org.apache.spark.ml.impl

Class Utils

Object

org.apache.spark.ml.impl.Utils

public class Utils
extends Object

Constructor Summary

Constructors

Constructor and Description
Utils()

Method Summary

Modifier and Type	Method and Description
static double	EPSILON()
static int	indexUpperTriangular(int n, int i, int j) Indexing in an array representing the upper triangular part of a matrix into an n * n array representing the full symmetric matrix (column major).
static double	log1pExp(double x) When x is positive and large, computing math.log(1 + math.exp(x)) will lead to arithmetic overflow.
static double[]	unpackUpperTriangular(int n, double[] triangularValues) Convert an n * (n + 1) / 2 dimension array representing the upper triangular part of a matrix into an n * n array representing the full symmetric matrix (column major).

Methods inherited from class Object

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

Constructor Detail

Utils

public Utils()

Method Detail

EPSILON

public static double EPSILON()

unpackUpperTriangular

public static double[] unpackUpperTriangular(int n,
                                             double[] triangularValues)

Convert an n * (n + 1) / 2 dimension array representing the upper triangular part of a matrix into an n * n array representing the full symmetric matrix (column major).

Parameters:

n - The order of the n by n matrix.

triangularValues - The upper triangular part of the matrix packed in an array (column major).

Returns:

A dense matrix which represents the symmetric matrix in column major.

indexUpperTriangular

public static int indexUpperTriangular(int n,
                                       int i,
                                       int j)

Indexing in an array representing the upper triangular part of a matrix into an n * n array representing the full symmetric matrix (column major). val symmetricValues = unpackUpperTriangularMatrix(n, triangularValues) val matrix = new DenseMatrix(n, n, symmetricValues) val index = indexUpperTriangularMatrix(n, i, j) then: symmetricValues(index) == matrix(i, j)

Parameters:

n - The order of the n by n matrix.

i - (undocumented)

j - (undocumented)

Returns:

(undocumented)

log1pExp

public static double log1pExp(double x)

When x is positive and large, computing math.log(1 + math.exp(x)) will lead to arithmetic overflow. This will happen when x > 709.78 which is not a very large number. It can be addressed by rewriting the formula into x + math.log1p(math.exp(-x)) when x is positive.

Parameters:

x - a floating-point value as input.

Returns:

the result of math.log(1 + math.exp(x)).