com.yahoo.tensor

Class MixedTensor

java.lang.Object

com.yahoo.tensor.MixedTensor

All Implemented Interfaces:

Tensor

public class MixedTensor
extends Object
implements Tensor

A mixed tensor type. This is class is currently suitable for serialization and deserialization, not yet for computation. A mixed tensor has a combination of mapped and indexed dimensions. By reordering the mapped dimensions before the indexed dimensions, one can think of mixed tensors as the mapped dimensions mapping to a dense tensor. This dense tensor is called a dense subspace.

Author:

lesters

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	MixedTensor.BoundBuilder Builder for mixed tensors with bound indexed dimensions.
static class	MixedTensor.Builder Base class for building mixed tensors.
private static class	MixedTensor.Index An immutable index into a list of cells.
static class	MixedTensor.UnboundBuilder Temporarily stores all cells to find bounds of indexed dimensions, then creates a tensor using BoundBuilder.

Nested classes/interfaces inherited from interface com.yahoo.tensor.Tensor

Tensor.Cell

Field Summary

Fields

Modifier and Type	Field and Description
private com.google.common.collect.ImmutableList<Tensor.Cell>	cells The list of cells in the tensor
private MixedTensor.Index	index An index structure over the cell list
private TensorType	type The dimension specification for this tensor

Constructor Summary

Constructors

Modifier	Constructor and Description
private	MixedTensor(TensorType type, com.google.common.collect.ImmutableList<Tensor.Cell> cells, MixedTensor.Index index)

Method Summary

Modifier and Type	Method and Description
Iterator<Tensor.Cell>	cellIterator() Returns an iterator over the cells of this tensor.
Map<TensorAddress,Double>	cells() Returns an immutable map of the cells of this in no particular order.
static TensorType	createPartialType(List<TensorType.Dimension> dimensions)
long	denseSubspaceSize() Returns the size of dense subspaces
boolean	equals(Object other) Returns whether this tensor and the given tensor is mathematically equal: That they have the same dimension *names* and the same content.
double	get(TensorAddress address) Returns the value at the given address
int	hashCode()
long	size() Returns the size of the tensor measured in number of cells
String	toString() Returns this tensor on the form {address1:value1,address2:value2,...} where each address is on the form {dimension1:label1,dimension2:label2,...}, and values are numbers.
TensorType	type() Returns the tensor type
Iterator<Double>	valueIterator() Returns an iterator over the values of this tensor.

Methods inherited from class java.lang.Object

clone, finalize, getClass, notify, notifyAll, wait, wait, wait

Methods inherited from interface com.yahoo.tensor.Tensor

add, approxEqual, approxEquals, approxEquals, argmax, argmin, asDouble, atan2, avg, avg, avg, concat, concat, contentToString, count, count, count, diag, divide, equal, equals, fmod, from, from, from, generate, isEmpty, join, l1Normalize, l2Normalize, larger, largerOrEqual, ldexp, map, matmul, max, max, max, max, min, min, min, min, multiply, notEqual, pow, prod, prod, prod, random, range, reduce, reduce, rename, rename, smaller, smallerOrEqual, softmax, subtract, sum, sum, sum, toStandardString, xwPlusB

Field Detail

type

private final TensorType type

The dimension specification for this tensor

cells

private final com.google.common.collect.ImmutableList<Tensor.Cell> cells

The list of cells in the tensor

index

private final MixedTensor.Index index

An index structure over the cell list

Constructor Detail

MixedTensor

private MixedTensor(TensorType type,
                    com.google.common.collect.ImmutableList<Tensor.Cell> cells,
                    MixedTensor.Index index)

Method Detail

type

public TensorType type()

Returns the tensor type

Specified by:

type in interface Tensor

size

public long size()

Returns the size of the tensor measured in number of cells

Specified by:

size in interface Tensor

get

public double get(TensorAddress address)

Returns the value at the given address

Specified by:

get in interface Tensor

cellIterator

public Iterator<Tensor.Cell> cellIterator()

Returns an iterator over the cells of this tensor. Cells are returned in order of increasing indexes in the indexed dimensions, increasing indexes of later dimensions in the dimension type before earlier. No guarantee is given for the order of sparse dimensions.

Specified by:

cellIterator in interface Tensor

valueIterator

public Iterator<Double> valueIterator()

Returns an iterator over the values of this tensor. The iteration order is the same as for cellIterator.

Specified by:

valueIterator in interface Tensor

cells

public Map<TensorAddress,Double> cells()

Description copied from interface: Tensor

Returns an immutable map of the cells of this in no particular order. This may be expensive for some implementations - avoid when possible

Specified by:

cells in interface Tensor

hashCode

public int hashCode()

Overrides:

hashCode in class Object

toString

public String toString()

Description copied from interface: Tensor

Returns this tensor on the form {address1:value1,address2:value2,...} where each address is on the form {dimension1:label1,dimension2:label2,...}, and values are numbers.

Cells are listed in the natural order of tensor addresses: Increasing size primarily and by element lexical order secondarily.

Note that while this is suggestive of JSON, it is not JSON.

Specified by:

toString in interface Tensor

Overrides:

toString in class Object

equals

public boolean equals(Object other)

Description copied from interface: Tensor

Returns whether this tensor and the given tensor is mathematically equal: That they have the same dimension *names* and the same content.

Specified by:

equals in interface Tensor

Overrides:

equals in class Object

denseSubspaceSize

public long denseSubspaceSize()

Returns the size of dense subspaces

createPartialType

public static TensorType createPartialType(List<TensorType.Dimension> dimensions)