Class CopyOp<T>

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

Inheritance

Object

CopyOp<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(Clone), "Copy<>", new Type[]{})]
[Quality(QualityBand.Mature)]
public static class CopyOp<T>

Type Parameters

Name	Description
T	The type of the variable being copied.

Methods

AverageLogFactor(T, T)

Evidence message for VMP.

Declaration

public static double AverageLogFactor(T copy, T Value)

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
T	Value	Incoming message from value.

Returns

Type	Description
Double	Zero.

Remarks

In Variational Message Passing, the evidence contribution of a deterministic factor is zero. Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

AverageLogFactor(TruncatedGaussian, Gaussian, Gaussian)

Evidence message for VMP.

Declaration

[Quality(QualityBand.Preview)]
public static double AverageLogFactor(TruncatedGaussian copy, Gaussian value, Gaussian to_value)

Parameters

Type	Name	Description
TruncatedGaussian	copy	Incoming message from copy.
Gaussian	value	Incoming message from value.
Gaussian	to_value	Previous outgoing message to value.

Returns

Type	Description
Double	Zero.

Remarks

In Variational Message Passing, the evidence contribution of a deterministic factor is zero. Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

AverageLogFactor<TDist>(TDist, TDist)

Evidence message for VMP.

Declaration

public static double AverageLogFactor<TDist>(TDist copy, TDist Value)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.
TDist	Value	Incoming message from value.

Returns

Type	Description
Double	Zero.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

In Variational Message Passing, the evidence contribution of a deterministic factor is zero. Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

AverageLogFactor<TDist>(TDist, T)

Evidence message for VMP.

Declaration

public static double AverageLogFactor<TDist>(TDist copy, T Value)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.
T	Value	Incoming message from value.

Returns

Type	Description
Double	Zero.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

In Variational Message Passing, the evidence contribution of a deterministic factor is zero. Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

AverageLogFactor<TDist>(T, TDist)

Evidence message for VMP.

Declaration

public static double AverageLogFactor<TDist>(T copy, TDist Value)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
TDist	Value	Incoming message from value.

Returns

Type	Description
Double	Zero.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

In Variational Message Passing, the evidence contribution of a deterministic factor is zero. Adding up these values across all factors and variables gives the log-evidence estimate for VMP.

CopyAverageConditional(T)

EP message to copy.

Declaration

public static T CopyAverageConditional(T Value)

Parameters

Type	Name	Description
T	Value	Incoming message from value.

Returns

Type	Description
T	The outgoing EP message to the copy argument.

Remarks

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

CopyAverageConditional(TruncatedGamma)

EP message to copy.

Declaration

[Quality(QualityBand.Preview)]
public static Gamma CopyAverageConditional(TruncatedGamma value)

Parameters

Type	Name	Description
TruncatedGamma	value	Incoming message from value.

Returns

Type	Description
Gamma	The outgoing EP message to the copy argument.

Remarks

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

CopyAverageConditional(TruncatedGaussian)

EP message to copy.

Declaration

[Quality(QualityBand.Preview)]
public static Gaussian CopyAverageConditional(TruncatedGaussian value)

Parameters

Type	Name	Description
TruncatedGaussian	value	Incoming message from value.

Returns

Type	Description
Gaussian	The outgoing EP message to the copy argument.

Remarks

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

CopyAverageConditional<TDist>(TDist)

EP message to copy.

Declaration

public static TDist CopyAverageConditional<TDist>(TDist Value)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	Value	Incoming message from value.

Returns

Type	Description
TDist	The outgoing EP message to the copy argument.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

CopyAverageLogarithm(Gaussian, Gaussian)

VMP message to copy.

Declaration

[Quality(QualityBand.Preview)]
public static TruncatedGaussian CopyAverageLogarithm(Gaussian value, Gaussian to_value)

Parameters

Type	Name	Description
Gaussian	value	Incoming message from value. Must be a proper distribution. If uniform, the result will be uniform.
Gaussian	to_value	Previous outgoing message to value.

Returns

Type	Description
TruncatedGaussian	The outgoing VMP message to the copy argument.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	value is not a proper distribution.

CopyAverageLogarithm(Gaussian, NonconjugateGaussian, Gaussian)

VMP message to copy.

Declaration

public static NonconjugateGaussian CopyAverageLogarithm(Gaussian value, NonconjugateGaussian copy, Gaussian to_value)

Parameters

Type	Name	Description
Gaussian	value	Incoming message from value.
NonconjugateGaussian	copy	Incoming message from copy.
Gaussian	to_value	Previous outgoing message to value.

Returns

Type	Description
NonconjugateGaussian	The outgoing VMP message to the copy argument.

Remarks

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

CopyAverageLogarithm(TruncatedGaussian)

VMP message to copy.

Declaration

[Quality(QualityBand.Preview)]
public static Gaussian CopyAverageLogarithm(TruncatedGaussian value)

Parameters

Type	Name	Description
TruncatedGaussian	value	Incoming message from value.

Returns

Type	Description
Gaussian	The outgoing VMP message to the copy argument.

Remarks

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

CopyAverageLogarithm<TDist>(TDist)

VMP message to copy.

Declaration

public static TDist CopyAverageLogarithm<TDist>(TDist Value)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	Value	Incoming message from value.

Returns

Type	Description
TDist	The outgoing VMP message to the copy argument.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

CopyDeriv<TDist>(TDist)

Declaration

public static TDist CopyDeriv<TDist>(TDist result)
    where TDist : IDistribution<T>, SettableToUniform

Parameters

Type	Name	Description
TDist	result

Returns

Type	Description
TDist

Type Parameters

Name	Description
TDist

LogAverageFactor(T, T)

Evidence message for EP.

Declaration

public static double LogAverageFactor(T copy, T value)

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
T	value	Incoming message from value.

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_(copy,value) p(copy,value) factor(copy,value)).

LogAverageFactor<TDist>(TDist, TDist)

Evidence message for EP.

Declaration

public static double LogAverageFactor<TDist>(TDist copy, TDist value)
    where TDist : CanGetLogAverageOf<TDist>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.
TDist	value	Incoming message from value.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

The formula for the result is log(sum_(copy,value) p(copy,value) factor(copy,value)).

LogAverageFactor<TDist>(TDist, T)

Evidence message for EP.

Declaration

public static double LogAverageFactor<TDist>(TDist copy, T value)
    where TDist : CanGetLogProb<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.
T	value	Incoming message from value.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

The formula for the result is log(sum_(copy,value) p(copy,value) factor(copy,value)).

LogAverageFactor<TDist>(T, TDist)

Evidence message for EP.

Declaration

public static double LogAverageFactor<TDist>(T copy, TDist value)
    where TDist : CanGetLogProb<T>

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
TDist	value	Incoming message from value.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

The formula for the result is log(sum_(copy,value) p(copy,value) factor(copy,value)).

LogEvidenceRatio(T, T)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(T copy, T value)

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
T	value	Incoming message from value.

Returns

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

Remarks

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

LogEvidenceRatio<TDist>(TDist)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<TDist>(TDist copy)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

LogEvidenceRatio<TDist>(T, TDist)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<TDist>(T copy, TDist value)
    where TDist : CanGetLogProb<T>

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
TDist	value	Incoming message from value.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

ValueAverageConditional(T)

EP message to value.

Declaration

public static T ValueAverageConditional(T copy)

Parameters

Type	Name	Description
T	copy	Incoming message from copy.

Returns

Type	Description
T	The outgoing EP message to the value argument.

Remarks

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

ValueAverageConditional(Gamma)

Declaration

[Quality(QualityBand.Preview)]
public static TruncatedGamma ValueAverageConditional(Gamma copy)

Parameters

Type	Name	Description
Gamma	copy

Returns

Type	Description
TruncatedGamma

ValueAverageConditional(Gaussian)

Declaration

[Quality(QualityBand.Preview)]
public static TruncatedGaussian ValueAverageConditional(Gaussian copy)

Parameters

Type	Name	Description
Gaussian	copy

Returns

Type	Description
TruncatedGaussian

ValueAverageConditional<TDist>(TDist)

EP message to value.

Declaration

public static TDist ValueAverageConditional<TDist>(TDist copy)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.

Returns

Type	Description
TDist	The outgoing EP message to the value argument.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

ValueAverageConditional<TDist>(T, TDist)

EP message to value.

Declaration

public static TDist ValueAverageConditional<TDist>(T copy, TDist result)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
TDist	result	Modified to contain the outgoing message.

Returns

Type	Description
TDist	result

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

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

ValueAverageLogarithm(Gaussian)

VMP message to value.

Declaration

[Quality(QualityBand.Preview)]
public static TruncatedGaussian ValueAverageLogarithm(Gaussian copy)

Parameters

Type	Name	Description
Gaussian	copy	Incoming message from copy.

Returns

Type	Description
TruncatedGaussian	The outgoing VMP message to the value argument.

Remarks

The outgoing message is the factor viewed as a function of value with copy integrated out. The formula is sum_copy p(copy) factor(copy,value).

ValueAverageLogarithm(NonconjugateGaussian, Gaussian, Gaussian)

VMP message to value.

Declaration

public static Gaussian ValueAverageLogarithm(NonconjugateGaussian copy, Gaussian value, Gaussian to_value)

Parameters

Type	Name	Description
NonconjugateGaussian	copy	Incoming message from copy.
Gaussian	value	Incoming message from value.
Gaussian	to_value	Previous outgoing message to value.

Returns

Type	Description
Gaussian	The outgoing VMP message to the value argument.

Remarks

The outgoing message is the factor viewed as a function of value with copy integrated out. The formula is sum_copy p(copy) factor(copy,value).

ValueAverageLogarithm(TruncatedGaussian, Gaussian, Gaussian)

VMP message to value.

Declaration

[Quality(QualityBand.Preview)]
public static Gaussian ValueAverageLogarithm(TruncatedGaussian copy, Gaussian value, Gaussian to_value)

Parameters

Type	Name	Description
TruncatedGaussian	copy	Incoming message from copy.
Gaussian	value	Incoming message from value. Must be a proper distribution. If uniform, the result will be uniform.
Gaussian	to_value	Previous outgoing message to value.

Returns

Type	Description
Gaussian	The outgoing VMP message to the value argument.

Remarks

The outgoing message is the factor viewed as a function of value with copy integrated out. The formula is sum_copy p(copy) factor(copy,value).

Exceptions

Type	Condition
ImproperMessageException	value is not a proper distribution.

ValueAverageLogarithm<TDist>(TDist)

VMP message to value.

Declaration

public static TDist ValueAverageLogarithm<TDist>(TDist copy)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	copy	Incoming message from copy.

Returns

Type	Description
TDist	The outgoing VMP message to the value argument.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

The outgoing message is the factor viewed as a function of value with copy integrated out. The formula is sum_copy p(copy) factor(copy,value).

ValueAverageLogarithm<TDist>(T, TDist)

VMP message to value.

Declaration

public static TDist ValueAverageLogarithm<TDist>(T copy, TDist result)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
T	copy	Incoming message from copy.
TDist	result	Modified to contain the outgoing message.

Returns

Type	Description
TDist	result

Type Parameters

Name	Description
TDist	The type of the distribution over the variable being copied.

Remarks

The outgoing message is the factor viewed as a function of value with copy integrated out. The formula is sum_copy p(copy) factor(copy,value).