tree Namespace Reference

Regular pointer-style trees (as opposed to THOR trees). More...

Functions
template<typename TKdTree , typename T >
success_t	LoadKdTree (datanode *module, GenMatrix< T > *matrix, TKdTree **tree_pp, ArrayList< index_t > *old_from_new)
	Loads a KD tree from a command-line parameter, creating a KD tree if necessary.
template<typename TKdTree , typename T >
TKdTree *	MakeKdTreeMidpoint (GenMatrix< T > &matrix, index_t leaf_size, GenVector< index_t > *old_from_new=NULL, GenVector< index_t > *new_from_old=NULL)
template<typename TKdTree , typename T >
TKdTree *	MakeKdTreeMidpoint (GenMatrix< T > &matrix, index_t leaf_size, ArrayList< index_t > *old_from_new=NULL, ArrayList< index_t > *new_from_old=NULL)
template<typename TKdTree , typename T >
TKdTree *	MakeKdTreeMidpointSelective (GenMatrix< T > &matrix, const Vector &split_dimensions, index_t leaf_size, GenVector< index_t > *old_from_new=NULL, GenVector< index_t > *new_from_old=NULL)
template<typename TKdTree , typename T >
TKdTree *	MakeKdTreeMidpointSelective (GenMatrix< T > &matrix, const Vector &split_dimensions, index_t leaf_size, ArrayList< index_t > *old_from_new=NULL, ArrayList< index_t > *new_from_old=NULL)
	Creates a KD tree from data, splitting on the midpoint.

---

Detailed Description

Regular pointer-style trees (as opposed to THOR trees).

---

Function Documentation

template<typename TKdTree , typename T >

success_t tree::LoadKdTree	(	datanode *	module,
		GenMatrix< T > *	matrix,
		TKdTree **	tree_pp,
		ArrayList< index_t > *	old_from_new
	)			[inline]

Loads a KD tree from a command-line parameter, creating a KD tree if necessary.

EXPERIMENTAL FEATURE.

This optionally allows the end user to write out the created KD tree to a file, as a convenience.

Requires a sub-module, with the root parameter of the submodule being the filename, and optional parameters leaflen, type, and save (see example below).

Example:

 MyKdTree *q_tree;
 Matrix q_matrix;
 ArrayList<index_t> q_permutation;
 LoadKdTree(fx_submodule(NULL, "q", "q"), &q_matrix, &q_tree,
    &q_permutation);

Command-line use:

 ./main --q=foo.txt                  # load from csv format
 ./main --q=foo.txt --q/leaflen=20   # leaf length

Parameters:

	module	the module to get parameters from
	matrix	the matrix to initialize, undefined on failure
	tree_pp	an unitialized pointer that will be set to the root of the tree, must still be freed on failure
	old_from_new	stores the permutation to get from the indices in the matrix returned to the original data point indices

Returns:

SUCCESS_PASS or SUCCESS_FAIL

Definition at line 231 of file kdtree.h.

References fx_param_int(), fx_param_str(), fx_timer_start(), fx_timer_stop(), and data::Load().

template<typename TKdTree , typename T >

TKdTree * tree::MakeKdTreeMidpointSelective	(	GenMatrix< T > &	matrix,
		const Vector &	split_dimensions,
		index_t	leaf_size,
		ArrayList< index_t > *	old_from_new = NULL,
		ArrayList< index_t > *	new_from_old = NULL
	)			[inline]

Creates a KD tree from data, splitting on the midpoint.

EXPERIMENTAL FEATURE.

This requires you to pass in two unitialized ArrayLists which will contain index mappings so you can account for the re-ordering of the matrix. (By unitialized I mean don't call Init on it)

Parameters:

	matrix	data where each column is a point, WHICH WILL BE RE-ORDERED
	leaf_size	the maximum points in a leaf
	old_from_new	pointer to an unitialized arraylist; it will map new indices to original
	new_from_old	pointer to an unitialized arraylist; it will map original indexes to new indices

Definition at line 82 of file kdtree.h.

References GenVector< T >::length(), and GenMatrix< T >::n_cols().