Package de.lmu.ifi.dbs.elki.distance.distanceresultlist

Classes for building and storing the results of distance-based queries The classes in this package essentially form three groups: KNNHeap for building kNN results.

See: Description

Interface Summary

Interface	Description
DistanceDBIDResult<D extends Distance<D>>	Collection of objects and their distances.
DistanceDBIDResultIter<D extends Distance<D>>	Iterator over distance-based query results.
DoubleDistanceDBIDResultIter	Iterator for double valued distance-based query results.
KNNHeap<D extends Distance<D>>	Interface for kNN heaps.
KNNResult<D extends Distance<D>>	Interface for kNN results.
ModifiableDistanceDBIDResult<D extends Distance<D>>	Modifiable API for Distance-DBID results

Class Summary

Class	Description
AbstractKNNHeap<P extends DistanceDBIDPair<D>,D extends Distance<D>>	Heap used for KNN management.
AbstractKNNHeap.Comp	Comparator to use.
DistanceDBIDResultUtil	Utility classes for distance DBID results.
DoubleDistanceDBIDList	Default class to keep a list of distance-object pairs.
DoubleDistanceKNNHeap	Heap for collecting double-valued KNN instances.
DoubleDistanceKNNHeap.Comp	Comparator to use.
DoubleDistanceKNNList	Finalized KNN List.
GenericDistanceDBIDList<D extends Distance<D>>	Default class to keep a list of distance-object pairs.
GenericKNNHeap<D extends Distance<D>>	Heap for collecting kNN candiates with arbitrary distance types.
GenericKNNList<D extends Distance<D>>	Finalized KNN List.
KNNUtil	Helper classes for kNN results.
KNNUtil.DistanceItr<D extends Distance<D>>	Proxy iterator for accessing DBIDs.
KNNUtil.DistanceView<D extends Distance<D>>	A view on the Distances of the result.
KNNUtil.KNNSubList<D extends Distance<D>>	Sublist of an existing result to contain only the first k elements.

Package de.lmu.ifi.dbs.elki.distance.distanceresultlist Description

Classes for building and storing the results of distance-based queries

The classes in this package essentially form three groups:

1. KNNHeap for building kNN results. It allows adding new candidates (and loses old candidates automatically), but it is not iterable.
   To get an instance, use KNNUtil.newHeap(de.lmu.ifi.dbs.elki.distance.distancefunction.DistanceFunction<?, D>, int)!
2. KNNResult is the final kNN result obtained by serializing a heap via KNNHeap.toKNNList(). It is iterable and totally ordered, but can no longer be modified (unless you call KNNUtil.newHeap(de.lmu.ifi.dbs.elki.distance.distancefunction.DistanceFunction<?, D>, int)!
3. GenericDistanceDBIDList and the optimized counterpart DoubleDistanceDBIDList, are modifiable, but not necessarily sorted lists of neighbors, useful for example for range queries.

Try to choose the most appropriate one! Heaps are optimized for updates but bad for reading, KNNResult is optimized for reading but unmodifiable, and the lists are easy to modify, but less efficient than heaps.