Classes
struct	Block
	Where this block sits in the global grid: `o` is the global index of inner cell (0,0,0). More...

Functions
Block	blockOf (const WyAdvector &a, I3 origin)

void	periodicFill (SField f, I3 e, int g, bool px, bool py, bool pz)
	Periodic wrap of a single-block extended field (serial reference path; the MPI path uses core's GridHalo instead).

void	clampFill (SField f, Block b, I3 gs, bool px, bool py, bool pz)
	Zero-gradient (Neumann) fill of the ghost cells that lie OUTSIDE the global domain on a non-periodic axis: each such cell takes the value at its globally CLAMPED index.

template<typename F >
void	forEachExtended (Block b, F fn)
	Visit every cell of the extended block with (local linear index, global cell index).

void	initSphere (SField c, Block b, double h, double cx, double cy, double cz, double r, int levels=4)
	Sphere of fluid (C = 1 inside), exact to ~5e-7 relative volume at R/h = 16 via the octree helper in plic.hpp.

void	initSlab (SField c, Block b, double h, int axis, double lo, double hi)
	Slab `zlo < x < zhi` along axis `axis`, exact fractions (axis-aligned planes).

void	initZalesak (SField c, Block b, double h, double cx, double cy, double r, double sw, double stop, int nsub=64)
	Zalesak's slotted disk in the (x,y) plane, uniform in z.

void	fillUniform (WyAdvector &a, Block b, double ux, double uy, double uz)
	Uniform translation. Discrete divergence is EXACTLY zero (identical values differenced).

void	fillRotation (WyAdvector &a, Block b, double h, double cx, double cy, double omega)
	Solid-body rotation about (cx,cy) in the (x,y) plane, w = 0.

double	lvSin2 (double t)
	The LeVeque 3D deformation field u = 2 sin^2(pi x) sin(2 pi y) sin(2 pi z) cos(pi t / T) v = - sin(2 pi x) sin^2(pi y) sin(2 pi z) cos(pi t / T) w = - sin(2 pi x) sin(2 pi y) sin^2(pi z) cos(pi t / T) sampled as the discrete curl of the edge vector potential A = ( 0 , -psi2(x,z) sin(2 pi y) , psi1(x,y) sin(2 pi z) ) cos(pi t / T), psi1 = sin^2(pi x) sin^2(pi y) / pi , psi2 = sin^2(pi x) sin^2(pi z) / pi .

double	lvAy (double x, double y, double z, double phase)
	A_y of the LeVeque vector potential (lives on y-edges: node x, centre y, node z).

double	lvAz (double x, double y, double z, double phase)
	A_z of the LeVeque vector potential (lives on z-edges: node x, node y, centre z).

void	fillLeVeque (WyAdvector &a, Block b, double h, double phase)

double	l1Diff (SField a, SField b, I3 e, I3 n, int g)
	sum \|a - b\| over the inner region (cell-fraction units).

double	lInfDiff (SField a, SField b, I3 e, I3 n, int g)
	max \|a - b\| over the inner region.

SField	copyOf (SField src, const char *name)

Function Documentation

◆ blockOf()

Block vofscene::blockOf	(	const WyAdvector &	a,
		I3	origin
	)

inline

Definition at line 39 of file vof_advect_scenes.hpp.

◆ periodicFill()

void vofscene::periodicFill	(	SField	f,
		I3	e,
		int	g,
		bool	px,
		bool	py,
		bool	pz
	)

inline

Periodic wrap of a single-block extended field (serial reference path; the MPI path uses core's GridHalo instead).

Sequential axis passes, so edges/corners come out right.

Definition at line 47 of file vof_advect_scenes.hpp.

References N.

◆ clampFill()

void vofscene::clampFill	(	SField	f,
		Block	b,
		I3	gs,
		bool	px,
		bool	py,
		bool	pz
	)

inline

Zero-gradient (Neumann) fill of the ghost cells that lie OUTSIDE the global domain on a non-periodic axis: each such cell takes the value at its globally CLAMPED index.

The clamp is per axis and global, which is what makes it decomposition-independent. The obvious alternative – sequential zero-gradient axis passes over the block – is NOT: a corner cell that is outside the domain in y but whose x-neighbourhood is an interior halo gets the x-extension of the true row on one decomposition and the x-extension of the y = 0 row on another, and that difference reaches the colour field through the MYC stencil of the inner corner cell. (Which is exactly the "MPI differs at the last bit only in ghost-adjacent cells" symptom the work order lists; it was found by reasoning about the fill, before it could be blamed on the halo.)

The clamped source is always inside the domain, hence either an inner cell of this block or a ghost the exchange has already filled: a ghost at global index < 0 on axis a only exists when origin[a] < g, so index 0 on that axis is inside the block's extended range. Call AFTER the halo exchange / periodic fill.

Definition at line 86 of file vof_advect_scenes.hpp.

◆ forEachExtended()

template<typename F >

void vofscene::forEachExtended	(	Block	b,
		F	fn
	)

inline

Visit every cell of the extended block with (local linear index, global cell index).

Definition at line 112 of file vof_advect_scenes.hpp.

Referenced by fillLeVeque(), fillRotation(), fillUniform(), initSlab(), initSphere(), and initZalesak().

◆ initSphere()

void vofscene::initSphere	(	SField	c,
		Block	b,
		double	h,
		double	cx,
		double	cy,
		double	cz,
		double	r,
		int	levels = `4`
	)

inline

Sphere of fluid (C = 1 inside), exact to ~5e-7 relative volume at R/h = 16 via the octree helper in plic.hpp.

Domain is [0,1]^3 with h = 1/gn.

Definition at line 127 of file vof_advect_scenes.hpp.

References forEachExtended().

◆ initSlab()

void vofscene::initSlab	(	SField	c,
		Block	b,
		double	h,
		int	axis,
		double	lo,
		double	hi
	)

inline

Slab zlo < x < zhi along axis axis, exact fractions (axis-aligned planes).

Definition at line 137 of file vof_advect_scenes.hpp.

References forEachExtended().

◆ initZalesak()

void vofscene::initZalesak	(	SField	c,
		Block	b,
		double	h,
		double	cx,
		double	cy,
		double	r,
		double	sw,
		double	stop,
		int	nsub = `64`
	)

inline

Zalesak's slotted disk in the (x,y) plane, uniform in z.

Domain [0,1]^3; the disk is a circle of radius r about (cx,cy) with a rectangular slot of half-width sw reaching up to y = stop. Fractions by n x n midpoint subsampling (the metric compares the transported field against THIS field, so the sampling error is common to both and does not enter the error).

Definition at line 151 of file vof_advect_scenes.hpp.

References forEachExtended().

◆ fillUniform()

void vofscene::fillUniform	(	WyAdvector &	a,
		Block	b,
		double	ux,
		double	uy,
		double	uz
	)

inline

Uniform translation. Discrete divergence is EXACTLY zero (identical values differenced).

Definition at line 176 of file vof_advect_scenes.hpp.

References forEachExtended().

◆ fillRotation()

void vofscene::fillRotation	(	WyAdvector &	a,
		Block	b,
		double	h,
		double	cx,
		double	cy,
		double	omega
	)

inline

Solid-body rotation about (cx,cy) in the (x,y) plane, w = 0.

Discrete divergence is EXACTLY zero: u does not depend on x and v does not depend on y, so each 1D difference is 0 bitwise.

Definition at line 188 of file vof_advect_scenes.hpp.

References forEachExtended().

◆ lvSin2()

double vofscene::lvSin2 ( double t )

The LeVeque 3D deformation field u = 2 sin^2(pi x) sin(2 pi y) sin(2 pi z) cos(pi t / T) v = - sin(2 pi x) sin^2(pi y) sin(2 pi z) cos(pi t / T) w = - sin(2 pi x) sin(2 pi y) sin^2(pi z) cos(pi t / T) sampled as the discrete curl of the edge vector potential A = ( 0 , -psi2(x,z) sin(2 pi y) , psi1(x,y) sin(2 pi z) ) cos(pi t / T), psi1 = sin^2(pi x) sin^2(pi y) / pi , psi2 = sin^2(pi x) sin^2(pi z) / pi .

(curl A reproduces the field exactly; the discrete curl reproduces it to O(h^2) while making the discrete face divergence identically zero in exact arithmetic.)

Definition at line 207 of file vof_advect_scenes.hpp.

Referenced by lvAy(), and lvAz().

◆ lvAy()

double vofscene::lvAy	(	double	x,
		double	y,
		double	z,
		double	phase
	)

A_y of the LeVeque vector potential (lives on y-edges: node x, centre y, node z).

Definition at line 212 of file vof_advect_scenes.hpp.

References lvSin2().

Referenced by fillLeVeque().

◆ lvAz()

double vofscene::lvAz	(	double	x,
		double	y,
		double	z,
		double	phase
	)

A_z of the LeVeque vector potential (lives on z-edges: node x, node y, centre z).

Definition at line 217 of file vof_advect_scenes.hpp.

References lvSin2().

Referenced by fillLeVeque().

◆ fillLeVeque()

void vofscene::fillLeVeque	(	WyAdvector &	a,
		Block	b,
		double	h,
		double	phase
	)

inline

Definition at line 222 of file vof_advect_scenes.hpp.

References forEachExtended(), lvAy(), and lvAz().

◆ l1Diff()

double vofscene::l1Diff	(	SField	a,
		SField	b,
		I3	e,
		I3	n,
		int	g
	)

inline

sum |a - b| over the inner region (cell-fraction units).

Definition at line 243 of file vof_advect_scenes.hpp.

◆ lInfDiff()

double vofscene::lInfDiff	(	SField	a,
		SField	b,
		I3	e,
		I3	n,
		int	g
	)

inline

max |a - b| over the inner region.

Definition at line 258 of file vof_advect_scenes.hpp.

◆ copyOf()

SField vofscene::copyOf	(	SField	src,
		const char *	name
	)

inline

Definition at line 272 of file vof_advect_scenes.hpp.

Classes

Functions

Function Documentation

◆ blockOf()

◆ periodicFill()

◆ clampFill()

◆ forEachExtended()

◆ initSphere()

◆ initSlab()

◆ initZalesak()

◆ fillUniform()

◆ fillRotation()

◆ lvSin2()

◆ lvAy()

◆ lvAz()

◆ fillLeVeque()

◆ l1Diff()

◆ lInfDiff()

◆ copyOf()