Class SubvectorOp

Provides outgoing messages for Subvector(Vector, Int32, Int32), given random arguments to the function.

Inheritance

Object

SubvectorOp

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(Vector), "Subvector", new Type[]{typeof(Vector), typeof(int), typeof(int)})]
[Buffers(new string[]{"SourceMean", "SourceVariance"})]
[Quality(QualityBand.Stable)]
public static class SubvectorOp

Methods

AverageLogFactor()

Evidence message for VMP.

Declaration

public static double AverageLogFactor()

Returns

Type	Description
Double	Zero.

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)). Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

AverageLogFactor(Vector, Vector, Int32)

Evidence message for VMP.

Declaration

public static double AverageLogFactor(Vector subvector, Vector source, int startIndex)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Vector	source	Constant value for source.
Int32	startIndex	Constant value for startIndex.

Returns

Type	Description
Double	Zero.

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)). Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

LogAverageFactor(VectorGaussian, VectorGaussian)

Evidence message for EP.

Declaration

public static double LogAverageFactor(VectorGaussian subvector, VectorGaussian to_subvector)

Parameters

Type	Name	Description
VectorGaussian	subvector	Incoming message from subvector.
VectorGaussian	to_subvector	Outgoing message to subvector.

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_(subvector) p(subvector) factor(subvector,source,startIndex,count)).

LogAverageFactor(Vector, Vector, PositiveDefiniteMatrix, Int32)

Evidence message for EP.

Declaration

public static double LogAverageFactor(Vector subvector, Vector SourceMean, PositiveDefiniteMatrix SourceVariance, int startIndex)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Vector	SourceMean	Buffer SourceMean.
PositiveDefiniteMatrix	SourceVariance	Buffer SourceVariance.
Int32	startIndex	Constant value for startIndex.

Returns

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

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)).

LogAverageFactor(Vector, Vector, Int32)

Evidence message for EP.

Declaration

public static double LogAverageFactor(Vector subvector, Vector source, int startIndex)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Vector	source	Constant value for source.
Int32	startIndex	Constant value for startIndex.

Returns

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

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)).

LogEvidenceRatio(VectorGaussian)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(VectorGaussian subvector)

Parameters

Type	Name	Description
VectorGaussian	subvector	Incoming message from subvector.

Returns

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

Remarks

The formula for the result is log(sum_(subvector) p(subvector) factor(subvector,source,startIndex,count) / sum_subvector p(subvector) messageTo(subvector)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

LogEvidenceRatio(Vector, Vector, PositiveDefiniteMatrix, Int32)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(Vector subvector, Vector SourceMean, PositiveDefiniteMatrix SourceVariance, int startIndex)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Vector	SourceMean	Buffer SourceMean.
PositiveDefiniteMatrix	SourceVariance	Buffer SourceVariance.
Int32	startIndex	Constant value for startIndex.

Returns

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

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

LogEvidenceRatio(Vector, Vector, Int32)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(Vector subvector, Vector source, int startIndex)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Vector	source	Constant value for source.
Int32	startIndex	Constant value for startIndex.

Returns

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

Remarks

The formula for the result is log(factor(subvector,source,startIndex,count)). Adding up these values across all factors and variables gives the log-evidence estimate for EP.

SourceAverageConditional(VectorGaussian, Int32, VectorGaussian)

EP message to source.

Declaration

public static VectorGaussian SourceAverageConditional(VectorGaussian subvector, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
VectorGaussian	subvector	Incoming message from subvector. Must be a proper distribution. If any element is uniform, the result will be uniform.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	subvector is not a proper distribution.

SourceAverageConditional(Vector, Int32, VectorGaussian)

EP message to source.

Declaration

public static VectorGaussian SourceAverageConditional(Vector subvector, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of source conditioned on the given values.

SourceAverageLogarithm(VectorGaussian, Int32, VectorGaussian)

VMP message to source.

Declaration

public static VectorGaussian SourceAverageLogarithm(VectorGaussian subvector, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
VectorGaussian	subvector	Incoming message from subvector. Must be a proper distribution. If any element is uniform, the result will be uniform.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of source with subvector integrated out. The formula is sum_subvector p(subvector) factor(subvector,source,startIndex,count).

Exceptions

Type	Condition
ImproperMessageException	subvector is not a proper distribution.

SourceAverageLogarithm(Vector, Int32, VectorGaussian)

VMP message to source.

Declaration

public static VectorGaussian SourceAverageLogarithm(Vector subvector, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
Vector	subvector	Constant value for subvector.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of source conditioned on the given values.

SourceMean(VectorGaussian, PositiveDefiniteMatrix, Vector)

Update the buffer SourceMean.

Declaration

public static Vector SourceMean(VectorGaussian Source, PositiveDefiniteMatrix SourceVariance, Vector result)

Parameters

Type	Name	Description
VectorGaussian	Source	Incoming message from source. Must be a proper distribution. If any element is uniform, the result will be uniform.
PositiveDefiniteMatrix	SourceVariance	Buffer SourceVariance.
Vector	result	Modified to contain the outgoing message.

Returns

Type	Description
Vector	result

Remarks

Exceptions

Type	Condition
ImproperMessageException	Source is not a proper distribution.

SourceMeanInit(VectorGaussian)

Initialize the buffer SourceMean.

Declaration

public static Vector SourceMeanInit(VectorGaussian Source)

Parameters

Type	Name	Description
VectorGaussian	Source	Incoming message from source.

Returns

Type	Description
Vector	Initial value of buffer SourceMean.

Remarks

SourceVariance(VectorGaussian, PositiveDefiniteMatrix)

Update the buffer SourceVariance.

Declaration

public static PositiveDefiniteMatrix SourceVariance(VectorGaussian Source, PositiveDefiniteMatrix result)

Parameters

Type	Name	Description
VectorGaussian	Source	Incoming message from source. Must be a proper distribution. If any element is uniform, the result will be uniform.
PositiveDefiniteMatrix	result	Modified to contain the outgoing message.

Returns

Type	Description
PositiveDefiniteMatrix	result

Remarks

Exceptions

Type	Condition
ImproperMessageException	Source is not a proper distribution.

SourceVarianceInit(VectorGaussian)

Initialize the buffer SourceVariance.

Declaration

public static PositiveDefiniteMatrix SourceVarianceInit(VectorGaussian Source)

Parameters

Type	Name	Description
VectorGaussian	Source	Incoming message from source.

Returns

Type	Description
PositiveDefiniteMatrix	Initial value of buffer SourceVariance.

Remarks

SubvectorAverageConditional(Vector, PositiveDefiniteMatrix, Int32, VectorGaussian)

EP message to subvector.

Declaration

public static VectorGaussian SubvectorAverageConditional(Vector SourceMean, PositiveDefiniteMatrix SourceVariance, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
Vector	SourceMean	Buffer SourceMean.
PositiveDefiniteMatrix	SourceVariance	Buffer SourceVariance.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of subvector conditioned on the given values.

SubvectorAverageConditionalInit(Int32)

Declaration

public static VectorGaussian SubvectorAverageConditionalInit(int count)

Parameters

Type	Name	Description
Int32	count	Constant value for count.

Returns

Type	Description
VectorGaussian

Remarks

SubvectorAverageLogarithm(Vector, PositiveDefiniteMatrix, Int32, VectorGaussian)

VMP message to subvector.

Declaration

public static VectorGaussian SubvectorAverageLogarithm(Vector SourceMean, PositiveDefiniteMatrix SourceVariance, int startIndex, VectorGaussian result)

Parameters

Type	Name	Description
Vector	SourceMean	Buffer SourceMean.
PositiveDefiniteMatrix	SourceVariance	Buffer SourceVariance.
Int32	startIndex	Constant value for startIndex.
VectorGaussian	result	Modified to contain the outgoing message.

Returns

Type	Description
VectorGaussian	result

Remarks

The outgoing message is the factor viewed as a function of subvector conditioned on the given values.

SubvectorAverageLogarithmInit(Int32)

Declaration

public static VectorGaussian SubvectorAverageLogarithmInit(int count)

Parameters

Type	Name	Description
Int32	count	Constant value for count.

Returns

Type	Description
VectorGaussian

Remarks