org.apache.mahout.math.hadoop.stochasticsvd

Class Omega

java.lang.Object

org.apache.mahout.math.hadoop.stochasticsvd.Omega

public class Omega
extends Object

simplistic implementation for Omega matrix in Stochastic SVD method

Constructor Summary

Constructors

Constructor and Description
Omega(long seed, int kp)

Method Summary

Modifier and Type	Method and Description
protected void	accumDots(int aIndex, double aElement, double[] yRow)
protected void	accumDots(int aIndex, double aElement, Vector yRow)
void	computeYRow(Vector aRow, double[] yRow) Deprecated.
void	computeYRow(Vector aRow, Vector yRowOut) A version to compute yRow as a sparse vector in case of extremely sparse matrices
double	getQuick(int row, int column) Get omega element at (x,y) uniformly distributed within [-1...1)
static long	murmur64(byte[] val, int offset, int len, long seed)
static long	murmur64(long val, int len, long seed) Shortened version for data < 8 bytes packed into len lowest bytes of val.
Vector	mutlithreadedTRightMultiply(Vector v)

Methods inherited from class java.lang.Object

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

Constructor Detail

Omega

public Omega(long seed,
             int kp)

Method Detail

getQuick

public double getQuick(int row,
                       int column)

Get omega element at (x,y) uniformly distributed within [-1...1)

Parameters:

row - omega row

column - omega column

computeYRow

@Deprecated
public void computeYRow(Vector aRow,
                                    double[] yRow)

Deprecated.

compute YRow=ARow*Omega.

Parameters:

aRow - row of matrix A (size n)

yRow - row of matrix Y (result) must be pre-allocated to size of (k+p)

computeYRow

public void computeYRow(Vector aRow,
                        Vector yRowOut)

A version to compute yRow as a sparse vector in case of extremely sparse matrices

Parameters:

aRow -

yRowOut -

mutlithreadedTRightMultiply

public Vector mutlithreadedTRightMultiply(Vector v)

accumDots

protected void accumDots(int aIndex,
                         double aElement,
                         double[] yRow)

accumDots

protected void accumDots(int aIndex,
                         double aElement,
                         Vector yRow)

murmur64

public static long murmur64(long val,
                            int len,
                            long seed)

Shortened version for data < 8 bytes packed into len lowest bytes of val.

Parameters:

val - the value

len - the length of data packed into this many low bytes of val

seed - the seed to use

Returns:

murmur hash

murmur64

public static long murmur64(byte[] val,
                            int offset,
                            int len,
                            long seed)