AMR driver that deposits/interpolates ParticleContainer particles across the hierarchy. More...

#include <CD_EBAMRParticleMesh.H>

Public Member Functions
Construction and lifetime
	EBAMRParticleMesh ()
	Default constructor. Leaves the object undefined; call define().

	~EBAMRParticleMesh ()=default
	Destructor.

	EBAMRParticleMesh (const EBAMRParticleMesh &)=delete
	Copy construction is deleted (shared RefCountedPtr state).

EBAMRParticleMesh &	operator= (const EBAMRParticleMesh &)=delete
	Copy assignment is deleted.

void	define (const Vector< RefCountedPtr< EBLevelGrid > > &a_eblgs, const Vector< int > &a_refRat, const Vector< Real > &a_dx, const RealVect &a_probLo, const int a_ghost, const int a_finestLevel)
	Define the per-level leaves, coarse-fine mesh operators, and halo/transition masks.

Interpolation
template<auto... Members, typename P , typename Traits >
void	interpolate (ParticleContainer< P, Traits > &a_particles, const EBAMRCellData &a_meshData, const DepositionType a_interpType, const bool a_forceIrregNGP) const
	Interpolate a mesh field onto one or more payload columns, across all levels.

template<typename P , typename Traits >
void	interpolateWeight (ParticleContainer< P, Traits > &a_particles, const EBAMRCellData &a_meshData, const DepositionType a_interpType, const bool a_forceIrregNGP) const
	Interpolate a mesh field onto the container-owned weight column, across all levels.

Deposition (all coarse-fine strategies)
template<typename P , typename Traits >
void	depositWeight (EBAMRCellData &a_meshData, const ParticleContainer< P, Traits > &a_particles, const DepositionType a_depositionType, const CoarseFineDeposition a_coarseFineDeposition, const IrregularDeposition a_irregularDeposition)
	Deposit the container-owned weight onto the AMR mesh.

template<typename P , typename Traits , typename CellGather >
void	depositGathered (EBAMRCellData &a_meshData, const ParticleContainer< P, Traits > &a_particles, const DepositionType a_depositionType, const CoarseFineDeposition a_coarseFineDeposition, const IrregularDeposition a_irregularDeposition, CellGather a_cellGather)
	Deposit a custom per-particle scalar (computed by a_cellGather) onto the AMR mesh.

template<auto... Members, typename P , typename Traits >
void	deposit (EBAMRCellData &a_meshData, const ParticleContainer< P, Traits > &a_particles, const DepositionType a_depositionType, const CoarseFineDeposition a_coarseFineDeposition, const IrregularDeposition a_irregularDeposition)
	Deposit one or more payload columns onto the AMR mesh.

Protected Member Functions
void	errorUnsupported () const
	Abort on an unrecognized coarse-fine deposition strategy.

void	defineEBParticleMesh ()
	Build the per-level per-patch EBParticleMesh leaves (valid grids + refined-coarse grids).

void	defineLevelMotion ()
	Build the per-level valid+ghost -> valid Copiers.

void	defineCoarseFineMotion ()
	Build the per-level coarse-fine mesh operators (null on level 0).

template<typename Container , typename PatchDeposit >
void	depositInterpCore (EBAMRCellData &a_meshData, const Container &a_particles, const int a_numComp, PatchDeposit a_patchDeposit) const
	The CoarseFineDeposition::Interp deposition algorithm, parameterized by the per-patch deposit.

void	defineOuterHaloMasks ()
	Build the per-width outer-halo masks: per level, the coarse cells just outside the fine patch.

template<int NCOMP, typename Container , typename PatchDeposit >
void	depositHaloCore (EBAMRCellData &a_meshData, const Container &a_particles, PatchDeposit a_patchDeposit) const
	The CoarseFineDeposition::Halo deposition algorithm.

template<int NCOMP, typename Container , typename PatchDeposit , typename PatchDepositNGP >
void	depositHaloNGPCore (EBAMRCellData &a_meshData, const Container &a_particles, PatchDeposit a_patch, PatchDepositNGP a_patchNGP) const
	The CoarseFineDeposition::HaloNGP deposition algorithm.

int	getTransitionMaskWidth (const DepositionType a_depositionType, const int a_refRat) const
	Transition-mask width for a kernel: CIC -> refRat/2, TSC -> refRat (NGP is handled upstream).

void	defineTransitionMasks ()
	Build the per-width transition masks: the fine-side transition band, on the refined-coarse grid.

template<typename Container >
void	transferMaskParticlesTransition (Container &a_particles, const DepositionType a_depositionType) const
	Move the coarse-side transition particles into the container mask holder (using the FiCo mask).

template<int NCOMP, typename Container , typename PatchDeposit >
void	depositTransitionCore (EBAMRCellData &a_meshData, const Container &a_particles, const DepositionType a_depositionType, PatchDeposit a_patch) const
	The CoarseFineDeposition::Transition deposition algorithm.

Protected Attributes
Vector< RefCountedPtr< EBLevelGrid > >	m_eblgs
	Per-level EB level grids.

Vector< int >	m_refRat
	Per-level refinement ratios.

Vector< Real >	m_dx
	Per-level grid spacing (isotropic).

RealVect	m_probLo
	Lower-left corner of the physical domain.

int	m_ghost
	Number of ghost cells in the mesh data holders.

int	m_finestLevel
	Finest AMR level index.

bool	m_isDefined
	Whether define() has been called.

Vector< RefCountedPtr< LayoutData< EBParticleMesh > > >	m_ebParticleMesh
	Per-level per-patch deposition/interpolation leaves.

Vector< Copier >	m_levelCopiers
	Per-level valid+ghost -> valid Copiers (fold ghost-deposited mass into the owner patch).

Vector< RefCountedPtr< EBCoarseFineParticleMesh > >	m_coarseFinePM
	Per-level coarse-fine mesh operators (null on level 0).

Vector< RefCountedPtr< LayoutData< EBParticleMesh > > >	m_ebParticleMeshFiCo
	Per-level per-patch leaves on the refined-coarse (FiCo) grid (null on level 0).

std::map< int, Vector< RefCountedPtr< LevelData< BaseFab< bool > > > > >	m_outerHaloMasks
	Outer-halo masks keyed by halo width: per-level coarse cells just outside the fine patch.

std::map< int, Vector< RefCountedPtr< LevelData< BaseFab< bool > > > > >	m_transitionMasks
	Transition masks keyed by width: per-level fine-side transition band, on the refined-coarse grid.

Detailed Description

AMR driver that deposits/interpolates ParticleContainer particles across the hierarchy.

EBAMRParticleMesh deposits the particles of a ParticleContainer onto AMR mesh data and interpolates mesh data back onto particle columns, by driving the per-patch EBParticleMesh leaf on every box of every level. define() builds the per-level leaves (on the valid grids and the refined-coarse grids), the level valid+ghost -> valid Copiers, the EBCoarseFineParticleMesh operators, and the halo/transition masks; it must be called before any deposit or interpolate.

Interpolation

interpolate() and interpolateWeight() are purely per-level/per-patch: the mesh is assumed to be valid+ghosted by the caller, so no coarse-fine coupling is needed.

Deposition

deposit(), depositWeight(), and depositGathered() conserve mass across refinement boundaries, and select how the coarse-fine overlap is handled through CoarseFineDeposition:

Interp – move the over-boundary mass between levels with the coarse-fine mesh operators.
Halo – re-deposit the coarse-side halo particles on the refined-coarse grid (via the container's mask holder and the FiCo buffer) and add them to the fine level.
HaloNGP – deposit the near-boundary halo particles with NGP so their clouds do not spread across the boundary (no refined-coarse buffer is needed).
Transition – deposit the coarse-side transition particles on the refined-coarse grid at the fine-grid width, distributing the result to both levels. Halo, HaloNGP, and Transition require at least one ghost cell.

Cut cells, and what this class does NOT do

The deposition routines take an IrregularDeposition, but this class only implements the part of it that the leaf kernels can answer: NGP in a cut cell, yes or no. Everything else is the CALLER's responsibility.

Native, NGP – handled here. These are the two kernel behaviours.
Mirror – NOT handled here. It is a second pass that reflects each band particle across the embedded boundary and deposits the image, so it needs an image container, a scratch field and a collective remap, none of which this class owns. See ItoSolver::mirrorPass.
Redistribute, RedistributeBlended – NOT handled here. They act on the field AFTER the deposit. See ItoSolver::redistributeAMR and McPhoto::depositHybridDivergence.

The consequence is worth stating plainly, because it is silent: passing Mirror or Redistribute to a deposition routine here and then omitting the caller's half produces Native behaviour, with no error and no diagnostic. The run completes and the output is bit-for-bit what Native would have given. A caller that selects one of those values owns the rest of its meaning.

This split is deliberate but not ideal – the enum is named in one place and interpreted in three, so there is no single source of truth for what a selector means. Centralizing the mirror and the redistribution here is the natural end state; it needs ParticleContainer::define's minimum block size, level tiles, valid cells and realm name, which this class is not currently given.

Member Function Documentation

◆ define()

void EBAMRParticleMesh::define	(	const Vector< RefCountedPtr< EBLevelGrid > > &	a_eblgs,
		const Vector< int > &	a_refRat,
		const Vector< Real > &	a_dx,
		const RealVect &	a_probLo,
		const int	a_ghost,
		const int	a_finestLevel
	)

inline

Define the per-level leaves, coarse-fine mesh operators, and halo/transition masks.

Parameters

[in]	a_eblgs	Per-level EB level grids.
[in]	a_refRat	Per-level refinement ratio (entry l is the ratio between level l and l+1).
[in]	a_dx	Per-level (isotropic) grid spacing.
[in]	a_probLo	Lower-left corner of the physical domain.
[in]	a_ghost	Number of ghost cells in the mesh data holders.
[in]	a_finestLevel	Finest AMR level index.

◆ defineOuterHaloMasks()

void EBAMRParticleMesh::defineOuterHaloMasks ( )

inlineprotected

Build the per-width outer-halo masks: per level, the coarse cells just outside the fine patch.

For each ghost width in [1, m_ghost] and each level with a finer level, marks the coarse cells within that width of the refinement boundary (on the coarse side), excluding the coarsened-fine valid region and its neighbours. The mask drives copyMaskParticles in the Halo deposit. Storage-agnostic (operates on the grids only). Empty BaseFabs are undefined.

◆ defineTransitionMasks()

void EBAMRParticleMesh::defineTransitionMasks ( )

inlineprotected

Build the per-width transition masks: the fine-side transition band, on the refined-coarse grid.

For each width in [1, m_ghost] and each level with a finer level, builds the fine-grid coarse-fine interface band (fine ghost cells just outside the fine boxes), transfers it to the refined-coarse (FiCo) grid of the finer level, and stores it as a boolean mask there. Storage- agnostic. Empty BaseFabs are undefined.

◆ deposit()

template<auto... Members, typename P , typename Traits >

void EBAMRParticleMesh::deposit	(	EBAMRCellData &	a_meshData,
		const ParticleContainer< P, Traits > &	a_particles,
		const DepositionType	a_depositionType,
		const CoarseFineDeposition	a_coarseFineDeposition,
		const IrregularDeposition	a_irregularDeposition
	)

inline

Deposit one or more payload columns onto the AMR mesh.

See depositWeight for the coarse-fine dispatch.

Template Parameters

Members	Pack of payload data-member pointers selecting the column(s) to deposit.
P	Payload type (deduced).
Traits	Column descriptor (deduced).

Parameters

[in,out]	a_meshData	AMR mesh data (sizeof...(Members) components; RESET then filled).
[in]	a_particles	Particles to deposit.
[in]	a_depositionType	Deposition kernel (NGP/CIC/TSC).
[in]	a_coarseFineDeposition	Coarse-fine handling (Interp / Halo / HaloNGP / Transition).
[in]	a_irregularDeposition	Cut-cell deposition strategy.

◆ depositGathered()

template<typename P , typename Traits , typename CellGather >

void EBAMRParticleMesh::depositGathered	(	EBAMRCellData &	a_meshData,
		const ParticleContainer< P, Traits > &	a_particles,
		const DepositionType	a_depositionType,
		const CoarseFineDeposition	a_coarseFineDeposition,
		const IrregularDeposition	a_irregularDeposition,
		CellGather	a_cellGather
	)

inline

Deposit a custom per-particle scalar (computed by a_cellGather) onto the AMR mesh.

Generalizes depositWeight() to an arbitrary per-particle value (a product of columns such as weight*mobility), dispatching on the coarse-fine strategy exactly like depositWeight(). Deposits one mesh component.

Template Parameters

P	Payload type (deduced).
Traits	Column descriptor (deduced).
CellGather	Callable (const ParticleSoA<P, Traits>& a_leaf, std::size_t a_i) -> Real.

Parameters

[in,out]	a_meshData	AMR mesh data (one component; RESET then filled).
[in]	a_particles	Particles to deposit.
[in]	a_depositionType	Deposition kernel (NGP/CIC/TSC).
[in]	a_coarseFineDeposition	Coarse-fine handling (Interp / Halo / HaloNGP / Transition).
[in]	a_irregularDeposition	Cut-cell deposition strategy.
[in]	a_cellGather	Per-particle value gatherer (leaf, index) -> Real.

◆ depositHaloCore()

template<int NCOMP, typename Container , typename PatchDeposit >

void EBAMRParticleMesh::depositHaloCore	(	EBAMRCellData &	a_meshData,
		const Container &	a_particles,
		PatchDeposit	a_patchDeposit
	)		const

inlineprotected

The CoarseFineDeposition::Halo deposition algorithm.

Like Interp for the on-level deposit + fine-ghost->coarse transfer, but instead of interpolating the coarse mass under the fine grid, it re-deposits the coarse-side halo particles (selected by the width-1 outer-halo mask, copied into the container mask holder) on the refined-coarse grid with widthScale = refinement ratio, and adds that to the fine level.

Template Parameters

NCOMP	Number of mesh components (compile-time; selects the FiCo buffer).
Container	ParticleContainer type (deduced).
PatchDeposit	Callable (const EBParticleMesh&, EBCellFAB&, const ParticleSoA&, Real widthScale).

Parameters

[in,out]	a_meshData	AMR mesh data (RESET then filled).
[in]	a_particles	Particle container (its mask holder is filled then cleared).
[in]	a_patchDeposit	Per-patch deposition functor.

◆ depositHaloNGPCore()

template<int NCOMP, typename Container , typename PatchDeposit , typename PatchDepositNGP >

void EBAMRParticleMesh::depositHaloNGPCore	(	EBAMRCellData &	a_meshData,
		const Container &	a_particles,
		PatchDeposit	a_patch,
		PatchDepositNGP	a_patchNGP
	)		const

inlineprotected

The CoarseFineDeposition::HaloNGP deposition algorithm.

Moves the coarse-side halo particles (width-1 mask) out of the valid holder, deposits the remaining (non-halo) particles with the requested kernel and the halo particles with NGP onto the same level mesh, folds fine-ghost mass to the coarse level, then transfers the halo particles back to the valid holder. NGP keeps the near-boundary clouds from spreading over the refinement boundary, so no refined-coarse buffer is needed.

Template Parameters

NCOMP	Number of mesh components (compile-time).
Container	ParticleContainer type (deduced).
PatchDeposit	Per-patch deposit with the requested kernel.
PatchDepositNGP	Per-patch deposit forced to NGP.

Parameters

[in,out]	a_meshData	AMR mesh data (RESET then filled).
[in]	a_particles	Particle container (halo particles are moved out then back).
[in]	a_patch	Per-patch deposit functor (requested kernel).
[in]	a_patchNGP	Per-patch deposit functor (NGP).

◆ depositInterpCore()

template<typename Container , typename PatchDeposit >

void EBAMRParticleMesh::depositInterpCore	(	EBAMRCellData &	a_meshData,
		const Container &	a_particles,
		const int	a_numComp,
		PatchDeposit	a_patchDeposit
	)		const

inlineprotected

The CoarseFineDeposition::Interp deposition algorithm, parameterized by the per-patch deposit.

Resets the mesh, deposits each level's valid particles (via a_patchDeposit), folds ghost-deposited mass into the owner patch (level Copier + EBAddOp), then moves the over-boundary mass between levels with the coarse-fine operators. Uses only valid particles.

Template Parameters

Container	ParticleContainer type (deduced).
PatchDeposit	Callable (const EBParticleMesh&, EBCellFAB&, const ParticleSoA&, Real widthScale).

Parameters

[in,out]	a_meshData	AMR mesh data (RESET then filled).
[in]	a_particles	Particle container.
[in]	a_numComp	Number of mesh components written.
[in]	a_patchDeposit	Per-patch deposition functor.

◆ depositTransitionCore()

template<int NCOMP, typename Container , typename PatchDeposit >

void EBAMRParticleMesh::depositTransitionCore	(	EBAMRCellData &	a_meshData,
		const Container &	a_particles,
		const DepositionType	a_depositionType,
		PatchDeposit	a_patch
	)		const

inlineprotected

The CoarseFineDeposition::Transition deposition algorithm.

NGP short-circuits to the Interp result. Otherwise: moves the coarse-side transition particles out of the valid holder; deposits the remaining particles with the kernel and folds fine-ghost mass to coarse; then deposits the transition particles on the refined-coarse grid with the fine-grid width and distributes that buffer to both the fine level (addFiCoDataToFine + exchangeAndAddFiCoData) and the coarse level (restrictAndAddFiCoDataToCoar); finally returns the transition particles to the valid holder.

Template Parameters

NCOMP	Number of mesh components (compile-time).
Container	ParticleContainer type (deduced).
PatchDeposit	Per-patch deposit with the requested kernel.

Parameters

[in,out]	a_meshData	AMR mesh data (RESET then filled).
[in]	a_particles	Particle container.
[in]	a_depositionType	Deposition kernel.
[in]	a_patch	Per-patch deposit functor (requested kernel).

◆ depositWeight()

template<typename P , typename Traits >

void EBAMRParticleMesh::depositWeight	(	EBAMRCellData &	a_meshData,
		const ParticleContainer< P, Traits > &	a_particles,
		const DepositionType	a_depositionType,
		const CoarseFineDeposition	a_coarseFineDeposition,
		const IrregularDeposition	a_irregularDeposition
	)

inline

Deposit the container-owned weight onto the AMR mesh.

Dispatches on the coarse-fine strategy (Interp / Halo / HaloNGP / Transition); see the class documentation. Halo/HaloNGP/Transition need ghost >= 1.

Template Parameters

P	Payload type (deduced).
Traits	Column descriptor (deduced).

Parameters

[in,out]	a_meshData	AMR mesh data (one component; RESET then filled).
[in]	a_particles	Particles to deposit.
[in]	a_depositionType	Deposition kernel (NGP/CIC/TSC).
[in]	a_coarseFineDeposition	Coarse-fine handling (Interp / Halo / HaloNGP / Transition).
[in]	a_irregularDeposition	Cut-cell deposition strategy.

◆ getTransitionMaskWidth()

int EBAMRParticleMesh::getTransitionMaskWidth	(	const DepositionType	a_depositionType,
		const int	a_refRat
	)		const

inlineprotected

Transition-mask width for a kernel: CIC -> refRat/2, TSC -> refRat (NGP is handled upstream).

Parameters

[in]	a_depositionType	Deposition kernel.
[in]	a_refRat	Refinement ratio between the level and the finer one.

Returns: The transition band width (in fine cells).

◆ interpolate()

template<auto... Members, typename P , typename Traits >

void EBAMRParticleMesh::interpolate	(	ParticleContainer< P, Traits > &	a_particles,
		const EBAMRCellData &	a_meshData,
		const DepositionType	a_interpType,
		const bool	a_forceIrregNGP
	)		const

inline

Interpolate a mesh field onto one or more payload columns, across all levels.

Per level, per patch: calls EBParticleMesh::interpolate. The mesh is assumed to be valid+ghosted by the caller (no coarse-fine coupling is needed for interpolation).

Template Parameters

Members	Pack of payload data-member pointers selecting the target column(s).
P	Payload type (deduced).
Traits	Column descriptor (deduced).

Parameters

[in,out]	a_particles	Particles whose target columns are overwritten.
[in]	a_meshData	AMR mesh data (>= sizeof...(Members) components).
[in]	a_interpType	Interpolation kernel (NGP/CIC/TSC).
[in]	a_forceIrregNGP	Force NGP in cut cells.

◆ interpolateWeight()

template<typename P , typename Traits >

void EBAMRParticleMesh::interpolateWeight	(	ParticleContainer< P, Traits > &	a_particles,
		const EBAMRCellData &	a_meshData,
		const DepositionType	a_interpType,
		const bool	a_forceIrregNGP
	)		const

inline

Interpolate a mesh field onto the container-owned weight column, across all levels.

Per level, per patch: calls EBParticleMesh::interpolateWeight. The mesh is assumed to be valid+ghosted by the caller (no coarse-fine coupling is needed for interpolation). This is the interpolation mirror of depositWeight.

Template Parameters

P	Payload type (deduced).
Traits	Column descriptor (deduced).

Parameters

[in,out]	a_particles	Particles whose weight column is overwritten.
[in]	a_meshData	AMR mesh data (>= 1 component).
[in]	a_interpType	Interpolation kernel (NGP/CIC/TSC).
[in]	a_forceIrregNGP	Force NGP in cut cells.

◆ operator=()

EBAMRParticleMesh & EBAMRParticleMesh::operator= ( const EBAMRParticleMesh & )

delete

Copy assignment is deleted.

Returns: Reference to this.

◆ transferMaskParticlesTransition()

template<typename Container >

void EBAMRParticleMesh::transferMaskParticlesTransition	(	Container &	a_particles,
		const DepositionType	a_depositionType
	)		const

inlineprotected

Move the coarse-side transition particles into the container mask holder (using the FiCo mask).

The transition mask lives on the refined-coarse grid, so this is done directly here (the container's transferMaskParticles works on the valid grids). Coarse particles whose fine-resolution cell is marked are moved (swap-and-pop) from the valid to the mask holder.

Template Parameters

Container ParticleContainer type (deduced).

Parameters

[in,out]	a_particles	Particle container.
[in]	a_depositionType	Deposition kernel (selects the mask width; NGP is a no-op).

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

Source/Particle/CD_EBAMRParticleMesh.H

Public Member Functions

Protected Member Functions

Protected Attributes

Detailed Description

Interpolation

Deposition

Cut cells, and what this class does NOT do

Member Function Documentation

◆ define()

◆ defineOuterHaloMasks()

◆ defineTransitionMasks()

◆ deposit()

◆ depositGathered()

◆ depositHaloCore()

◆ depositHaloNGPCore()

◆ depositInterpCore()

◆ depositTransitionCore()

◆ depositWeight()

◆ getTransitionMaskWidth()

◆ interpolate()

◆ interpolateWeight()

◆ operator=()

◆ transferMaskParticlesTransition()