org.neo4j.collection.primitive.hopscotch

Class HopScotchHashingAlgorithm

java.lang.Object

org.neo4j.collection.primitive.hopscotch.HopScotchHashingAlgorithm

public class HopScotchHashingAlgorithm
extends Object

An implementation of the hop-scotch algorithm, see http://en.wikipedia.org/wiki/Hopscotch_hashing. It's a static set of methods that implements the essence of the algorithm, where storing and retrieving data is abstracted into the Table interface. Also things like monitoring and choice of HopScotchHashingAlgorithm.HashFunction gets passed in.

About hop scotching and hop bits: Each index in the table has hop bits which is a list of which nearby indexes contain entries with a key hashing to the same index as itself - its neighborhood. A neighbor can at most be H indexes away. This is the collision resolution method of choice for the hop scotch algorithm. Getting and putting entries at an index first checks the index at hand. If occupied by another entry its neighbors, dictated by the hop bits, are checked. When putting an entry and the index and all its neighbors are occupied, the hop scotching begins, where a free index further away is picked and iteratively moved closer and closer until it's within the neighborhood of the intended index. The entry is then placed at that newly "freed" index. Removing an entry will put some effort into do reverse hop scotching as well, i.e. moving a neighbor into the newly removed index and iteratively also move neighbors of the moved neighbor, and so forth. This behavior has the benefit of keeping entries hashing to the same index very close together, and so will take more advantage of CPU caches and pre-fetching in general, especially for lookups.

Why are these methods (like put(Table, Monitor, HashFunction, long, Object, ResizeMonitor), get(Table, Monitor, HashFunction, long) a.s.o. static? To reduce garbage and also reduce overhead of each set or map object making use of hop-scotch hashing where they won't need to have a reference to an algorithm object, merely use its static methods. Also, all essential state is managed by Table.

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static interface	HopScotchHashingAlgorithm.HashFunction
static interface	HopScotchHashingAlgorithm.Monitor Monitor for what how a HopScotchHashingAlgorithm changes the items in a Table.
static interface	HopScotchHashingAlgorithm.ResizeMonitor<VALUE>

Field Summary

Fields

Modifier and Type	Field and Description
static int	DEFAULT_H Default number of hop bits per index, i.e.
static HopScotchHashingAlgorithm.HashFunction	DEFAULT_HASHING The default hash function is based on a pseudo-random number generator, which uses the input value as a seed to the generator.
static HopScotchHashingAlgorithm.HashFunction	JUL_HASHING Same hash function as that used by the standard library hash collections.
static HopScotchHashingAlgorithm.Monitor	NO_MONITOR

Constructor Summary

Constructors

Constructor and Description
HopScotchHashingAlgorithm()

Method Summary

Methods

Modifier and Type	Method and Description
static <VALUE> VALUE	get(Table<VALUE> table, HopScotchHashingAlgorithm.Monitor monitor, HopScotchHashingAlgorithm.HashFunction hashFunction, long key)
static <VALUE> VALUE	put(Table<VALUE> table, HopScotchHashingAlgorithm.Monitor monitor, HopScotchHashingAlgorithm.HashFunction hashFunction, long key, VALUE value, HopScotchHashingAlgorithm.ResizeMonitor<VALUE> resizeMonitor)
static <VALUE> VALUE	remove(Table<VALUE> table, HopScotchHashingAlgorithm.Monitor monitor, HopScotchHashingAlgorithm.HashFunction hashFunction, long key)

Methods inherited from class java.lang.Object

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

Field Detail

DEFAULT_H

public static final int DEFAULT_H

Default number of hop bits per index, i.e. size of neighborhood.

See Also:

Constant Field Values

NO_MONITOR

public static final HopScotchHashingAlgorithm.Monitor NO_MONITOR

JUL_HASHING

public static final HopScotchHashingAlgorithm.HashFunction JUL_HASHING

Same hash function as that used by the standard library hash collections. It generates a hash by splitting the input value into segments, and then re-distributing those segments, so the end result is effectively a striped and then jumbled version of the input data. For randomly distributed keys, this has a good chance at generating an even hash distribution over the full hash space. It performs exceptionally poorly for sequences of numbers, as the sequence increments all end up in the same stripe, generating hash values that will end up in the same buckets in collections.

DEFAULT_HASHING

public static final HopScotchHashingAlgorithm.HashFunction DEFAULT_HASHING

The default hash function is based on a pseudo-random number generator, which uses the input value as a seed to the generator. This is very fast, and performs well for most input data. However, it is not guaranteed to generate a superb distribution, only a "decent" one.

Constructor Detail

HopScotchHashingAlgorithm

public HopScotchHashingAlgorithm()

Method Detail

get

public static <VALUE> VALUE get(Table<VALUE> table,
                HopScotchHashingAlgorithm.Monitor monitor,
                HopScotchHashingAlgorithm.HashFunction hashFunction,
                long key)

remove

public static <VALUE> VALUE remove(Table<VALUE> table,
                   HopScotchHashingAlgorithm.Monitor monitor,
                   HopScotchHashingAlgorithm.HashFunction hashFunction,
                   long key)

put

public static <VALUE> VALUE put(Table<VALUE> table,
                HopScotchHashingAlgorithm.Monitor monitor,
                HopScotchHashingAlgorithm.HashFunction hashFunction,
                long key,
                VALUE value,
                HopScotchHashingAlgorithm.ResizeMonitor<VALUE> resizeMonitor)