jsat.linear

Class VecOps

java.lang.Object

jsat.linear.VecOps

public class VecOps
extends Object

This class provides efficient implementations of use full vector operations and updates. The implementations are efficient for both dense and sparse vectors.

Author:

Edward Raff

Constructor Summary

Constructors

Constructor and Description
VecOps()

Method Summary

Modifier and Type	Method and Description
static double	accumulateSum(Vec w, Vec x, Vec y, Function f) Computes the result of ∑∀ i ∈ |w| wi f(xi-yi)
static double	weightedDot(Vec w, Vec x, Vec y) Computes the weighted dot product of ∑∀ i ∈ |w| w_i x_i y_i

Methods inherited from class java.lang.Object

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

Constructor Detail

VecOps

public VecOps()

Method Detail

accumulateSum

public static double accumulateSum(Vec w,
                                   Vec x,
                                   Vec y,
                                   Function f)

Computes the result of ∑_{∀ i ∈ |w|} w_i f(x_i-y_i)

Parameters:

w - the vector of weight values to multiply on the results

x - the first vector of values in the difference

y - the second vector of values in the difference

f - the single variate function to apply to the difference computed

Returns:

the accumulated sum of the evaluations

weightedDot

public static double weightedDot(Vec w,
                                 Vec x,
                                 Vec y)

Computes the weighted dot product of ∑_{∀ i ∈ |w|} w_i x_i y_i

Parameters:

w - the vector containing the weights, it is assumed to be random access

x - the first vector of the dot product

y - the second vector of the dot product

Returns:

the weighted dot product, which is equivalent to the sum of the products of each index for each vector