Class IntCasesOp

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

Inheritance

Object

IntCasesOp

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(new string[]{"Cases", "i", "count"}, typeof(Gate), "CasesInt", new Type[]{typeof(int), typeof(int)})]
[Quality(QualityBand.Mature)]
public static class IntCasesOp

Methods

AverageLogFactor(IList<Bernoulli>, Discrete)

Evidence message for VMP.

Declaration

public static double AverageLogFactor([SkipIfAllUniform] IList<Bernoulli> cases, Discrete i)

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from casesInt. Must be a proper distribution. If all elements are uniform, the result will be uniform.
Discrete	i	Incoming message from i.

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.

Exceptions

Type	Condition
ImproperMessageException	cases is not a proper distribution.

CasesAverageConditional(Discrete, Int32)

EP message to casesInt.

Declaration

public static Bernoulli CasesAverageConditional(Discrete i, int resultIndex)

Parameters

Type	Name	Description
Discrete	i	Incoming message from i.
Int32	resultIndex	Index of the casesInt for which a message is desired.

Returns

Type	Description
Bernoulli	The outgoing EP message to the casesInt argument.

Remarks

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

CasesAverageLogarithm(Discrete, Int32)

VMP message to casesInt.

Declaration

public static Bernoulli CasesAverageLogarithm(Discrete i, int resultIndex)

Parameters

Type	Name	Description
Discrete	i	Incoming message from i.
Int32	resultIndex	Index of the casesInt for which a message is desired.

Returns

Type	Description
Bernoulli	The outgoing VMP message to the casesInt argument.

Remarks

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

CasesAverageLogarithmInit(Discrete)

Declaration

public static DistributionStructArray<Bernoulli, bool> CasesAverageLogarithmInit(Discrete i)

Parameters

Type	Name	Description
Discrete	i	Incoming message from i.

Returns

Type	Description
DistributionStructArray<Bernoulli, Boolean>

Remarks

IAverageConditional(IList<Bernoulli>, Discrete)

EP message to i.

Declaration

public static Discrete IAverageConditional(IList<Bernoulli> cases, Discrete result)

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from casesInt. Must be a proper distribution. If any element is uniform, the result will be uniform.
Discrete	result	Modified to contain the outgoing message.

Returns

Type	Description
Discrete	result

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	cases is not a proper distribution.

IAverageLogarithm(IList<Bernoulli>, Discrete)

VMP message to i.

Declaration

public static Discrete IAverageLogarithm(IList<Bernoulli> cases, Discrete result)

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from casesInt. Must be a proper distribution. If any element is uniform, the result will be uniform.
Discrete	result	Modified to contain the outgoing message.

Returns

Type	Description
Discrete	result

Remarks

The outgoing message is the factor viewed as a function of i with casesInt integrated out. The formula is sum_casesInt p(casesInt) factor(casesInt,i,count).

Exceptions

Type	Condition
ImproperMessageException	cases is not a proper distribution.

LogEvidenceRatio(IList<Bernoulli>, Discrete)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(IList<Bernoulli> cases, Discrete i)

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from casesInt. Must be a proper distribution. If any element is uniform, the result will be uniform.
Discrete	i	Incoming message from i.

Returns

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

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	cases is not a proper distribution.

LogEvidenceRatio(IList<Bernoulli>, Int32)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio(IList<Bernoulli> cases, int i)

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from casesInt.
Int32	i	Constant value for i.

Returns

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

Remarks

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