFunctions
	Domain flux functions.
int	m_verbosity
	Solver verbosity.
int	m_timeStep
	Time step.
Real	m_time
	Current time.
Real	m_dt
	Last time step increment.
Redistribution	m_whichRedistribution
	Which type of redistribution to use.
bool	m_blendConservation
	Flag for blending the hybrid divergence.
bool	m_isDiffusive
	Is the solver diffusive or not.
bool	m_isMobile
	Solve for advection/convection or not.
bool	m_plotPhi
	If true, m_phi is added to plot files.
bool	m_plotVelocity
	Output velocities.
bool	m_plotDiffusionCoefficient
	Output diffusion coefficients.
bool	m_plotEbFlux
	Output EB fluxes.
bool	m_plotSource
	Output source term.
bool	m_plotNumbers
	Plot numbers or densities.
bool	m_regridSlopes
	Use slopes when regridding.
int	m_seed
	RNG seed.

Additional Inherited Members
Protected Types inherited from CdrMultigrid
enum class	BottomSolverType { Simple , BiCGStab , GMRES }
	Enum class for supported bottom solvers in multigrid.
enum class	MultigridType { VCycle , WCycle }
	Enum for multigrid cycle types.
Protected Types inherited from CdrSolver
enum class	Redistribution { VolumeWeighted , None }
	Redistribution method.
Static Protected Attributes inherited from CdrSolver
static constexpr int	m_comp = 0
	Component number in data holder.
static constexpr int	m_nComp = 1
	Number of components that this solver solves for.

Detailed Description

Godunov implementation for advection.

The CdrGodunov class perform Godunov-type advection with time-extrapolated slope-limiting to faces. Note that this class implements CdrMultigrid which is the base class that implements second order diffusion (either explicit or implicit). The rationale for doing time-extrapolation ala Godunov is that one can use centered differencing phi^(k+1) = phi^k - dt*div[v^(k+1/2) * phi^(k+1/2)] for a second order accurate discretization in time. This class uses Chombo functionality from EBAdvectLevelIntegrator which uses the Graves/Trebotich discretization in "An adaptive finite volume method for the incompressible Navier–Stokes equations in complex geometries" (DOI: 10.2140/camcos.2015.10.43). Note that if source terms are involved the extrapolation is more involved.

Member Function Documentation

advectToFaces()

void CdrGodunov::advectToFaces ( EBAMRFluxData &  a_facePhi, const EBAMRCellData &  a_cellPhi, const Real  a_extrapDt  )

overrideprotectedvirtual

Godunov face extrapolation method for advection.

Parameters

[out]	a_facePhi	Phi on face centers.
[in]	a_cellPhi	Phi on cell centers.
[in]	a_extrapDt	Time centering (i.e. extrapolation) of the face-centered states.

This functions calls Chombo code which implements the Graves/Trebotich discretization.

Implements CdrMultigrid.

allocate()

void CdrGodunov::allocate ( )

overridevirtual

Allocate internal data holders in class object.

Reimplemented from CdrMultigrid.

computeAdvectionDt()

Real CdrGodunov::computeAdvectionDt ( )

overridevirtual

Compute the largest possible advective time step (for explicit methods).

This computes dt = dx/max(|vx|,|vy|,|vz|), minimized over all grid levels and patches.

Note

This is the appropriate time step routine for the BCG reconstruction.

Returns

Maximum stable advection time step.

Reimplemented from CdrSolver.

parseExtrapolateSourceTerm()

void CdrGodunov::parseExtrapolateSourceTerm ( )

protectedvirtual

®brief Parse whether or not to include a source-term in the Godunov extrapolation.

parseOptions()

void CdrGodunov::parseOptions ( )

overridevirtual

Parse class options to put object in usable state.

Implements CdrMultigrid.

parseRuntimeOptions()

void CdrGodunov::parseRuntimeOptions ( )

overridevirtual

Parse runtime options.

Implements CdrMultigrid.

---

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

• Source/ConvectionDiffusionReaction/CD_CdrGodunov.H
• Source/ConvectionDiffusionReaction/CD_CdrGodunov.cpp