Class SubarrayOp<T>

Provides outgoing messages for Subarray<T>(IReadOnlyList<T>, IReadOnlyList<Int32>), given random arguments to the function.

Inheritance

Object

SubarrayOp<T>

Inherited Members

Object.Equals(Object)

Object.Equals(Object, Object)

Object.GetHashCode()

Object.GetType()

Object.MemberwiseClone()

Object.ReferenceEquals(Object, Object)

Object.ToString()

Namespace: Microsoft.ML.Probabilistic.Factors

Assembly: Microsoft.ML.Probabilistic.dll

Syntax

[FactorMethod(typeof(Collection), "Subarray<>", new Type[]{})]
[Quality(QualityBand.Mature)]
public static class SubarrayOp<T>

Type Parameters

Name	Description
T	The type of an array item.

Methods

ArrayAverageConditional<DistributionType, ArrayType>(IReadOnlyList<DistributionType>, IReadOnlyList<Int32>, ArrayType)

EP message to array.

Declaration

public static ArrayType ArrayAverageConditional<DistributionType, ArrayType>([SkipIfAllUniform] IReadOnlyList<DistributionType> items, IReadOnlyList<int> indices, ArrayType result)
    where DistributionType : SettableTo<DistributionType> where ArrayType : IList<DistributionType>, SettableToUniform

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	items	Incoming message from items. Must be a proper distribution. If all elements are uniform, the result will be uniform.
IReadOnlyList<Int32>	indices	Constant value for indices.
ArrayType	result	Modified to contain the outgoing message.

Returns

Type	Description
ArrayType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ArrayType	The type of the outgoing message.

Remarks

The outgoing message is a distribution matching the moments of array as the random arguments are varied. The formula is proj[p(array) sum_(items) p(items) factor(items,array,indices)]/p(array).

Exceptions

Type	Condition
ImproperMessageException	items is not a proper distribution.

ArrayAverageConditional<DistributionType, ArrayType>(IReadOnlyList<T>, IReadOnlyList<Int32>, ArrayType)

EP message to array.

Declaration

public static ArrayType ArrayAverageConditional<DistributionType, ArrayType>(IReadOnlyList<T> items, IReadOnlyList<int> indices, ArrayType result)
    where DistributionType : HasPoint<T> where ArrayType : IList<DistributionType>, SettableToUniform

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<Int32>	indices	Constant value for indices.
ArrayType	result	Modified to contain the outgoing message.

Returns

Type	Description
ArrayType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ArrayType	The type of the outgoing message.

Remarks

The outgoing message is a distribution matching the moments of array as the random arguments are varied. The formula is proj[p(array) sum_(items) p(items) factor(items,array,indices)]/p(array).

ArrayAverageLogarithm<DistributionType, ArrayType>(IReadOnlyList<DistributionType>, IReadOnlyList<Int32>, ArrayType)

VMP message to array.

Declaration

public static ArrayType ArrayAverageLogarithm<DistributionType, ArrayType>([SkipIfAllUniform] IReadOnlyList<DistributionType> items, IReadOnlyList<int> indices, ArrayType result)
    where DistributionType : SettableTo<DistributionType> where ArrayType : IList<DistributionType>, SettableToUniform

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	items	Incoming message from items. Must be a proper distribution. If all elements are uniform, the result will be uniform.
IReadOnlyList<Int32>	indices	Constant value for indices.
ArrayType	result	Modified to contain the outgoing message.

Returns

Type	Description
ArrayType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ArrayType	The type of the outgoing message.

Remarks

The outgoing message is the factor viewed as a function of array with items integrated out. The formula is sum_items p(items) factor(items,array,indices).

Exceptions

Type	Condition
ImproperMessageException	items is not a proper distribution.

ArrayAverageLogarithm<DistributionType, ArrayType>(IReadOnlyList<T>, IReadOnlyList<Int32>, ArrayType)

VMP message to array.

Declaration

public static ArrayType ArrayAverageLogarithm<DistributionType, ArrayType>(IReadOnlyList<T> items, IReadOnlyList<int> indices, ArrayType result)
    where DistributionType : HasPoint<T> where ArrayType : IList<DistributionType>, SettableToUniform

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<Int32>	indices	Constant value for indices.
ArrayType	result	Modified to contain the outgoing message.

Returns

Type	Description
ArrayType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ArrayType	The type of the outgoing message.

Remarks

The outgoing message is the factor viewed as a function of array with items integrated out. The formula is sum_items p(items) factor(items,array,indices).

AverageLogFactor()

Evidence message for VMP.

Declaration

public static double AverageLogFactor()

Returns

Type	Description
Double	Zero.

Remarks

The formula for the result is log(factor(items,array,indices)). Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

ItemsAverageConditional<DistributionType, ResultType>(IReadOnlyList<DistributionType>, IReadOnlyList<Int32>, ResultType)

EP message to items.

Declaration

public static ResultType ItemsAverageConditional<DistributionType, ResultType>([SkipIfAllUniform] IReadOnlyList<DistributionType> array, IReadOnlyList<int> indices, ResultType result)
    where DistributionType : SettableTo<DistributionType> where ResultType : IList<DistributionType>

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	array	Incoming message from array. Must be a proper distribution. If all elements are uniform, the result will be uniform.
IReadOnlyList<Int32>	indices	Constant value for indices.
ResultType	result	Modified to contain the outgoing message.

Returns

Type	Description
ResultType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ResultType	The type of the outgoing message.

Remarks

The outgoing message is a distribution matching the moments of items as the random arguments are varied. The formula is proj[p(items) sum_(array) p(array) factor(items,array,indices)]/p(items).

Exceptions

Type	Condition
ImproperMessageException	array is not a proper distribution.

ItemsAverageConditionalInit<TDist>(DistributionRefArray<TDist, T>, IReadOnlyList<Int32>)

Declaration

public static DistributionRefArray<TDist, T> ItemsAverageConditionalInit<TDist>(DistributionRefArray<TDist, T> array, IReadOnlyList<int> indices)
    where TDist : class, SettableTo<TDist>, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToPower<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>, CanGetLogAverageOfPower<TDist>, CanGetAverageLog<TDist>, IDistribution<T>, Sampleable<T>

Parameters

Type	Name	Description
DistributionRefArray<TDist, T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
DistributionRefArray<TDist, T>

Type Parameters

Name	Description
TDist	The type of a distribution over an array item.

Remarks

ItemsAverageConditionalInit<TDist>(DistributionStructArray<TDist, T>, IReadOnlyList<Int32>)

Declaration

public static DistributionStructArray<TDist, T> ItemsAverageConditionalInit<TDist>(DistributionStructArray<TDist, T> array, IReadOnlyList<int> indices)
    where TDist : struct, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToPower<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>, CanGetLogAverageOfPower<TDist>, CanGetAverageLog<TDist>, IDistribution<T>, Sampleable<T>

Parameters

Type	Name	Description
DistributionStructArray<TDist, T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
DistributionStructArray<TDist, T>

Type Parameters

Name	Description
TDist	The type of a distribution over an array item.

Remarks

ItemsAverageLogarithm<DistributionType, ResultType>(IReadOnlyList<DistributionType>, IReadOnlyList<Int32>, ResultType)

VMP message to items.

Declaration

public static ResultType ItemsAverageLogarithm<DistributionType, ResultType>([SkipIfAllUniform] IReadOnlyList<DistributionType> array, IReadOnlyList<int> indices, ResultType result)
    where DistributionType : SettableTo<DistributionType> where ResultType : IList<DistributionType>

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	array	Incoming message from array. Must be a proper distribution. If all elements are uniform, the result will be uniform.
IReadOnlyList<Int32>	indices	Constant value for indices.
ResultType	result	Modified to contain the outgoing message.

Returns

Type	Description
ResultType	result

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.
ResultType	The type of the outgoing message.

Remarks

The outgoing message is a distribution matching the moments of items as the random arguments are varied. The formula is proj[sum_(array) p(array) factor(items,array,indices)].

Exceptions

Type	Condition
ImproperMessageException	array is not a proper distribution.

ItemsAverageLogarithmInit<TDist>(DistributionRefArray<TDist, T>, IReadOnlyList<Int32>)

Declaration

public static DistributionRefArray<TDist, T> ItemsAverageLogarithmInit<TDist>(DistributionRefArray<TDist, T> array, IReadOnlyList<int> indices)
    where TDist : class, SettableTo<TDist>, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToPower<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>, CanGetLogAverageOfPower<TDist>, CanGetAverageLog<TDist>, IDistribution<T>, Sampleable<T>

Parameters

Type	Name	Description
DistributionRefArray<TDist, T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
DistributionRefArray<TDist, T>

Type Parameters

Name	Description
TDist	The type of a distribution over an array item.

Remarks

ItemsAverageLogarithmInit<TDist>(DistributionStructArray<TDist, T>, IReadOnlyList<Int32>)

Declaration

public static DistributionStructArray<TDist, T> ItemsAverageLogarithmInit<TDist>(DistributionStructArray<TDist, T> array, IReadOnlyList<int> indices)
    where TDist : struct, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToPower<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>, CanGetLogAverageOfPower<TDist>, CanGetAverageLog<TDist>, IDistribution<T>, Sampleable<T>

Parameters

Type	Name	Description
DistributionStructArray<TDist, T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
DistributionStructArray<TDist, T>

Type Parameters

Name	Description
TDist	The type of a distribution over an array item.

Remarks

ItemsDeriv<ResultType>(ResultType)

Declaration

public static ResultType ItemsDeriv<ResultType>(ResultType result)
    where ResultType : SettableToUniform

Parameters

Type	Name	Description
ResultType	result

Returns

Type	Description
ResultType

Type Parameters

Name	Description
ResultType

LogAverageFactor(IReadOnlyList<T>, IReadOnlyList<T>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogAverageFactor(IReadOnlyList<T> items, IReadOnlyList<T> array, IReadOnlyList<int> indices)

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's average value across the given argument distributions.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices)).

LogAverageFactor<DistributionType>(IReadOnlyList<DistributionType>, IReadOnlyList<DistributionType>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogAverageFactor<DistributionType>(IReadOnlyList<DistributionType> items, IReadOnlyList<DistributionType> array, IReadOnlyList<int> indices)
    where DistributionType : CanGetLogAverageOf<DistributionType>

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	items	Incoming message from items.
IReadOnlyList<DistributionType>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's average value across the given argument distributions.

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices)).

LogAverageFactor<DistributionType>(IReadOnlyList<DistributionType>, IReadOnlyList<T>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogAverageFactor<DistributionType>(IReadOnlyList<DistributionType> items, IReadOnlyList<T> array, IReadOnlyList<int> indices)
    where DistributionType : CanGetLogProb<T>

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	items	Incoming message from items.
IReadOnlyList<T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's average value across the given argument distributions.

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices)).

LogAverageFactor<DistributionType>(IReadOnlyList<T>, IReadOnlyList<DistributionType>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogAverageFactor<DistributionType>(IReadOnlyList<T> items, IReadOnlyList<DistributionType> array, IReadOnlyList<int> indices)
    where DistributionType : CanGetLogProb<T>

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<DistributionType>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's average value across the given argument distributions.

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices)).

LogEvidenceRatio(IReadOnlyList<T>, IReadOnlyList<T>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(IReadOnlyList<T> items, IReadOnlyList<T> array, IReadOnlyList<int> indices)

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<T>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's contribution the EP model evidence.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices) / sum_items p(items) messageTo(items)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

LogEvidenceRatio<DistributionType>(IReadOnlyList<DistributionType>)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<DistributionType>(IReadOnlyList<DistributionType> items)
    where DistributionType : CanGetLogAverageOf<DistributionType>

Parameters

Type	Name	Description
IReadOnlyList<DistributionType>	items	Incoming message from items.

Returns

Type	Description
Double	Logarithm of the factor's contribution the EP model evidence.

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.

Remarks

The formula for the result is log(sum_(items) p(items) factor(items,array,indices) / sum_items p(items) messageTo(items)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

LogEvidenceRatio<DistributionType>(IReadOnlyList<T>, IReadOnlyList<DistributionType>, IReadOnlyList<Int32>)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<DistributionType>(IReadOnlyList<T> items, IReadOnlyList<DistributionType> array, IReadOnlyList<int> indices)
    where DistributionType : CanGetLogProb<T>

Parameters

Type	Name	Description
IReadOnlyList<T>	items	Incoming message from items.
IReadOnlyList<DistributionType>	array	Incoming message from array.
IReadOnlyList<Int32>	indices	Constant value for indices.

Returns

Type	Description
Double	Logarithm of the factor's contribution the EP model evidence.

Type Parameters

Name	Description
DistributionType	The type of a distribution over an array item.

Remarks

The formula for the result is log(sum_(items,array) p(items,array) factor(items,array,indices) / sum_items p(items) messageTo(items)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.