Class that solves certain problems on a regular 2D grid, consisting of n x m nodes. More...

#include <grid.h>

Collaboration diagram for oofem::Grid:

Public Member Functions
	Grid (int N, int M)
	Constructor (N = horizontal dimension, M = vertical dimension) More...

	~Grid ()
	Destructor. More...

int	giveSizeHorizontal ()
	Size information. More...

int	giveSizeVertical ()

int	giveSize (int dir)

void	unFreeze ()
	Set all flags to "unfrozen". More...

void	setMethod (int o, double i, int c)
	Set the details of the algorithm to be used. More...

void	setPrescribedFieldValue (int i, int j, double val)
	Method setting the values of the prescribed field (e.g., front velocities) More...

FloatMatrix *	givePrescribedField ()

void	setZeroValues (FloatMatrix *gridCoords)
	Methods setting the values of the unknown field at selected points (e.g., zero times) More...

void	setSolutionValueAt (int i, int j, double value)

double	giveSolutionValueAt (int i, int j)
	Output methods. More...

void	printSolutionAsData ()

Private Member Functions
void	fastMarch (int &eFlag)
	Fast marching method, solving the eikonal equation. More...

double	calcTime (int i, int j, double Fij, int ord, int &eFlag)
	Auxiliary function that evaluates the tentative value at one grid point, exploited by the fast marching method. More...

int	ij2ind (int i, int j, int m)
	Utility methods. More...

int	ind2i (int ind, int m)

int	ind2j (int ind, int m)

bool	isInDomain (int i, int j, int m, int n)

Private Attributes
int	n
	Grid dimensions: number of grid nodes horizontally (n) and vertically (m) More...

int	m

int	order
	Algorithmic parameters. More...

double	initDiag

int	centDiff

bool	solutionAvailable
	Flag indicating whether the solution has been computed. More...

FloatMatrix *	T
	Matrix storing the values of the unknown (computed) field (e.g., arrival times) More...

FloatMatrix *	F
	Matrix storing the values of the prescribed field (e.g., front velocities) More...

bool *	Frozen
	Array storing indicators of frozen points (at which the solution is already known) More...

Heap *	narrowBand
	Heap used for efficient sorting and detection of smallest candidate. More...

Detailed Description

Class that solves certain problems on a regular 2D grid, consisting of n x m nodes.

Currently it implements the fast marching algorithm that solves the eikonal equation. This can be useful for instance for tracing evolving curves. One could also implement finite differences or similar techniques.

The indexing convention is a bit unusual but it follows the original code, which was developed for image analysis. Subscripts "i" and "j" refer to rows and columns of the matrix, but this means that "i" grows in the vertical direction and corresponds to physical coordinate "y". i ... from 1 to m, corresponds to coordinate y j ... from 1 to n, corresponds to coordinate x Actually, for the numerical method it does not matter which direction is horizontal and which one is vertical, but the user should know how to interpret the data. The constructor should get the horizontal dimension first and the vertical one second (n,m). The coordinates of points at which zero value is prescribed should be passed as (x,y). The prescribed field (e.g., propagating front velocity) should be stored by rows, i.e., (1,1), (1,2), (1,3), ... (1,m), (2,1), (2,2,) ... (n,m). If the solution is printed as a matrix, it is arranged on the screen in the natural way. If the solution is printed as data, each output line corresponds to (x,y,value), where x=j and y=i. Conversion to grids with non-unit spacing and shifter corner must be done by the user.

Definition at line 32 of file grid.h.