Class GateExitOp<T>

Provides outgoing messages for Exit<T>(Boolean[], T[]), given random arguments to the function.

Inheritance

Object

GateExitOp<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(Gate), "Exit<>", new Type[]{})]
[Quality(QualityBand.Mature)]
public static class GateExitOp<T>

Type Parameters

Name	Description
T	The type of the variable exiting the gate.

Methods

AverageLogFactor<TDist>(TDist, TDist)

Evidence message for VMP.

Declaration

public static double AverageLogFactor<TDist>(TDist exit, TDist to_exit)
    where TDist : IDistribution<T>, CanGetAverageLog<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit. Must be a proper distribution. If uniform, the result will be uniform.
TDist	to_exit	Outgoing message to exit.

Returns

Type	Description
Double	Zero.

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

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	exit is not a proper distribution.

CasesAverageConditional<TDist>(TDist, TDist, Int32)

EP message to cases.

Declaration

public static Bernoulli CasesAverageConditional<TDist>(TDist exit, TDist values, int resultIndex)
    where TDist : IDistribution<T>, CanGetLogAverageOf<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit. Must be a proper distribution. If uniform, the result will be uniform.
TDist	values	Incoming message from values.
Int32	resultIndex	Index of the cases for which a message is desired.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	exit is not a proper distribution.

CasesAverageConditional<TDist>(TDist, IList<T>, Int32)

EP message to cases.

Declaration

public static Bernoulli CasesAverageConditional<TDist>(TDist exit, IList<T> values, int resultIndex)
    where TDist : CanGetLogProb<T>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<T>	values	Incoming message from values.
Int32	resultIndex	Index of the cases for which a message is desired.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

Remarks

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

CasesAverageLogarithm<TDist>(TDist, IList<TDist>, Int32)

VMP message to cases.

Declaration

public static Bernoulli CasesAverageLogarithm<TDist>(TDist exit, [SkipIfAllUniform] IList<TDist> values, int resultIndex)
    where TDist : CanGetAverageLog<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit. Must be a proper distribution. If uniform, the result will be uniform.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
Int32	resultIndex	Index of the cases for which a message is desired.

Returns

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

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

Remarks

The outgoing message is the exponential of the average log-factor value, where the average is over all arguments except cases. Because the factor is deterministic, exit is integrated out before taking the logarithm. The formula is exp(sum_(values) p(values) log(sum_exit p(exit) factor(exit,cases,values))).

Exceptions

Type	Condition
ImproperMessageException	exit is not a proper distribution.
ImproperMessageException	values is not a proper distribution.

ExitAverageConditional<TDist>(TDist, IList<Bernoulli>, IList<TDist>, TDist)

EP message to exit.

Declaration

public static TDist ExitAverageConditional<TDist>(TDist exit, IList<Bernoulli> cases, IList<TDist> values, TDist result)
    where TDist : IDistribution<T>, SettableTo<TDist>, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Bernoulli>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageConditional<TDist>(TDist, IList<Bernoulli>, IList<T>, TDist)

EP message to exit.

Declaration

public static TDist ExitAverageConditional<TDist>(TDist exit, IList<Bernoulli> cases, IList<T> values, TDist result)
    where TDist : ICloneable, HasPoint<T>, SettableToWeightedSum<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Bernoulli>	cases	Incoming message from cases.
IList<T>	values	Incoming message from values.
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 exiting the gate.

Remarks

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

ExitAverageConditional<TDist>(TDist, IList<Boolean>, IList<TDist>, TDist)

EP message to exit.

Declaration

public static TDist ExitAverageConditional<TDist>(TDist exit, IList<bool> cases, IList<TDist> values, TDist result)
    where TDist : SettableTo<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Boolean>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageConditional1<TDist>(TDist, IList<Bernoulli>, IList<TDist>, TDist)

EP message to exit.

Declaration

public static TDist ExitAverageConditional1<TDist>(TDist exit, IList<Bernoulli> cases, IList<TDist> values, TDist result)
    where TDist : IDistribution<T>, SettableTo<TDist>, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToWeightedSum<TDist>, CanGetLogAverageOf<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Bernoulli>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageConditional2<TDist>(TDist, IList<Bernoulli>, IList<TDist>, TDist)

EP message to exit.

Declaration

public static TDist ExitAverageConditional2<TDist>(TDist exit, IList<Bernoulli> cases, [SkipIfAllUniform] IList<TDist> values, TDist result)
    where TDist : SettableTo<TDist>, ICloneable, SettableToProduct<TDist>, SettableToRatio<TDist>, SettableToWeightedSum<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Bernoulli>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageConditionalInit<TDist>(IList<TDist>)

Declaration

public static TDist ExitAverageConditionalInit<TDist>(IList<TDist> values)
    where TDist : ICloneable

Parameters

Type	Name	Description
IList<TDist>	values	Incoming message from values.

Returns

Type	Description
TDist

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

Remarks

ExitAverageLogarithm<TDist>(TDist, IList<Boolean>, IList<TDist>, TDist)

VMP message to exit.

Declaration

public static TDist ExitAverageLogarithm<TDist>(TDist exit, IList<bool> cases, IList<TDist> values, TDist result)
    where TDist : SettableTo<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Boolean>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageLogarithm<TDist>(IList<Bernoulli>, IList<TDist>, TDist)

VMP message to exit.

Declaration

public static TDist ExitAverageLogarithm<TDist>(IList<Bernoulli> cases, [SkipIfAllUniform] IList<TDist> values, TDist result)
    where TDist : ICloneable, SettableToProduct<TDist>, SettableToPower<TDist>, CanGetAverageLog<TDist>, SettableToUniform, SettableTo<TDist>, SettableToRatio<TDist>, SettableToWeightedSum<TDist>

Parameters

Type	Name	Description
IList<Bernoulli>	cases	Incoming message from cases.
IList<TDist>	values	Incoming message from values. Must be a proper distribution. If any element is uniform, the result will be uniform.
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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	values is not a proper distribution.

ExitAverageLogarithmInit<TDist>(IList<TDist>)

Declaration

public static TDist ExitAverageLogarithmInit<TDist>(IList<TDist> values)
    where TDist : ICloneable

Parameters

Type	Name	Description
IList<TDist>	values	Incoming message from values.

Returns

Type	Description
TDist

Type Parameters

Name	Description
TDist	The type of the distribution over the variable exiting the gate.

Remarks

LogEvidenceRatio<TDist>(TDist, TDist)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<TDist>(TDist exit, TDist to_exit)
    where TDist : IDistribution<T>, CanGetLogAverageOf<TDist>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit. Must be a proper distribution. If uniform, the result will be uniform.
TDist	to_exit	Outgoing message to exit.

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 exiting the gate.

Remarks

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

Exceptions

Type	Condition
ImproperMessageException	exit is not a proper distribution.

LogEvidenceRatio<TDist>(TDist, IList<Boolean>)

Evidence message for EP.

Declaration

public static double LogEvidenceRatio<TDist>(TDist exit, IList<bool> cases)
    where TDist : IDistribution<T>

Parameters

Type	Name	Description
TDist	exit	Incoming message from exit.
IList<Boolean>	cases	Incoming message from cases.

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 exiting the gate.

Remarks

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

ValuesAverageConditional<TExit>(TExit, TExit[])

EP message to values.

Declaration

public static TExit[] ValuesAverageConditional<TExit>(TExit exit, TExit[] result)

Parameters

Type	Name	Description
TExit	exit	Incoming message from exit.
TExit[]	result	Modified to contain the outgoing message.

Returns

Type	Description
TExit[]	result

Type Parameters

Name	Description
TExit	The type of the message from exit.

Remarks

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

ValuesAverageConditional<TExit, TResultList>(TExit, TResultList)

EP message to values.

Declaration

public static TResultList ValuesAverageConditional<TExit, TResultList>(TExit exit, TResultList result)
    where TResultList : CanSetAllElementsTo<TExit>

Parameters

Type	Name	Description
TExit	exit	Incoming message from exit.
TResultList	result	Modified to contain the outgoing message.

Returns

Type	Description
TResultList	result

Type Parameters

Name	Description
TExit	The type of the message from exit.
TResultList	The type of the outgoing message.

Remarks

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

ValuesAverageLogarithm<TExit>(TExit, TExit[])

Declaration

public static TExit[] ValuesAverageLogarithm<TExit>(TExit exit, TExit[] result)

Parameters

Type	Name	Description
TExit	exit
TExit[]	result

Returns

Type	Description
TExit[]

Type Parameters

Name	Description
TExit	The type of the message from exit.

ValuesAverageLogarithm<TExit, TResultList>(TExit, TResultList)

VMP message to values.

Declaration

public static TResultList ValuesAverageLogarithm<TExit, TResultList>(TExit exit, TResultList result)
    where TResultList : CanSetAllElementsTo<TExit>

Parameters

Type	Name	Description
TExit	exit	Incoming message from exit.
TResultList	result	Modified to contain the outgoing message.

Returns

Type	Description
TResultList	result

Type Parameters

Name	Description
TExit	The type of the message from exit.
TResultList	The type of the outgoing message.

Remarks

The outgoing message is the factor viewed as a function of values with exit integrated out. The formula is sum_exit p(exit) factor(exit,cases,values).