edu.stanford.nlp.loglinear.model

Class NDArray<T>

java.lang.Object

edu.stanford.nlp.loglinear.model.NDArray<T>

All Implemented Interfaces:

java.lang.Iterable<int[]>

Direct Known Subclasses:

ConcatVectorTable

public class NDArray<T>
extends java.lang.Object
implements java.lang.Iterable<int[]>

Created on 8/12/15.

Author:

keenon

Holds and provides access to an N-dimensional array.

Yes, generics will lead to unfortunate boxing and unboxing in the TableFactor case, we'll handle that if it becomes a problem.

Constructor Summary

Constructors

Constructor and Description
NDArray(int[] dimensions) Constructor takes a list of neighbor variables to use for this factor.

Method Summary

Modifier and Type	Method and Description
NDArray<T>	cloneArray() Clones the table, but keeps the values by reference.
int	combinatorialNeighborStatesCount()
java.util.Iterator<int[]>	fastPassByReferenceIterator() This is its own function because people will inevitably attempt this optimization of not cloning the array we hand to the iterator, to save on GC, and it should not be default behavior.
T	getAssignmentValue(int[] assignment) Retrieve a single value for an assignment.
int[]	getDimensions()
java.util.Iterator<int[]>	iterator() WARNING: This is pass by reference to avoid massive GC overload during heavy iterations, and because the standard use case is to use the assignments array as an accessor.
void	setAssignmentValue(int[] assignment, T value) Set a single value in the factor table.

Methods inherited from class java.lang.Object

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

Methods inherited from interface java.lang.Iterable

forEach, spliterator

Constructor Detail

NDArray

public NDArray(int[] dimensions)

Constructor takes a list of neighbor variables to use for this factor. This must not change after construction, and the number of states of those variables must also not change.

Parameters:

dimensions - list of neighbor variables assignment range sizes

Method Detail

setAssignmentValue

public void setAssignmentValue(int[] assignment,
                               T value)

Set a single value in the factor table.

Parameters:

assignment - a list of variable settings, in the same order as the neighbors array of the factor

value - the value to put into the factor table

getAssignmentValue

public T getAssignmentValue(int[] assignment)

Retrieve a single value for an assignment.

Parameters:

assignment - a list of variable settings, in the same order as the neighbors array of the factor

Returns:

the value for the given assignment. Can be null if not been set yet.

getDimensions

public int[] getDimensions()

Returns:

the size array of the neighbors of the feature factor, passed by value to ensure immutability.

iterator

public java.util.Iterator<int[]> iterator()

WARNING: This is pass by reference to avoid massive GC overload during heavy iterations, and because the standard use case is to use the assignments array as an accessor. Please, clone if you save a copy, otherwise the array will mutate underneath you.

Specified by:

iterator in interface java.lang.Iterable<int[]>

Returns:

an iterator over all possible assignments to this factor

fastPassByReferenceIterator

public java.util.Iterator<int[]> fastPassByReferenceIterator()

This is its own function because people will inevitably attempt this optimization of not cloning the array we hand to the iterator, to save on GC, and it should not be default behavior. If you know what you're doing, then this may be the iterator for you.

Returns:

an iterator that will mutate the value it returns to you, so you must clone if you want to keep a copy

combinatorialNeighborStatesCount

public int combinatorialNeighborStatesCount()

Returns:

the total number of states this factor must represent to include all neighbors.

cloneArray

public NDArray<T> cloneArray()

Clones the table, but keeps the values by reference.

Returns:

a new NDArray, a perfect replica of this one