Common algorithms used in FVM implementations. More...

Classes
struct	CLimiterDetails< LimiterKind >
	A traits class for limiters, see notes for "computeLimiters_impl()". More...

struct	LimiterHelpers< Type >
	Common small functions used by limiters. More...

struct	CLimiterDetails< LIMITER::BARTH_JESPERSEN >
	Barth-Jespersen specialization. More...

struct	CLimiterDetails< LIMITER::VENKATAKRISHNAN >
	Venkatakrishnan specialization. More...

struct	CLimiterDetails< LIMITER::NISHIKAWA_R3 >
	Nishikawa's R3 limiter specialization. More...

struct	CLimiterDetails< LIMITER::NISHIKAWA_R4 >
	Nishikawa's R4 limiter specialization. More...

struct	CLimiterDetails< LIMITER::NISHIKAWA_R5 >
	Nishikawa's R5 limiter specialization. More...

struct	CLimiterDetails< LIMITER::VENKATAKRISHNAN_WANG >
	Venkatakrishnan-Wang specialization. More...

struct	CLimiterDetails< LIMITER::SHARP_EDGES >
	Venkatakrishnan with sharp edge modification. More...

struct	CLimiterDetails< LIMITER::WALL_DISTANCE >
	Venkatakrishnan with wall distance modification. More...

Functions
template<size_t nDim, class FieldType , class GradientType >
void	detail::computeGradientsGreenGauss (CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm, CGeometry &geometry, const CConfig &config, const FieldType &field, const size_t varBegin, const size_t varEnd, const int idxVel, GradientType &gradient)
	Compute the gradient of a field using the Green-Gauss theorem.

template<class FieldType , class GradientType >
void	computeGradientsGreenGauss (CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm, CGeometry &geometry, const CConfig &config, const FieldType &field, const size_t varBegin, const size_t varEnd, const int idxVel, GradientType &gradient)
	Instantiations for 2D and 3D.

void	detail::computeSmatrix (su2double r11, su2double r12, su2double r13, su2double r22, su2double r23, su2double r33, su2double detR2, su2double Smatrix[][2])
	Prepare Smatrix for 2D.

void	detail::computeSmatrix (su2double r11, su2double r12, su2double r13, su2double r22, su2double r23, su2double r33, su2double detR2, su2double Smatrix[][3])
	Prepare Smatrix for 3D.

template<size_t nDim, bool periodic, class GradientType , class RMatrixType >
void	detail::solveLeastSquares (size_t iPoint, size_t varBegin, size_t varEnd, const RMatrixType &Rmatrix, GradientType &gradient)
	Solve the least-squares problem for one point.

template<size_t nDim, class FieldType , class GradientType , class RMatrixType >
void	detail::computeGradientsLeastSquares (CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm, CGeometry &geometry, const CConfig &config, bool weighted, const FieldType &field, const size_t varBegin, const size_t varEnd, const int idxVel, GradientType &gradient, RMatrixType &Rmatrix)
	Compute the gradient of a field using inverse-distance-weighted or unweighted Least-Squares approximation.

template<class FieldType , class GradientType , class RMatrixType >
void	computeGradientsLeastSquares (CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm, CGeometry &geometry, const CConfig &config, bool weighted, const FieldType &field, const size_t varBegin, const size_t varEnd, const int idxVel, GradientType &gradient, RMatrixType &Rmatrix)
	Instantiations for 2D and 3D.

template<size_t nDim, class Matrix , class Scalar >
void	detail::correctGradient (const size_t varBegin, const size_t varEnd, const int idxVel, const Scalar *n, Matrix &&gradients)
	Correct the gradient on a symmetry plane.

template<size_t nDim, class GradientType >
void	correctGradientsSymmetry (CGeometry &geometry, const CConfig &config, const size_t varBegin, const size_t varEnd, const int idxVel, GradientType &gradient)
	Correct gradients on symmetry and Euler (slip) markers to respect the conditions:

template<class FieldType , class GradientType >
void	computeLimiters (LIMITER LimiterKind, CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm1, PERIODIC_QUANTITIES kindPeriodicComm2, CGeometry &geometry, const CConfig &config, size_t varBegin, size_t varEnd, su2double umusclKappa, const FieldType &field, const GradientType &gradient, FieldType &fieldMin, FieldType &fieldMax, FieldType &limiter)
	A wrapper funtion that calls specialized implementations depending on "LimiterKind". Those implementations are generated by instantiating versions of "computeLimiters_impl" with appropriate specializations of "CLimiterDetails". See corresponding hpp files for further details.

template<size_t nDim, LIMITER LimiterKind, class FieldType , class GradientType >
void	computeLimiters_impl (CSolver *solver, MPI_QUANTITIES kindMpiComm, PERIODIC_QUANTITIES kindPeriodicComm1, PERIODIC_QUANTITIES kindPeriodicComm2, CGeometry &geometry, const CConfig &config, size_t varBegin, size_t varEnd, su2double umusclKappa, const FieldType &field, const GradientType &gradient, FieldType &fieldMin, FieldType &fieldMax, FieldType &limiter)
	Generic limiter computation for methods based on one limiter value per point (as opposed to one per edge) and per variable.

Detailed Description

Common algorithms used in FVM implementations.

Function Documentation

◆ computeGradientsGreenGauss() [1/2]

template<size_t nDim, class FieldType , class GradientType >

void detail::computeGradientsGreenGauss	(	CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm,
		CGeometry &	geometry,
		const CConfig &	config,
		const FieldType &	field,
		const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		GradientType &	gradient
	)

Compute the gradient of a field using the Green-Gauss theorem.

Note: Template nDim to allow efficient unrolling of inner loops.; Gradients can be computed only for a contiguous range of variables, defined by [varBegin, varEnd[ (e.g. 0,1 computes the gradient of the 1st variable). This can be used, for example, to compute only velocity gradients.; The function uses an optional solver object to perform communications, if none (nullptr) is provided the function does not fail (the objective of this is to improve test-ability).

Parameters

[in]	solver	- Optional, solver associated with the field (used only for MPI).
[in]	kindMpiComm	- Type of MPI communication required.
[in]	kindPeriodicComm	- Type of periodic communication required.
[in]	geometry	- Geometric grid properties.
[in]	config	- Configuration of the problem, used to identify types of boundaries.
[in]	field	- Generic object implementing operator (iPoint, iVar).
[in]	varBegin	- Index of first variable for which to compute the gradient.
[in]	varEnd	- Index of last variable for which to compute the gradient.
[in]	idxVel	- Index of velocity, or -1 if no velocity present.
[out]	gradient	- Generic object implementing operator (iPoint, iVar, iDim).

◆ computeGradientsGreenGauss() [2/2]

template<class FieldType , class GradientType >

void computeGradientsGreenGauss	(	CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm,
		CGeometry &	geometry,
		const CConfig &	config,
		const FieldType &	field,
		const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		GradientType &	gradient
	)

Instantiations for 2D and 3D.

◆ computeGradientsLeastSquares() [1/2]

template<size_t nDim, class FieldType , class GradientType , class RMatrixType >

void detail::computeGradientsLeastSquares	(	CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm,
		CGeometry &	geometry,
		const CConfig &	config,
		bool	weighted,
		const FieldType &	field,
		const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		GradientType &	gradient,
		RMatrixType &	Rmatrix
	)

Compute the gradient of a field using inverse-distance-weighted or unweighted Least-Squares approximation.

Note: See notes from computeGradientsGreenGauss.hpp.

Parameters

[in]	solver	- Optional, solver associated with the field (used only for MPI).
[in]	kindMpiComm	- Type of MPI communication required.
[in]	kindPeriodicComm	- Type of periodic communication required.
[in]	geometry	- Geometric grid properties.
[in]	weighted	- Use inverse-distance weights.
[in]	config	- Configuration of the problem, used to identify types of boundaries.
[in]	field	- Generic object implementing operator (iPoint, iVar).
[in]	varBegin	- Index of first variable for which to compute the gradient.
[in]	varEnd	- Index of last variable for which to compute the gradient.
[in]	idxVel	- Index to velocity, -1 if no velocity is present in the solver.
[out]	gradient	- Generic object implementing operator (iPoint, iVar, iDim).
[out]	Rmatrix	- Generic object implementing operator (iPoint, iDim, iDim).

◆ computeGradientsLeastSquares() [2/2]

template<class FieldType , class GradientType , class RMatrixType >

void computeGradientsLeastSquares	(	CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm,
		CGeometry &	geometry,
		const CConfig &	config,
		bool	weighted,
		const FieldType &	field,
		const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		GradientType &	gradient,
		RMatrixType &	Rmatrix
	)

Instantiations for 2D and 3D.

◆ computeLimiters()

template<class FieldType , class GradientType >

void computeLimiters	(	LIMITER	LimiterKind,
		CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm1,
		PERIODIC_QUANTITIES	kindPeriodicComm2,
		CGeometry &	geometry,
		const CConfig &	config,
		size_t	varBegin,
		size_t	varEnd,
		su2double	umusclKappa,
		const FieldType &	field,
		const GradientType &	gradient,
		FieldType &	fieldMin,
		FieldType &	fieldMax,
		FieldType &	limiter
	)

A wrapper funtion that calls specialized implementations depending on "LimiterKind". Those implementations are generated by instantiating versions of "computeLimiters_impl" with appropriate specializations of "CLimiterDetails". See corresponding hpp files for further details.

◆ computeLimiters_impl()

template<size_t nDim, LIMITER LimiterKind, class FieldType , class GradientType >

void computeLimiters_impl	(	CSolver *	solver,
		MPI_QUANTITIES	kindMpiComm,
		PERIODIC_QUANTITIES	kindPeriodicComm1,
		PERIODIC_QUANTITIES	kindPeriodicComm2,
		CGeometry &	geometry,
		const CConfig &	config,
		size_t	varBegin,
		size_t	varEnd,
		su2double	umusclKappa,
		const FieldType &	field,
		const GradientType &	gradient,
		FieldType &	fieldMin,
		FieldType &	fieldMax,
		FieldType &	limiter
	)

Generic limiter computation for methods based on one limiter value per point (as opposed to one per edge) and per variable.

Note: This implementation can be used to derive most common methods by specializing the limiter functions (e.g. Venkatakrishnan) and the geometric modifications (e.g. sharp edges), this is done via specialization of "CLimiterDetails" (all its methods). Then, a call to this function with the specialized "LimiterKind" should be added to the body of "computeLimiters()". See also the notes in computeGradientsGreenGauss.hpp

Arguments:

Parameters

[in]	solver	- Optional, solver associated with the field (used only for MPI).
[in]	kindMpiComm	- Type of MPI communication required.
[in]	kindPeriodicComm1	- Type of periodic comm. to determine min/max.
[in]	kindPeriodicComm2	- Type of periodic comm. to adjust limiters.
[in]	geometry	- Geometric grid properties.
[in]	config	- Configuration of the problem.
[in]	varBegin	- First variable index for which to compute limiters.
[in]	varEnd	- End of computation range (nVar = end-begin).
[in]	umusclKappa	- Blending parameter for U-MUSCL reconstruction.
[in]	field	- Variable field.
[in]	gradient	- Gradient of the field.
[out]	fieldMin	- Minimum field values over direct neighbors of each point.
[out]	fieldMax	- As above but maximum values.
[out]	limiter	- Reconstruction limiter for the field.

Template parameters:

Parameters

nDim	- Number of dimensions.
LimiterKind	- Used to instantiate the right details class.
FieldType	- Generic object with operator (iPoint,iVar).
GradientType	- Generic object with operator (iPoint,iVar,iDim).

◆ computeSmatrix() [1/2]

void detail::computeSmatrix	(	su2double	r11,
		su2double	r12,
		su2double	r13,
		su2double	r22,
		su2double	r23,
		su2double	r33,
		su2double	detR2,
		su2double	Smatrix[][2]
	)

inline

Prepare Smatrix for 2D.

◆ computeSmatrix() [2/2]

void detail::computeSmatrix	(	su2double	r11,
		su2double	r12,
		su2double	r13,
		su2double	r22,
		su2double	r23,
		su2double	r33,
		su2double	detR2,
		su2double	Smatrix[][3]
	)

inline

Prepare Smatrix for 3D.

◆ correctGradient()

template<size_t nDim, class Matrix , class Scalar >

void detail::correctGradient	(	const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		const Scalar *	n,
		Matrix &&	gradients
	)

inline

Correct the gradient on a symmetry plane.

Parameters

[in]	varBegin	- Start of the variables.
[in]	varEnd	- End of the variables.
[in]	idxVel	- Variable index where velocity gradients start (-1 if all variables are scalars).
[in]	n	- Normal direction.
[in,out]	gradients	- The gradients to be modified.

◆ correctGradientsSymmetry()

template<size_t nDim, class GradientType >

void correctGradientsSymmetry	(	CGeometry &	geometry,
		const CConfig &	config,
		const size_t	varBegin,
		const size_t	varEnd,
		const int	idxVel,
		GradientType &	gradient
	)

Correct gradients on symmetry and Euler (slip) markers to respect the conditions:

n.grad(phi) = 0
n.grad(v.t) = 0

t.grad(v.n) = 0

Note: See Blazek eq. 8.40.

Parameters

[in]	geometry	- Geometric grid properties.
[in]	config	- Configuration of the problem, used to identify types of boundaries.
[in]	varBegin	- Index of first variable for which to compute the gradient.
[in]	varEnd	- Index of last variable for which to compute the gradient.
[in]	idxVel	- Index to velocity, -1 if no velocity is present in the solver.
[in,out]	gradient	- Generic object implementing operator (iPoint, iVar, iDim).

◆ solveLeastSquares()

template<size_t nDim, bool periodic, class GradientType , class RMatrixType >

void detail::solveLeastSquares	(	size_t	iPoint,
		size_t	varBegin,
		size_t	varEnd,
		const RMatrixType &	Rmatrix,
		GradientType &	gradient
	)

inline

Solve the least-squares problem for one point.

Note: See detail::computeGradientsLeastSquares for the purpose of template "nDim" and "periodic".

Classes

Functions

Detailed Description

Function Documentation

◆ computeGradientsGreenGauss() [1/2]

◆ computeGradientsGreenGauss() [2/2]

◆ computeGradientsLeastSquares() [1/2]

◆ computeGradientsLeastSquares() [2/2]

◆ computeLimiters()

◆ computeLimiters_impl()

◆ computeSmatrix() [1/2]

◆ computeSmatrix() [2/2]

◆ correctGradient()

◆ correctGradientsSymmetry()

◆ solveLeastSquares()