ReinitializationEllipticOperatorNewton< dim, number > Class Template Reference

Developer Documentation: MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number > Class Template Reference

Developer Documentation

Operator for the matrix-free evaluation of diffusive reinitialization of the level set function. More...

#include <reinitialization_elliptic_operator_CG_newton.hpp>

Inheritance diagram for MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >:

[legend]

Collaboration diagram for MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >:

[legend]

Public Types
using	VectorType = dealii::LinearAlgebra::distributed::Vector< number >

using	VectorizedArrayType = dealii::VectorizedArray< number >

using	MappingInfoType = CutUtil::MappingInfoType< dim, number >

using	PointEvaluationType = dealii::FEPointEvaluation< 1, dim, dim, VectorizedArrayType >

Public Types inherited from MeltPoolDG::OperatorMatrixFree< dim, number >
using	VectorType = dealii::LinearAlgebra::distributed::Vector< number >

using	BlockVectorType = dealii::LinearAlgebra::distributed::BlockVector< number >

using	SparseMatrixType = dealii::TrilinosWrappers::SparseMatrix

Public Member Functions
	ReinitializationEllipticOperatorNewton (const MeltPoolDG::ScratchData< dim, dim, number > &scratch_data_in, const ReinitializationData< number > &reinit_data_in, const unsigned int reinit_dof_idx_in, const unsigned int reinit_quad_idx_in, const MappingInfoType &mapping_info_surface_in, const unsigned int ls_dof_idx_in)
	Constructor.

void	vmult (VectorType &dst, const VectorType &src) const final
	Apply the operator in a matrix-free fashion. Top-level function for the evaluation of the left-hand side.

void	create_residual (VectorType &rhs, const VectorType &level_set) const
	Create the right-hand side vector using a matrix-free evaluation. Top-level function.

void	reinit () final
	Reinitialize internal data structures, typically after mesh changes.

void	compute_system_matrix_from_matrixfree (dealii::TrilinosWrappers::SparseMatrix &system_matrix) const final
	Compute and assemble the system matrix from matrix-free operator evaluations. Used by the preconditioner.

void	compute_inverse_diagonal_from_matrixfree (VectorType &diagonal) const final
	Compute the inverse diagonal of the system matrix. Used by the preconditioner.

Public Member Functions inherited from MeltPoolDG::OperatorMatrixFree< dim, number >
virtual	~OperatorMatrixFree ()=default

virtual void	create_rhs (VectorType &, const VectorType &) const
	Compute the right-hand side vector.

virtual void	create_rhs (BlockVectorType &, const VectorType &) const
	Compute the right-hand side vector for block vectors `dst`.

virtual void	create_rhs (VectorType &, const BlockVectorType &) const
	Compute the right-hand side vector for block vectors `src`.

virtual void	vmult (BlockVectorType &, const BlockVectorType &) const

virtual void	compute_inverse_diagonal_from_matrixfree (BlockVectorType &) const
	Compute the inverse diagonal using matrix-free techniques for block vectors.

Public Member Functions inherited from MeltPoolDG::OperatorBase< number >
virtual	~OperatorBase ()=default

void	reset_dof_index (const unsigned int dof_idx_in)
	Set the index for the current degrees of freedom (DoF).

void	reset_time_increment (const number dt)
	Set the time increment for the operator.

virtual void	pre ()
	Prepare for computations (e.g. update ghost values). Can be overridden by derived classes.

virtual void	post ()
	Finalize for computations (e.g. zero out ghost values). Can be overridden by derived classes.

Public Attributes
VectorType	solution_old
	Solution vector from the previous iteration or time step (used in matrix-free mode).

Private Member Functions
template<int n_components>
VectorizedArrayType	evaluate_coefficient (const FECellIntegrator< dim, n_components, number > &phi_old, const unsigned int q_index) const
	This evaluates the coefficient for the rhs integral.

template<int n_components>
VectorizedArrayType	evaluate_coefficient_derivative (const FECellIntegrator< dim, n_components, number > &phi_old, const unsigned int q_index) const
	This evaluates the coefficient derivative for the analytical form.

template<int n_components>
VectorizedArrayType	evaluate_coefficient_derivative (const FEFaceIntegrator< dim, n_components, number > &phi_old, const unsigned int q_index) const

void	residual_laplace_operation (FECellIntegrator< dim, 1, number > &cell_eval) const
	Calculate the contribution of the weigted Laplace operator of a cell to the rhs.

void	tangent_laplace_operation (FECellIntegrator< dim, 1, number > &cell_eval, FECellIntegrator< dim, 1, number > &phi_old) const
	Calculate the contribution of the weighted Laplace operator to the lhs.

void	cell_operation (FECellIntegrator< dim, 1, number > &interface_penalty, FECellIntegrator< dim, 1, number > &cell_eval, PointEvaluationType &interface_penalty_surface) const
	Compute the contribution of a single cell integral to the right-hand side. First-level function for the evaluation of the volume integrals.

void	interface_penalty_cell_operation (PointEvaluationType &interface_penalty_surface, FECellIntegrator< dim, 1, number > &interface_penalty, const unsigned int lane, const number penalty_coefficient) const
	Calculate the contribution of a single cell surface integral to the penalty term for the weak imposition of interface Dirichlet boundary condition.

Private Attributes
const ScratchData< dim, dim, number > &	scratch_data
	Reference to scratch data containing mesh, geometry, and FE evaluation utilities.

const ReinitializationData< number > &	reinit_data
	Reference to the reinitialization parameters and coefficients.

const unsigned int	reinit_quad_idx
	Quadrature index for evaluating terms during reinitialization.

const MappingInfoType &	mapping_info_surface
	Mapping information for integration over the interface.

dealii::FE_Q< dim >	fe_point_level_set
	Temporary FE object required by FEPointEvaluation.

const unsigned int	n_dofs_per_cell
	Number of DoFs per cell of the temporary FE.

const unsigned int	ls_dof_idx
	DOF handler index for the level set field.

VectorType	zero_interface

Additional Inherited Members
Protected Attributes inherited from MeltPoolDG::OperatorBase< number >
unsigned int	dof_idx = dealii::numbers::invalid_unsigned_int

number	time_increment = numbers::invalid_double

number	time_increment_inv = numbers::invalid_double

Detailed Description

template<int dim, typename number>
class MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >

Operator for the matrix-free evaluation of diffusive reinitialization of the level set function.

The implementation is based on the elliptic level set reinitialization method described in:

T. Adams, S. Giani, and W. M. Coombs, "A high-order elliptic PDE based level set reinitialisation method using a discontinuous Galerkin discretisation," Journal of Computational Physics, vol. 379, pp. 373–391, 2019. https://doi.org/10.1016/j.jcp.2018.12.003

Template Parameters

dim	Spatial dimension.
number	Scalar number type (e.g., double or float).

Member Typedef Documentation

◆ MappingInfoType

template<int dim, typename number >

using MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::MappingInfoType = CutUtil::MappingInfoType<dim, number>

◆ PointEvaluationType

template<int dim, typename number >

using MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::PointEvaluationType = dealii::FEPointEvaluation<1, dim, dim, VectorizedArrayType>

◆ VectorizedArrayType

template<int dim, typename number >

using MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::VectorizedArrayType = dealii::VectorizedArray<number>

◆ VectorType

template<int dim, typename number >

using MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::VectorType = dealii::LinearAlgebra::distributed::Vector<number>

Constructor & Destructor Documentation

◆ ReinitializationEllipticOperatorNewton()

template<int dim, typename number >

MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::ReinitializationEllipticOperatorNewton	(	const MeltPoolDG::ScratchData< dim, dim, number > &	scratch_data_in,
		const ReinitializationData< number > &	reinit_data_in,
		const unsigned int	reinit_dof_idx_in,
		const unsigned int	reinit_quad_idx_in,
		const MappingInfoType &	mapping_info_surface_in,
		const unsigned int	ls_dof_idx_in
	)

Constructor.

Operator for the matrix-free solution of the nonlinear elliptic reinitialization.

Parameters

scratch_data_in	Reference to the used ScratchData object.
reinit_data_in	Reference to the object for reinitialization-specific data.
reinit_dof_idx_in	Index of the used dof-handler object in `scratch_data_in`.
reinit_quad_idx_in	Index of the used quadrature object in `scratch_data_in`.
mapping_info_surface_in	Mapping information for the interface surface.
ls_dof_idx_in	DOF handler index for the level set function.

Member Function Documentation

◆ cell_operation()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::cell_operation	(	FECellIntegrator< dim, 1, number > &	interface_penalty,
		FECellIntegrator< dim, 1, number > &	cell_eval,
		PointEvaluationType &	interface_penalty_surface
	)		const

private

Compute the contribution of a single cell integral to the right-hand side. First-level function for the evaluation of the volume integrals.

Parameters

interface_penalty	Cell integrator for the penalty term.
cell_eval	Cell integrator for the Laplace operator (plus, provides function values for the penalty term).
interface_penalty_surface	Point evaluation object for the surface integral.

◆ compute_inverse_diagonal_from_matrixfree()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::compute_inverse_diagonal_from_matrixfree ( VectorType & diagonal ) const

finalvirtual

Compute the inverse diagonal of the system matrix. Used by the preconditioner.

Parameters

diagonal Output vector containing the diagonal inverse values.

Reimplemented from MeltPoolDG::OperatorMatrixFree< dim, number >.

◆ compute_system_matrix_from_matrixfree()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::compute_system_matrix_from_matrixfree ( dealii::TrilinosWrappers::SparseMatrix & system_matrix ) const

finalvirtual

Compute and assemble the system matrix from matrix-free operator evaluations. Used by the preconditioner.

Parameters

system_matrix Output sparse matrix.

Reimplemented from MeltPoolDG::OperatorMatrixFree< dim, number >.

◆ create_residual()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::create_residual	(	VectorType &	rhs,
		const VectorType &	level_set
	)		const

Create the right-hand side vector using a matrix-free evaluation. Top-level function.

Parameters

rhs	Right-hand side vector.
old_level_set	The level set vector where the signed distance property needs to be restored.

◆ evaluate_coefficient()

template<int dim, typename number >

template<int n_components>

ReinitializationEllipticOperatorNewton< dim, number >::VectorizedArrayType MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::evaluate_coefficient	(	const FECellIntegrator< dim, n_components, number > &	phi_old,
		const unsigned int	q_index
	)		const

private

This evaluates the coefficient for the rhs integral.

Parameters

psi_old	Cell integrator for the old level set values.
q_index	Index of the quadrature point.

Returns: Value of the coefficient.

◆ evaluate_coefficient_derivative() [1/2]

template<int dim, typename number >

template<int n_components>

ReinitializationEllipticOperatorNewton< dim, number >::VectorizedArrayType MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::evaluate_coefficient_derivative	(	const FECellIntegrator< dim, n_components, number > &	phi_old,
		const unsigned int	q_index
	)		const

private

This evaluates the coefficient derivative for the analytical form.

Parameters

psi_old	Cell integrator for the old level set values.
q_index	Index of the quadrature point.

Returns: Value of the coefficient derivative.

◆ evaluate_coefficient_derivative() [2/2]

template<int dim, typename number >

template<int n_components>

ReinitializationEllipticOperatorNewton< dim, number >::VectorizedArrayType MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::evaluate_coefficient_derivative	(	const FEFaceIntegrator< dim, n_components, number > &	phi_old,
		const unsigned int	q_index
	)		const

private

◆ interface_penalty_cell_operation()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::interface_penalty_cell_operation	(	PointEvaluationType &	interface_penalty_surface,
		FECellIntegrator< dim, 1, number > &	interface_penalty,
		const unsigned int	lane,
		const number	penalty_coefficient
	)		const

private

Calculate the contribution of a single cell surface integral to the penalty term for the weak imposition of interface Dirichlet boundary condition.

Parameters

interface_penalty_surface	Point evaluation object for the surface integral.
interface_penalty	Cell integrator for the penalty term.
lane	The SIMD lane for which the operation is performed. Corresponds to one cell in the cell batch.
penalty_coefficient	Penalty coefficient for the interface penalty term.

◆ reinit()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::reinit ( )

finalvirtual

Reinitialize internal data structures, typically after mesh changes.

Reimplemented from MeltPoolDG::OperatorBase< number >.

◆ residual_laplace_operation()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::residual_laplace_operation ( FECellIntegrator< dim, 1, number > & cell_eval ) const

private

Calculate the contribution of the weigted Laplace operator of a cell to the rhs.

Parameters

cell_eval Cell integrator.

◆ tangent_laplace_operation()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::tangent_laplace_operation	(	FECellIntegrator< dim, 1, number > &	cell_eval,
		FECellIntegrator< dim, 1, number > &	phi_old
	)		const

private

Calculate the contribution of the weighted Laplace operator to the lhs.

Parameters

cell_eval Cell integrator.

◆ vmult()

template<int dim, typename number >

void MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::vmult	(	VectorType &	dst,
		const VectorType &	src
	)		const

finalvirtual

Apply the operator in a matrix-free fashion. Top-level function for the evaluation of the left-hand side.

Parameters

dst	Output vector.
src	Input vector.

Reimplemented from MeltPoolDG::OperatorMatrixFree< dim, number >.

Member Data Documentation

◆ fe_point_level_set

template<int dim, typename number >

dealii::FE_Q<dim> MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::fe_point_level_set

private

Temporary FE object required by FEPointEvaluation.

◆ ls_dof_idx

template<int dim, typename number >

const unsigned int MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::ls_dof_idx

private

DOF handler index for the level set field.

◆ mapping_info_surface

template<int dim, typename number >

const MappingInfoType& MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::mapping_info_surface

private

Mapping information for integration over the interface.

◆ n_dofs_per_cell

template<int dim, typename number >

const unsigned int MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::n_dofs_per_cell

private

Number of DoFs per cell of the temporary FE.

◆ reinit_data

template<int dim, typename number >

const ReinitializationData<number>& MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::reinit_data

private

Reference to the reinitialization parameters and coefficients.

◆ reinit_quad_idx

template<int dim, typename number >

const unsigned int MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::reinit_quad_idx

private

Quadrature index for evaluating terms during reinitialization.

◆ scratch_data

template<int dim, typename number >

const ScratchData<dim, dim, number>& MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::scratch_data

private

Reference to scratch data containing mesh, geometry, and FE evaluation utilities.

◆ solution_old

template<int dim, typename number >

VectorType MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::solution_old

Solution vector from the previous iteration or time step (used in matrix-free mode).

◆ zero_interface

template<int dim, typename number >

VectorType MeltPoolDG::LevelSet::ReinitializationEllipticOperatorNewton< dim, number >::zero_interface

private

This vector is used to create an empty cell integrator buffer for cut cells. The buffer accumulates penalty contribution. This vector is initialized to zero every reinit() operation.

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

include/meltpooldg/level_set/reinitialization_elliptic_operator_CG_newton.hpp
source/level_set/reinitialization_elliptic_operator_CG_newton.cpp

ReinitializationEllipticOperatorNewton&lt; dim, number &gt; Class Template Reference

Public Types

Public Member Functions

Public Attributes

Private Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Member Typedef Documentation

◆ MappingInfoType

◆ PointEvaluationType

◆ VectorizedArrayType

◆ VectorType

Constructor & Destructor Documentation

◆ ReinitializationEllipticOperatorNewton()

Member Function Documentation

◆ cell_operation()

◆ compute_inverse_diagonal_from_matrixfree()

◆ compute_system_matrix_from_matrixfree()

◆ create_residual()

◆ evaluate_coefficient()

◆ evaluate_coefficient_derivative() [1/2]

◆ evaluate_coefficient_derivative() [2/2]

◆ interface_penalty_cell_operation()

◆ reinit()

◆ residual_laplace_operation()

◆ tangent_laplace_operation()

◆ vmult()

Member Data Documentation

◆ fe_point_level_set

◆ ls_dof_idx

◆ mapping_info_surface

◆ n_dofs_per_cell

◆ reinit_data

◆ reinit_quad_idx

◆ scratch_data

◆ solution_old

◆ zero_interface

ReinitializationEllipticOperatorNewton< dim, number > Class Template Reference