Package elki.index.laesa

Class LAESA.LAESAKNNSearcher<Q>

java.lang.Object

elki.index.laesa.LAESA.LAESAKNNSearcher<Q>

Type Parameters:

Q - query object

All Implemented Interfaces:

KNNSearcher<Q>

Direct Known Subclasses:

LAESA.LAESAKNNByDBIDSearcher, LAESA.LAESAKNNByObjectSearcher

Enclosing class:

LAESA<O>

public abstract class LAESA.LAESAKNNSearcher<Q>
extends java.lang.Object
implements KNNSearcher<Q>

KNN searcher class

Author:

Robert Gehde

Field Summary

Fields

Modifier and Type	Field	Description
(package private) int	maxnc	Maximum number of distance computations before also pruning reference points.

Constructor Summary

Constructors

Constructor	Description
LAESAKNNSearcher(int k)	Constructor.

Method Summary

Modifier and Type	Method	Description
private int	findInRef(DBIDRef p, DBIDArrayIter ref)	Find the given DBIDVar p in the reference iter
protected void	laesaKNNSearch(KNNHeap knns)	Search the k nearest neighbors
private int	processPoints(ModifiableDoubleDBIDList cands, DoubleDBIDListMIter iter, double threshold, double dxs, DoubleDataStore rdists, int nc)	Process a set of points
protected abstract double	queryDistance(DBIDRef p)	Compute the distance to a candidate object.

Methods inherited from class java.lang.Object

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

Methods inherited from interface elki.database.query.knn.KNNSearcher

getKNN

Field Detail

maxnc

int maxnc

Maximum number of distance computations before also pruning reference points.

Constructor Detail

LAESAKNNSearcher

public LAESAKNNSearcher(int k)

Constructor.

Parameters:

k - condition parameter

Method Detail

laesaKNNSearch

protected void laesaKNNSearch(KNNHeap knns)

Search the k nearest neighbors

Parameters:

knns - result heap

processPoints

private int processPoints(ModifiableDoubleDBIDList cands,
                          DoubleDBIDListMIter iter,
                          double threshold,
                          double dxs,
                          DoubleDataStore rdists,
                          int nc)

Process a set of points

Parameters:

cands - Points to process

iter - iterator

threshold - Pruning distance

dxs - Distance to current point

rdists - Precomputed distances

nc - Number of distance computations

Returns:

Index of best candidate

findInRef

private int findInRef(DBIDRef p,
                      DBIDArrayIter ref)

Find the given DBIDVar p in the reference iter

Parameters:

p - Object to find

ref - Reference iterator

Returns:

index into the reference points list

queryDistance

protected abstract double queryDistance(DBIDRef p)

Compute the distance to a candidate object.

Parameters:

p - Object

Returns:

Distance