Class IndexOfMaximumStochasticOp

Provides outgoing messages for IndexOfMaximumDouble(IList<Double>), given random arguments to the function.

Inheritance

Object

IndexOfMaximumStochasticOp

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(MMath), "IndexOfMaximumDouble", new Type[]{}, Default = true)]
[Quality(QualityBand.Experimental)]
[Buffers(new string[]{"Buffers"})]
public static class IndexOfMaximumStochasticOp

Methods

Buffers<GaussianList>(IndexOfMaximumBuffer[], GaussianList, Discrete)

Update the buffer Buffers.

Declaration

public static IndexOfMaximumBuffer[] Buffers<GaussianList>(IndexOfMaximumBuffer[] Buffers, GaussianList list, Discrete IndexOfMaximumDouble)
    where GaussianList : IList<Gaussian>

Parameters

Type	Name	Description
IndexOfMaximumBuffer[]	Buffers	Buffer Buffers.
GaussianList	list	Incoming message from list. Must be a proper distribution. If any element is uniform, the result will be uniform.
Discrete	IndexOfMaximumDouble	Incoming message from indexOfMaximumDouble.

Returns

Type	Description
IndexOfMaximumBuffer[]	New value of buffer Buffers.

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

Exceptions

Type	Condition
ImproperMessageException	list is not a proper distribution.

BuffersInit<GaussianList>(GaussianList, Discrete)

Initialize the buffer Buffers.

Declaration

public static IndexOfMaximumBuffer[] BuffersInit<GaussianList>(GaussianList list, Discrete IndexOfMaximumDouble)
    where GaussianList : IList<Gaussian>

Parameters

Type	Name	Description
GaussianList	list	Incoming message from list.
Discrete	IndexOfMaximumDouble	Incoming message from indexOfMaximumDouble.

Returns

Type	Description
IndexOfMaximumBuffer[]	Initial value of buffer Buffers.

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

IndexOfMaximumDoubleAverageConditional<GaussianList>(GaussianList, IndexOfMaximumBuffer[])

EP message to indexOfMaximumDouble.

Declaration

public static Discrete IndexOfMaximumDoubleAverageConditional<GaussianList>(GaussianList list, IndexOfMaximumBuffer[] Buffers)
    where GaussianList : DistributionStructArray<Gaussian, double>

Parameters

Type	Name	Description
GaussianList	list	Incoming message from list.
IndexOfMaximumBuffer[]	Buffers	Buffer Buffers.

Returns

Type	Description
Discrete	The outgoing EP message to the indexOfMaximumDouble argument.

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

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

ListAverageConditional<GaussianList>(IndexOfMaximumBuffer[], GaussianList, Discrete, GaussianList)

EP message to list.

Declaration

public static GaussianList ListAverageConditional<GaussianList>(IndexOfMaximumBuffer[] Buffers, GaussianList list, Discrete IndexOfMaximumDouble, GaussianList result)
    where GaussianList : DistributionStructArray<Gaussian, double>

Parameters

Type	Name	Description
IndexOfMaximumBuffer[]	Buffers	Buffer Buffers.
GaussianList	list	Incoming message from list.
Discrete	IndexOfMaximumDouble	Incoming message from indexOfMaximumDouble. Must be a proper distribution. If uniform, the result will be uniform.
GaussianList	result	Modified to contain the outgoing message.

Returns

Type	Description
GaussianList	result

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	IndexOfMaximumDouble is not a proper distribution.

LogAverageFactor<GaussianList>(GaussianList, GaussianList, IndexOfMaximumBuffer[], Discrete)

Evidence message for EP.

Declaration

public static double LogAverageFactor<GaussianList>(GaussianList list, GaussianList to_list, IndexOfMaximumBuffer[] Buffers, Discrete IndexOfMaximumDouble)
    where GaussianList : DistributionStructArray<Gaussian, double>

Parameters

Type	Name	Description
GaussianList	list	Incoming message from list. Must be a proper distribution. If any element is uniform, the result will be uniform.
GaussianList	to_list	Previous outgoing message to list.
IndexOfMaximumBuffer[]	Buffers	Buffer Buffers.
Discrete	IndexOfMaximumDouble	Incoming message from indexOfMaximumDouble.

Returns

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

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

The formula for the result is log(sum_(list,indexOfMaximumDouble) p(list,indexOfMaximumDouble) factor(indexOfMaximumDouble,list)).

Exceptions

Type	Condition
ImproperMessageException	list is not a proper distribution.

LogEvidenceRatio<GaussianList>(GaussianList, GaussianList, Discrete, Discrete, IndexOfMaximumBuffer[])

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<GaussianList>(GaussianList list, GaussianList to_list, Discrete IndexOfMaximumDouble, Discrete to_IndexOfMaximumDouble, IndexOfMaximumBuffer[] Buffers)
    where GaussianList : DistributionStructArray<Gaussian, double>

Parameters

Type	Name	Description
GaussianList	list	Incoming message from list.
GaussianList	to_list	Previous outgoing message to list.
Discrete	IndexOfMaximumDouble	Incoming message from indexOfMaximumDouble.
Discrete	to_IndexOfMaximumDouble	Previous outgoing message to IndexOfMaximumDouble.
IndexOfMaximumBuffer[]	Buffers	Buffer Buffers.

Returns

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

Type Parameters

Name	Description
GaussianList	The type of an incoming message from list.

Remarks

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