DiFfRG::def Namespace Reference

Default implementations and building blocks for user models, see model.hh. More...

Namespaces
namespace	BoundaryStencilIndex
	Positions in the boundary stencil, from lower to upper along the axis normal to the boundary face.
namespace	limiter_utils

Classes
class	AbstractModel
	Base class of every model: declares all hooks an assembler may call, with defaults. More...
class	AD
	All jacobians by forward-mode AD, including the dependence through the extractors and the variables' jacobian from dt_variables(). The default choice. More...
class	ADjacobian_extractors
	AD jacobian of extract() with respect to the FE values (to = 0), gradients (1) or hessians (2) at the extractor point. More...
class	ADjacobian_mass
	AD jacobian of mass() with respect to \(u\) (dot = 0) or \(\partial_t u\) (dot = 1), per point. More...
class	ADjacobian_variables
	AD jacobian of dt_variables() with respect to the variables. More...
struct	BatchADJacobians
	Marker: flux, source and numerical-flux jacobians are the AD of evaluate_batch (and of the batched numerical fluxes built on it). Inherited through def::AD and def::FE_AD; a model without it provides the per-point jacobian_* callbacks itself. More...
class	CentralLimiter
	No limiting: the central slope \(\tfrac12(a + b)\) of the backward and forward slopes \(a, b\). More...
class	ConstrainOriginBoundaryPointToZero
	Constrain the boundary dofs of scalar FE function component_name closest to the origin to zero. More...
class	ConstrainOriginSupportPointToZero
	As ConstrainOriginBoundaryPointToZero, but among all dofs of the component, not only boundary ones. More...
class	CorrectedWeightedLeastSquaresDiffusionReconstructor
	The gradients the KT diffusion flux sees at a face between cells \(K\) (side -) and \(N\) (side +). More...
struct	DefaultIndicator
	Return type of AbstractModel's cell_indicator and face_indicator, by which HAdaptivity tells them apart. More...
struct	DiffusionFaceState
	The inputs of the diffusion flux at one face: cell averages and corrected gradients of both cells, and means. More...
class	FE_AD
	As def::AD, but only with respect to the FE functions: extractors are frozen and no variables jacobian is computed. Cheaper per Newton step when extractors or variables are expensive. More...
class	FirstOrderReconstructor
	Piecewise constant reconstruction: every gradient and third derivative is zero. More...
class	FlowBoundaries
	Boundary flux = the model's own flux of the inner trace, i.e. the boundary does not constrain the flow (outflow). Also for the LDG levels, with ldg_flux. More...
struct	FlowDirections
	Per component of an LDG level: the direction \(d_i\) (0 to dim - 1) of its flux. More...
class	fRG
	Keeps the RG time \(t\) and the scale \(k = \Lambda e^{-t}\) with its powers, as protected members t, k, k2, ..., k6 and Lambda. Use either this or def::Time. More...
struct	FrozenExtractors
	Marker: the extractors are frozen within a Newton step; the AD jacobians skip their directions (def::FE_AD). More...
class	FVDefaultBoundaries
	Affine extrapolation on both boundaries: the ghosts continue the straight line through the boundary cell and the first interior cell. 1D and 2D. More...
class	LDGUpDownFluxes
	One-sided (up/down-wind) numerical fluxes for the LDG levels. More...
class	LLFFlux
	Local Lax-Friedrichs (Rusanov) numerical flux on interior faces: More...
class	MinModLimiter
	MinMod limiter, \(\tfrac12(\mathrm{sgn}\,a + \mathrm{sgn}\,b)\min(|a|, |b|)\) for the backward and forward slopes \(a, b\). More...
class	NoAffineConstraints
	No affine constraints (same as inheriting nothing). More...
class	NoJacobians
	Every jacobian callback is defined and does nothing, i.e. the jacobians are zero. A model using it shadows the callbacks it can provide by hand (e.g. KT's per-point jacobian_source*). More...
class	NoNumFlux
	Zero numerical flux on interior faces. More...
class	OriginOddLinearExtrapolationBoundaries
	Lower boundary: odd reflection about the centre of the boundary cell, which is meant to sit at the origin \(\sigma = 0\); that cell's value is set to 0 in the stencil. Upper boundary: affine extrapolation. More...
class	OriginShiftedOddLinearExtrapolationBoundaries
	As OriginOddLinearExtrapolationBoundaries, but \(u_c - b_c\) is odd: the boundary cell takes the value \(b_c\). 1D only. More...
struct	ReconstructionDerivativeData
	The derivatives of a reconstructed value, gradient and third derivatives by one stencil value. More...
class	SuperbeeLimiter
	Superbee limiter (Roe): for backward and forward slopes \(a, b\) of the same sign, \(\mathrm{sgn}\,a\, \max\big(\min(2|a|, |b|), \min(|a|, 2|b|)\big)\), else 0. More...
class	Time
	Keeps the time \(t\) for models that are not RG flows. Use either this or def::fRG. More...
class	TVDReconstructor
	Slope-limited linear reconstruction on a rectangular mesh. More...
struct	UpDown
	Per component of an LDG level: def::from_right or def::from_left. More...
struct	UpDownFlux
	The numerical flux of one LDG level: UpDownFlux<FlowDirections<...>, UpDown<...>>. More...
class	VanAlbadaLimiter
	van Albada limiter: for backward and forward slopes \(a, b\) of the same sign, \(ab(a + b)/(a^2 + b^2)\), else 0. More...

Concepts
concept	HasSlopeLimiter
	Concept of a slope limiter, the Limiter template parameter of def::TVDReconstructor.
concept	HasReconstructor
	Concept of a slope reconstruction, the Reconstructor_ / JacobianReconstructor_ of FV::KurganovTadmor::Assembler. Implementations: TVDReconstructor, FirstOrderReconstructor.
concept	HasWaveSpeed
	Concept of a wave-speed strategy, the WaveSpeedStrategy_ of FV::KurganovTadmor::Assembler.
concept	HasAbsTolerances
	Whether a model provides per-component absolute tolerances for the IDA timesteppers.

Typedefs
template<int dim, typename NumberType , size_t n_components>
using	GradientType = std::array<dealii::Tensor<1, dim, NumberType>, n_components>
	One gradient per component.
template<int dim, typename NumberType , size_t n_components>
using	ThirdDerivativeType = std::array<dealii::Tensor<3, dim, NumberType>, n_components>
	One tensor of third derivatives per component.
template<int dim, typename NumberType , std::size_t n_components>
using	BoundaryStencilValues = std::array<std::array<NumberType, n_components>, 2 * dim + 3>
	Cell averages of the boundary stencil: [stencil index][component], see fv_boundaries.hh.
template<int dim>
using	BoundaryStencilPoints = std::array<Point<dim>, 2 * dim + 3>
	Cell centres of the boundary stencil, see fv_boundaries.hh.

Variables
template<int dim>
constexpr std::size_t	n_faces = 2 * dim
	Faces of a rectangular cell in dim dimensions.
constexpr int	no_wave_speed = -1
	Block id for a component that carries no hyperbolic flux.
constexpr uint	from_right = 0
	LDGUpDownFluxes: take the flux from the cell on the + side (larger coordinate) of the face.
constexpr uint	from_left = 1
	LDGUpDownFluxes: take the flux from the cell on the - side (smaller coordinate) of the face.

Detailed Description

Default implementations and building blocks for user models, see model.hh.

Typedef Documentation

BoundaryStencilPoints

template<int dim>

using BoundaryStencilPoints = std::array<Point<dim>, 2 * dim + 3>

Cell centres of the boundary stencil, see fv_boundaries.hh.

BoundaryStencilValues

template<int dim, typename NumberType , std::size_t n_components>

using BoundaryStencilValues = std::array<std::array<NumberType, n_components>, 2 * dim + 3>

Cell averages of the boundary stencil: [stencil index][component], see fv_boundaries.hh.

GradientType

template<int dim, typename NumberType , size_t n_components>

using GradientType = std::array<dealii::Tensor<1, dim, NumberType>, n_components>

One gradient per component.

ThirdDerivativeType

template<int dim, typename NumberType , size_t n_components>

using ThirdDerivativeType = std::array<dealii::Tensor<3, dim, NumberType>, n_components>

One tensor of third derivatives per component.

Variable Documentation

from_left

uint from_left = 1

constexpr

LDGUpDownFluxes: take the flux from the cell on the - side (smaller coordinate) of the face.

from_right

uint from_right = 0

constexpr

LDGUpDownFluxes: take the flux from the cell on the + side (larger coordinate) of the face.

n_faces

template<int dim>

std::size_t n_faces = 2 * dim

inlineconstexpr

Faces of a rectangular cell in dim dimensions.

no_wave_speed

int no_wave_speed = -1

inlineconstexpr

Block id for a component that carries no hyperbolic flux.

See also

AbstractModel::wave_speed_blocks.