[frames] | no frames]

# Module victor_utils

source code

This module contains generic utilities that can be used by any model.

 Functions
double
 dot(u, v) Computes and returns the dot product of given two vectors source code

 grad(w, b, x, y) source code
vector
 scale(w, b) Scales the given vector with the given scale factor and return it source code
vector
 scale_and_add(x, y, s) Scales given vector y and add it to the x and return it source code
double
 dot_ds(dx, sy) Computes and returns dot product of two vectors (at least one of them is sparse) source code
double
 sigma(v) Returns sigma valu source code

 scale_i(x, c) Scales the given vector source code

 scale_and_add_i(x, y, c) Scales y and adds it to x x += y * c source code

 scale_and_add_ds(dx, sz, c) Scales sz (which is sparse) and adds it to x dx += sz * c source code
double
 dot_dss(w, ks, vs) Calculates and returns dot products of w and vs source code

 scale_and_add_dss(w, index, vectors, c) Scales vectors and adds it to w source code

 l2_project(x, B) l2 project function source code

 l1_shrink(x, u) l1 shrink function source code

 l1_shrink_mask(x, u, indexes) l1 shrink function source code

 incremental_average(w_hat, w, steps) Incremental average function source code
 Function Details

### dot(u, v)

source code

Computes and returns the dot product of given two vectors

Parameters:
• `u` (vector) - vector1
• `v` (vector) - vector2
Returns: double
dot product value of the vectors

### scale(w, b)

source code

Scales the given vector with the given scale factor and return it

Parameters:
• `w` (vector) - vector to be scaled
• `b` (double) - scale factor
Returns: vector
scaled vector

source code

Scales given vector y and add it to the x and return it

Parameters:
• `x` (vector) - vector to be added with the scaled one
• `y` (vector) - vector to be scaled
• `s` (double) - scale factor
Returns: vector
x + y * s

### dot_ds(dx, sy)

source code

Computes and returns dot product of two vectors (at least one of them is sparse)

Parameters:
• `dx` (vector) - first component of the dot product
• `sy` (vector) - second component of the dot product
Returns: double
dot product of dx and sy

### sigma(v)

source code

Returns sigma valu

Parameters:
• `v` (double) - v
Returns: double
sigma value of v

### scale_i(x, c)

source code

Scales the given vector

Parameters:
• `x` (vector) - vector to be scaled
• `c` (double) - scale factor

source code

Scales y and adds it to x x += y * c

Parameters:
• `x` (vector) - vector to be modified
• `y` (vector) - vector to be scaled and added to x
• `c` (double) - scale factor

source code

Scales sz (which is sparse) and adds it to x dx += sz * c

Parameters:
• `dx` (vector) - vector to be modified
• `sz` (vector) - vector to be scaled and added to dx
• `c` (double) - scale factor

### dot_dss(w, ks, vs)

source code

Calculates and returns dot products of w and vs

Parameters:
• `w` (vector) - first component of the dot product
• `ks` (vector) - indexes of the sparse vector
• `vs` (vector) - values of the sparse vector
Returns: double
dot product of two vectors

source code

Scales vectors and adds it to w

Parameters:
• `w` (vector) - vector to be modified
• `index` (vector) - indexes of the sparse vector
• `vectors` (vector) - values of the sparse vector
• `c` (double) - scale factor

### l2_project(x, B)

source code

l2 project function

Parameters:
• `x` (vector) - x
• `B` (double) - B

### l1_shrink(x, u)

source code

l1 shrink function

Parameters:
• `x` (vector) - x
• `u` (double) - u

source code

l1 shrink function

Parameters:
• `x` (vector) - x
• `u` (double) - B
• `indexes` (vector) - indexes of the sparse vector

### incremental_average(w_hat, w, steps)

source code

Incremental average function

Parameters:
• `w_hat` (vector) - w hat
• `w` (vector) - model vector
• `steps` (number) - steps

 Generated by Epydoc 3.0.1 on Mon Feb 28 00:56:44 2011 http://epydoc.sourceforge.net