org.apache.flink.graph.generator

Class RMatGraph<T extends org.apache.commons.math3.random.RandomGenerator>

java.lang.Object

org.apache.flink.graph.generator.GraphGeneratorBase<LongValue,NullValue,NullValue>

org.apache.flink.graph.generator.RMatGraph<T>

All Implemented Interfaces:

GraphGenerator<LongValue,NullValue,NullValue>

public class RMatGraph<T extends org.apache.commons.math3.random.RandomGenerator>
extends GraphGeneratorBase<LongValue,NullValue,NullValue>

See Also:

R-MAT: A Recursive Model for Graph Mining

Field Summary

Fields

Modifier and Type	Field and Description
static float	DEFAULT_A
static float	DEFAULT_B
static float	DEFAULT_C
static float	DEFAULT_NOISE
static int	MINIMUM_EDGE_COUNT
static int	MINIMUM_VERTEX_COUNT

Fields inherited from class org.apache.flink.graph.generator.GraphGeneratorBase

parallelism

Constructor Summary

Constructors

Constructor and Description
RMatGraph(ExecutionEnvironment env, RandomGenerableFactory<T> randomGeneratorFactory, long vertexCount, long edgeCount) A directed power-law multigraph generated using the stochastic Recursive Matrix (R-Mat) model.

Method Summary

Modifier and Type	Method and Description
Graph<LongValue,NullValue,NullValue>	generate() Generates the configured graph.
RMatGraph<T>	setConstants(float a, float b, float c) The parameters for recursively subdividing the adjacency matrix.
RMatGraph<T>	setNoise(boolean noiseEnabled, float noise) Enable and configure noise.

Methods inherited from class org.apache.flink.graph.generator.GraphGeneratorBase

setParallelism

Methods inherited from class java.lang.Object

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

Field Detail

MINIMUM_VERTEX_COUNT

public static final int MINIMUM_VERTEX_COUNT

See Also:

Constant Field Values

MINIMUM_EDGE_COUNT

public static final int MINIMUM_EDGE_COUNT

See Also:

Constant Field Values

DEFAULT_A

public static final float DEFAULT_A

See Also:

Constant Field Values

DEFAULT_B

public static final float DEFAULT_B

See Also:

Constant Field Values

DEFAULT_C

public static final float DEFAULT_C

See Also:

Constant Field Values

DEFAULT_NOISE

public static final float DEFAULT_NOISE

See Also:

Constant Field Values

Constructor Detail

RMatGraph

public RMatGraph(ExecutionEnvironment env,
                 RandomGenerableFactory<T> randomGeneratorFactory,
                 long vertexCount,
                 long edgeCount)

A directed power-law multigraph generated using the stochastic Recursive Matrix (R-Mat) model.

Parameters:

env - the Flink execution environment

randomGeneratorFactory - source of randomness

vertexCount - number of vertices

edgeCount - number of edges

Method Detail

setConstants

public RMatGraph<T> setConstants(float a,
                                 float b,
                                 float c)

The parameters for recursively subdividing the adjacency matrix.

Setting A = B = C = 0.25 emulates the Erdős–Rényi model.

Graph500 uses A = 0.57, B = C = 0.19.

Parameters:

a - likelihood of source bit = 0, target bit = 0

b - likelihood of source bit = 0, target bit = 1

c - likelihood of source bit = 1, target bit = 0

Returns:

this

setNoise

public RMatGraph<T> setNoise(boolean noiseEnabled,
                             float noise)

Enable and configure noise. Each edge is generated independently, but when noise is enabled the parameters A, B, and C are randomly increased or decreased, then normalized, by a fraction of the noise factor during the computation of each bit.

Parameters:

noiseEnabled - whether to enable noise perturbation

noise - strength of noise perturbation

Returns:

this

generate

public Graph<LongValue,NullValue,NullValue> generate()

Description copied from interface: GraphGenerator

Generates the configured graph.

Returns:

generated graph