com.mockturtlesolutions.snifflib.util

Class Qdecomp

java.lang.Object

com.mockturtlesolutions.snifflib.util.Qdecomp

All Implemented Interfaces:

Displayable

public class Qdecomp
extends java.lang.Object
implements Displayable

Qdecomp intends to be a Collection backed by N-dimensional data stored in DblMatrix arrays.

Field Summary

Fields

Modifier and Type	Field and Description
protected int	Dimension
protected DblMatrix	LowerBounds
protected int	MaxCount
protected DblMatrix	MaxExtent
int	MaxLevel
protected java.lang.String	Method
protected int	MinCount
protected DblMatrix	MinExtent
QuadTreeFun	QuadFun
protected Nhood	QuadNhood
QTreeNode[]	QuadTree
protected java.lang.String	ResponseType
protected int	SplitSize
protected DblMatrix	Threshold
protected DblMatrix	UpperBounds

Constructor Summary

Constructors

Constructor and Description
Qdecomp(DblMatrix[] XData, DblMatrix YData) Default constructor.

Method Summary

Methods

Modifier and Type	Method and Description
void	coalesce() Iteratively perform rejoin operations until only a single quad is left.
void	coalesce(int Level) Iteratively perform rejoin operations until specified level is reached.
void	decompose() Fully decompose the data using the quadtree approach.
int[]	getNeighbors(int Qind) Find the neighbors of quad I.
void	growQuadtree() Reallocated the QuadTree for 2 additional levels.
boolean	hasChildren(int I) Determine if the node at index I has children.
static int[]	indexOfChildren(int I, int dim, int level) Return the quadtree indices of the Children of the current node.
static int	indexOfParent(int I, int dim, int level) Return the index of a given quad's parent quad.
static int	indToQind(int Dim, int Level, int I) Convert ND matrix index into quadtree index.
int	locate(DblMatrix[] refpt) Search the quadtree for a quad that contains the given coordinate.
int	maxLevel() Return the maximum level of any quad in the decomposition.
static int	qindToInd(int Dim, int Level, int I) Convert quadtree index into ND matrix index.
void	quadBlkProc() Process data using the current QuadFun.
void	rcoalesce(int I, int Level) Helper for coalesce.
void	rejoin(int I) Replace the children of the indicated quad's parent with the union of the data from each of the children.
int	rlocate(DblMatrix[] refpt, int queryLevel, int quadIndex) Search the quadtree for a quad that contains the given coordinate.
void	setMinCount(int K) Set the minimum observation count within a quad.
void	setQuadResponse(DblMatrix newYData, int I) Set the response data for the given quad.
void	setQuadTree(QTreeNode[] newQ) Set pointer to the input QuadTree[].
void	show() Show the decomposition.
void	show(java.lang.String Label) Show the decomposition using a given label.
int[]	Size() Return the dimensions of the quad tree relative to a N-D cube.
static DblMatrix[][]	solveBin(DblMatrix[] XData, DblMatrix YData, DblMatrix CenterOfQuad, java.lang.String Method, DblMatrix Indices) Solve the binning problem for a newly formed quad.
void	updateQuadNhood() Update the Nhood size to equal that of the QuadTree.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

QuadTree

public QTreeNode[] QuadTree

Method

protected java.lang.String Method

Dimension

protected int Dimension

SplitSize

protected int SplitSize

ResponseType

protected java.lang.String ResponseType

MaxCount

protected int MaxCount

MinCount

protected int MinCount

MaxExtent

protected DblMatrix MaxExtent

MinExtent

protected DblMatrix MinExtent

MaxLevel

public int MaxLevel

Threshold

protected DblMatrix Threshold

QuadFun

public QuadTreeFun QuadFun

QuadNhood

protected Nhood QuadNhood

UpperBounds

protected DblMatrix UpperBounds

LowerBounds

protected DblMatrix LowerBounds

Constructor Detail

Qdecomp

public Qdecomp(DblMatrix[] XData,
       DblMatrix YData)

Default constructor.

Method Detail

setMinCount

public void setMinCount(int K)

Set the minimum observation count within a quad. Quads that would have descendents with less than this number of observations will not be split.

indToQind

public static int indToQind(int Dim,
            int Level,
            int I)

Convert ND matrix index into quadtree index.

qindToInd

public static int qindToInd(int Dim,
            int Level,
            int I)

Convert quadtree index into ND matrix index.

solveBin

public static DblMatrix[][] solveBin(DblMatrix[] XData,
                     DblMatrix YData,
                     DblMatrix CenterOfQuad,
                     java.lang.String Method,
                     DblMatrix Indices)

Solve the binning problem for a newly formed quad. This algorithm requires input of the possibly multivariate coordinate data, the response data, the **center** of the multivariate bin and the binning method. There are 2 possible methods for the binning. Methods: -------- Unitized : The XData is in index order and can simply be indexed into to give the correct results. Bisection : The XData is unordered and must be partitioned. The output is a DblMatrix[][] backed by the original input data. Changes in the input data will be reflected in the output data and vice versa. The output is formatted as XData(Q1,C1), XData(Q1,C2), ... XData(Q1,Cn), YData(Q1) XData(Q2,C1), XData(Q2,C2), ... XData(Q2,Cn), YData(Q2) . . . XData(Qm,C1), XData(Qm,C2), ... XData(Qm,Cn), YData(Q2) for quad 1 through m and coordinate dimensions 1 through n. Note that the response data is always at the end of the array.

indexOfChildren

public static int[] indexOfChildren(int I,
                    int dim,
                    int level)

Return the quadtree indices of the Children of the current node. indexOfChildren(int I,int dim,int level) where I is the index, dim is the dimension and level is the level at which the kinship is evaluated. The indices of the Children are given.

growQuadtree

public void growQuadtree()

Reallocated the QuadTree for 2 additional levels. This function allows the array storing the quadtree to grow. The additional quadtrees will all be null initially.

setQuadTree

public void setQuadTree(QTreeNode[] newQ)

Set pointer to the input QuadTree[]. The methods overwrites the QuadTree array in the Qdecomp with the one that is input.

show

public void show(java.lang.String Label)

Show the decomposition using a given label.

Specified by:

show in interface Displayable

show

public void show()

Show the decomposition.

Specified by:

show in interface Displayable

updateQuadNhood

public void updateQuadNhood()

Update the Nhood size to equal that of the QuadTree. Then number of quads is determined as the maxiumum number of quads assuming all quads are at the current maximum level of any quad (i.e. 2^Dimension(Level-1) entries in each dimension).

decompose

public void decompose()

Fully decompose the data using the quadtree approach.

hasChildren

public boolean hasChildren(int I)

Determine if the node at index I has children.

coalesce

public void coalesce()

Iteratively perform rejoin operations until only a single quad is left.

maxLevel

public int maxLevel()

Return the maximum level of any quad in the decomposition.

Size

public int[] Size()

Return the dimensions of the quad tree relative to a N-D cube.

coalesce

public void coalesce(int Level)

Iteratively perform rejoin operations until specified level is reached.

rcoalesce

public void rcoalesce(int I,
             int Level)

Helper for coalesce. This method begins the search for the non-null quad at a given index.

rejoin

public void rejoin(int I)

Replace the children of the indicated quad's parent with the union of the data from each of the children. Thus the quads that are joined are the indicated quad's Children.

locate

public int locate(DblMatrix[] refpt)

Search the quadtree for a quad that contains the given coordinate.

rlocate

public int rlocate(DblMatrix[] refpt,
          int queryLevel,
          int quadIndex)

Search the quadtree for a quad that contains the given coordinate.

indexOfParent

public static int indexOfParent(int I,
                int dim,
                int level)

Return the index of a given quad's parent quad.

setQuadResponse

public void setQuadResponse(DblMatrix newYData,
                   int I)

Set the response data for the given quad. The input data must be equally dimensioned relative to the quad's current response data. Note:Setting of the coordinate data is not allowed as this has too many pitfalls and can easily break multiple splitting conditions in ways that are difficult to manage.

quadBlkProc

public void quadBlkProc()

Process data using the current QuadFun.

getNeighbors

public int[] getNeighbors(int Qind)

Find the neighbors of quad I. The Nhood QuadNhood neighborhood type is used in establishing neighbors. Users may modify the settings of the QuadNhood to their specific needs.