com.cloudera.oryx.common.math

Class VectorMath

java.lang.Object

com.cloudera.oryx.common.math.VectorMath

public final class VectorMath
extends Object

Utility class with simple vector-related operations.

Method Summary

Modifier and Type	Method and Description
static double	cosineSimilarity(float[] x, float[] y, double normY) Computes cosine similarity of values in two given arrays, when the norm of one array is known in advance, which is a not-uncommon case.
static double	dot(float[] x, float[] y)
static double	norm(double[] x)
static double	norm(float[] x)
static double[]	parseVector(String[] values)
static float[]	randomVectorF(int features, org.apache.commons.math3.random.RandomGenerator random)
static double[]	transposeTimesSelf(Collection<float[]> M)

Methods inherited from class java.lang.Object

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

Method Detail

dot

public static double dot(float[] x,
                         float[] y)

Parameters:

x - one array

y - the other array

Returns:

dot product of the two given arrays

norm

public static double norm(float[] x)

Parameters:

x - vector for whom norm to be calculated

Returns:

the L2 norm of vector x

norm

public static double norm(double[] x)

Parameters:

x - vector for whom norm to be calculated

Returns:

the L2 norm of vector x

cosineSimilarity

public static double cosineSimilarity(float[] x,
                                      float[] y,
                                      double normY)

Computes cosine similarity of values in two given arrays, when the norm of one array is known in advance, which is a not-uncommon case.

Parameters:

x - one array

y - the other array

normY - norm of y

Returns:

cosine similarity = dot(x,y) / (norm(x) * norm(y))

transposeTimesSelf

public static double[] transposeTimesSelf(Collection<float[]> M)

Parameters:

M - tall, skinny matrix

Returns:

MT * M as a dense lower-triangular matrix, represented in packed row-major form.

parseVector

public static double[] parseVector(String[] values)

Parameters:

values - numeric values as Strings

Returns:

values parsed as double[]

randomVectorF

public static float[] randomVectorF(int features,
                                    org.apache.commons.math3.random.RandomGenerator random)

Parameters:

features - dimension of vector

random - random number generator

Returns:

vector whose direction from the origin is chosen uniformly at random, but which is not normalized