Detailed Description

The kD trees in LALInference are composed of cells.

Each cell represents a rectangular region in parameter space, defined by \(\mathrm{lowerLeft} <= x <= \mathrm{upperRight}\). It also contains two sub-cells, each of which represents rectangular regions of half the size. The dimension along which a cell at a particular level in the tree is split in half is given by its level (mod the dimension of the space).

Each cell contains some number (which may be zero) of points. Periodically, the cell will compute the mean and covariance of its points, and the eigenvectors of the covariance matrix (principal axes) of an ellipse fitting the the points. It will also compute the (tight) bounds of a box enclosing the points in a coordinate system aligned with the principal axes. When this has been done, the eigenFrameStale element will be set to zero. If points are subsequently added to the cell, the eigenFrameStale flag will be set to a non-zero value until the next re-computation of the principal axes.

Definition at line 915 of file LALInference.h.

Data Fields
size_t	npts

size_t	ptsSize
	Stores the number of tree points that lie in the cell. More...

size_t	dim
	Size of the pts buffer. More...

REAL8 **	pts
	Dimension of the system. More...

REAL8 *	ptsMean

REAL8 *	lowerLeft
	Mean of pts. More...

REAL8 *	upperRight
	Lower left (i.e. More...

REAL8 **	ptsCov
	Upper right (i.e. More...

REAL8 **	ptsCovEigenVects
	dim-by-dim covariance matrix. More...

REAL8 *	eigenMin
	Eigenvectors of the covariance matrix: [i][j] is the jth component of the ith eigenvector. More...

REAL8 *	eigenMax
	Minimum coordinates of points in the eigen-frame. More...

int	eigenFrameStale
	Maximum coordinates of points in the eigen-frame. More...

struct tagLALInferenceKDTree *	left
	== 1 when the mean, covariance, and eigenvectors are out of date (i.e. More...

struct tagLALInferenceKDTree *	right
	Left (i.e. More...

Field Documentation

◆ npts

size_t LALInferenceKDTree::npts

Definition at line 916 of file LALInference.h.

◆ ptsSize

size_t LALInferenceKDTree::ptsSize

Stores the number of tree points that lie in the cell.

Definition at line 917 of file LALInference.h.

◆ dim

size_t LALInferenceKDTree::dim

Size of the pts buffer.

Definition at line 918 of file LALInference.h.

◆ pts

REAL8** LALInferenceKDTree::pts

Dimension of the system.

Definition at line 919 of file LALInference.h.

◆ ptsMean

REAL8* LALInferenceKDTree::ptsMean

Definition at line 920 of file LALInference.h.

◆ lowerLeft

REAL8* LALInferenceKDTree::lowerLeft

Mean of pts.

Definition at line 921 of file LALInference.h.

◆ upperRight

REAL8* LALInferenceKDTree::upperRight

Lower left (i.e.

coordinate minimum) bound; length is ndim from LALInferenceKDTree.

Definition at line 923 of file LALInference.h.

◆ ptsCov

REAL8** LALInferenceKDTree::ptsCov

Upper right (i.e.

coordinate maximum) bound.

Definition at line 924 of file LALInference.h.

◆ ptsCovEigenVects

REAL8** LALInferenceKDTree::ptsCovEigenVects

dim-by-dim covariance matrix.

Definition at line 925 of file LALInference.h.

◆ eigenMin

REAL8* LALInferenceKDTree::eigenMin

Eigenvectors of the covariance matrix: [i][j] is the jth component of the ith eigenvector.

Definition at line 928 of file LALInference.h.

◆ eigenMax

REAL8* LALInferenceKDTree::eigenMax

Minimum coordinates of points in the eigen-frame.

Definition at line 929 of file LALInference.h.

◆ eigenFrameStale

int LALInferenceKDTree::eigenFrameStale

Maximum coordinates of points in the eigen-frame.

Definition at line 930 of file LALInference.h.

◆ left

struct tagLALInferenceKDTree* LALInferenceKDTree::left

== 1 when the mean, covariance, and eigenvectors are out of date (i.e.

more points added).

Definition at line 933 of file LALInference.h.

◆ right

struct tagLALInferenceKDTree* LALInferenceKDTree::right

Left (i.e.

lower-coordinate) sub-tree, may be NULL if empty.

Definition at line 936 of file LALInference.h.