[DataContract]
public abstract class GenericDiscreteBase<T, TThis> : IDistribution<T>, IDistribution, ICloneable, Diffable, SettableToUniform, HasPoint<T>, CanGetLogProb<T>, SettableTo<TThis>, SettableToProduct<TThis>, SettableToProduct<TThis, TThis>, SettableToRatio<TThis>, SettableToRatio<TThis, TThis>, SettableToPower<TThis>, SettableToWeightedSumExact<TThis>, SettableToWeightedSum<TThis>, SettableToPartialUniform<TThis>, CanGetLogAverageOf<TThis>, CanGetLogAverageOfPower<TThis>, CanGetAverageLog<TThis>, Sampleable<T> where TThis : GenericDiscreteBase<T, TThis>, new()

Type Parameters

Name	Description
T	The domain type of this distribution
TThis	The type of the subclass (allows factory methods return the correct type)

Remarks

This class makes it straightforward to implement discrete distributions over types other than int. It is only necessary to make a subclass of this abstract class, add a constructor and implement two methods to convert the desired type T to and from an integer.

Internally, this class wraps a Discrete distribution and uses it to provide all functionality.

Constructors

GenericDiscreteBase(Int32, Sparsity)

Creates a discrete

Declaration

protected GenericDiscreteBase(int dimension, Sparsity sparsity)

Parameters

Type	Name	Description
Int32	dimension	The dimension of the underlying discrete
Sparsity	sparsity	The sparsity of the underlying discrete

Fields

disc

Declaration

[DataMember]
protected Discrete disc

Field Value

Type	Description
Discrete

Properties

Dimension

The dimension of this discrete distribution.

Declaration

[IgnoreDataMember]
public int Dimension { get; }

Property Value

Type	Description
Int32

IndexRange

Declaration

public IntPair IndexRange { get; }

Property Value

Type	Description
IntPair

IsPointMass

Returns true if the distribution is a point mass.

Declaration

[IgnoreDataMember]
public bool IsPointMass { get; }

Property Value

Type	Description
Boolean

Item[T]

Gets the probability of a given value.

Declaration

public double this[T value] { get; }

Parameters

Type	Name	Description
T	value

Property Value

Type	Description
Double

Point

Point property

Declaration

[IgnoreDataMember]
public T Point { get; set; }

Property Value

Type	Description
T

ValueRange

Returns the domain values corresponding to the first and last indices of non-zero probabilities.

Declaration

public (T, T) ValueRange { get; }

Property Value

Type	Description
(T1, T2)<T, T>

Methods

Clone()

Returns a clone of this distribution.

Declaration

public virtual object Clone()

Returns

Type	Description
Object

ConvertFromInt(Int32)

Converts an integer to an item of type T.

Declaration

public abstract T ConvertFromInt(int i)

Parameters

Type	Name	Description
Int32	i

Returns

Type	Description
T

ConvertToInt(T)

Converts an item of type T to an integer.

Declaration

public abstract int ConvertToInt(T value)

Parameters

Type	Name	Description
T	value

Returns

Type	Description
Int32

Equals(Object)

Override of the Equals method

Declaration

public override bool Equals(object obj)

Parameters

Type	Name	Description
Object	obj	The instance to compare to

Returns

Type	Description
Boolean	True if the two distributions are the same in value, false otherwise

Overrides

Object.Equals(Object)

FromDiscrete(Discrete)

Creates a distribution using the probabilities from the given discrete distribution.

Declaration

public static TThis FromDiscrete(Discrete d)

Parameters

Type	Name	Description
Discrete	d

Returns

Type	Description
TThis

FromProbs(Double[])

Creates a distribution from the given probabilities.

Declaration

public static TThis FromProbs(params double[] probs)

Parameters

Type	Name	Description
Double[]	probs

Returns

Type	Description
TThis

FromVector(Vector)

Creates a distribution from the given vector of probabilities.

Declaration

[Construction(new string[]{"GetWorkspace"})]
public static TThis FromVector(Vector probs)

Parameters

Type	Name	Description
Vector	probs

Returns

Type	Description
TThis

GetAverageLog(TThis)

The expected logarithm of that distribution under this distribution.

Declaration

public double GetAverageLog(TThis that)

Parameters

Type	Name	Description
TThis	that	The distribution to take the logarithm of.

Returns

Type	Description
Double	`sum_x this.Evaluate(x)*Math.Log(that.Evaluate(x))`

Remarks

This is also known as the cross entropy.

GetHashCode()

Override of GetHashCode method

Declaration

public override int GetHashCode()

Returns

Type	Description
Int32	The hash code for this instance

Overrides

Object.GetHashCode()

GetInternalDiscrete()

Gets the internal discrete distribution that this distribution wraps.

Declaration

public Discrete GetInternalDiscrete()

Returns

Type	Description
Discrete	The internal discrete distribution

GetLogAverageOf(TThis)

The log of the integral of the product of this discrete distribution and that discrete distribution

Declaration

public double GetLogAverageOf(TThis that)

Parameters

Type	Name	Description
TThis	that	That discrete distribution

Returns

Type	Description
Double	The log inner product

GetLogAverageOfPower(TThis, Double)

Get the integral of this distribution times another distribution raised to a power.

Declaration

public double GetLogAverageOfPower(TThis that, double power)

Parameters

Type	Name	Description
TThis	that
Double	power

Returns

Type	Description
Double

GetLogProb(T)

Evaluates the log density at the specified domain value

Declaration

public double GetLogProb(T value)

Parameters

Type	Name	Description
T	value	The point at which to evaluate

Returns

Type	Description
Double	The log density

GetLogProbs()

Gets the log-probability at each index.

Declaration

public Vector GetLogProbs()

Returns

Type	Description
Vector	The vector of log-probabilities

GetMode()

Gets the mode of the distribution

Declaration

public T GetMode()

Returns

Type	Description
T	The value with the highest probability

GetProbs()

Gets the probability at each index.

Declaration

public Vector GetProbs()

Returns

Type	Description
Vector	The vector of probabilities

GetWorkspace()

Gets a Vector of size this.Dimension.

Declaration

public Vector GetWorkspace()

Returns

Type	Description
Vector	A pointer to the internal probs Vector of the object.

Remarks

This function is intended to be used with SetProbs, to avoid allocating a new Vector. The return value should not be interpreted as a probs vector, but only a workspace filled with unknown data that can be overwritten. Until SetProbs is called, the distribution object is invalid once this workspace is modified.

IsPartialUniform()

Checks whether the distribution is uniform over its support.

Declaration

public bool IsPartialUniform()

Returns

Type	Description
Boolean	True if the distribution is uniform over its support, false otherwise.

IsUniform()

Returns true if the distribution is uniform.

Declaration

public bool IsUniform()

Returns

Type	Description
Boolean	True if uniform

MaxDiff(Object)

Gets the maximum difference between the parameters of this discrete and that discrete

Declaration

public double MaxDiff(object that)

Parameters

Type	Name	Description
Object	that	That discrete

Returns

Type	Description
Double	The maximum difference

PointMass(T)

Creates a point mass distribution.

Declaration

[Construction(new string[]{"Point"}, UseWhen = "IsPointMass")]
public static TThis PointMass(T value)

Parameters

Type	Name	Description
T	value	The allowed value.

Returns

Type	Description
TThis

Sample()

Returns a sample from the distribution

Declaration

public T Sample()

Returns

Type	Description
T	The sample value

Sample(T)

Returns a sample from the distribution

Declaration

public T Sample(T result)

Parameters

Type	Name	Description
T	result	Not used

Returns

Type	Description
T	The sample value

SetProbs(Vector)

Sets the probability of each index.

Declaration

public void SetProbs(Vector probs)

Parameters

Type	Name	Description
Vector	probs	A vector of non-negative, finite numbers. Need not sum to 1.

Remarks

Instead of allocating your own Vector to pass to SetProbs, you can call GetWorkspace(), fill in the resulting Vector, and then pass it to SetProbs.

SetTo(TThis)

Sets the parameters of this instance to the parameters of that instance

Declaration

public void SetTo(TThis value)

Parameters

Type	Name	Description
TThis	value	That instance

SetToPartialUniform()

Sets the distribution to be uniform over its support.

Declaration

public void SetToPartialUniform()

SetToPartialUniformOf(TThis)

Sets the distribution to be uniform over the support of a given distribution.

Declaration

public void SetToPartialUniformOf(TThis dist)

Parameters

Type	Name	Description
TThis	dist	The distribution which support will be used to setup the current distribution.

SetToPower(TThis, Double)

Sets the parameters to represent the power of a discrete distributions.

Declaration

public void SetToPower(TThis value, double exponent)

Parameters

Type	Name	Description
TThis	value	The discrete distribution
Double	exponent	The exponent

SetToProduct(TThis, TThis)

Sets the parameters to represent the product of two discrete distributions.

Declaration

public void SetToProduct(TThis a, TThis b)

Parameters

Type	Name	Description
TThis	a	The first discrete distribution
TThis	b	The second discrete distribution

SetToRatio(TThis, TThis, Boolean)

Sets the parameters to represent the ratio of two discrete distributions.

Declaration

public void SetToRatio(TThis numerator, TThis denominator, bool forceProper)

Parameters

Type	Name	Description
TThis	numerator	The numerator discrete distribution
TThis	denominator	The denominator discrete distribution
Boolean	forceProper	Whether to force the returned distribution to be proper

SetToSum(Double, TThis, Double, TThis)

Sets the parameters to represent the weighted sum of two discrete distributions.

Declaration

public void SetToSum(double weight1, TThis value1, double weight2, TThis value2)

Parameters

Type	Name	Description
Double	weight1	The first weight
TThis	value1	The first discrete distribution
Double	weight2	The second weight
TThis	value2	The second discrete distribution

SetToUniform()

Sets this instance to a uniform discrete (i.e. probabilities all equal)

Declaration

public void SetToUniform()

ToString()

Declaration

public override string ToString()

Returns

Type	Description
String

Overrides

Object.ToString()

ToString(T)

Implements a custom ToString() for items of type T.

Declaration

protected virtual string ToString(T value)

Parameters

Type	Name	Description
T	value

Returns

Type	Description
String

ToString(String)

Declaration

public virtual string ToString(string format)

Parameters

Type	Name	Description
String	format

Returns

Type	Description
String

ToString(String, String)

Returns a string representation of this distribution.

Declaration

public virtual string ToString(string format, string delimiter)

Parameters

Type	Name	Description
String	format
String	delimiter

Returns

Type	Description
String

Uniform()

Creates a uniform distribution.

Declaration

[Construction(UseWhen = "IsUniform")]
public static TThis Uniform()

Returns

Type	Description
TThis

UniformInRange(T, T)

Creates a Discrete distribution which is uniform over values from start to end inclusive.

Declaration

public static TThis UniformInRange(T start, T end)

Parameters

Type	Name	Description
T	start	The first value included in the distribution
T	end	The last value included in the distribution

Returns

Type	Description
TThis	Discrete which is uniform over the specified range (and zero elsewhere).

UniformInRanges(T[])

Creates a Discrete distribution which is uniform over values in multiple ranges specified by pairs of start and end values. These pairs are specified as adjacent values in an array whose length must therefore be even.

Declaration

public static TThis UniformInRanges(params T[] startEndPairs)

Parameters

Type	Name	Description
T[]	startEndPairs	Sequence of start and end pairs

Returns

Type	Description
TThis	Discrete which is uniform over the specified range (and zero elsewhere).

UniformInRanges(IEnumerable<T>)

Creates a Discrete distribution which is uniform over values in multiple ranges specified by pairs of start and end values. These pairs are specified as adjacent values in an enumerable whose length must therefore be even.

Declaration

public static TThis UniformInRanges(IEnumerable<T> startEndPairs)

Parameters

Type	Name	Description
IEnumerable<T>	startEndPairs	Sequence of start and end pairs

Returns

Type	Description
TThis	Discrete which is uniform over the specified range (and zero elsewhere).

UniformOver(T[])

Creates a Discrete distribution which is uniform over the specified set of values and zero elsewhere.

Declaration

public static TThis UniformOver(params T[] values)

Parameters

Type	Name	Description
T[]	values	The values.

Returns

Type	Description
TThis	Discrete which is uniform over the specified set of values and zero elsewhere.

UniformOver(IEnumerable<T>)

Creates a Discrete distribution which is uniform over the specified set of values and zero elsewhere.

Declaration

public static TThis UniformOver(IEnumerable<T> values)

Parameters

Type	Name	Description
IEnumerable<T>	values	The values.

Returns

Type	Description
TThis	Discrete which is uniform over the specified set of values and zero elsewhere.

Explicit Interface Implementations

CanGetLogProb<T>.GetLogProb(T)

Returns the log probability of the value under this distribution.

Declaration

double CanGetLogProb<T>.GetLogProb(T value)

Parameters

Type	Name	Description
T	value

Returns

Type	Description
Double

Implements

IDistribution<T>

IDistribution

System.ICloneable

Diffable

SettableToUniform

HasPoint<T>

CanGetLogProb<T>

SettableTo<T>

SettableToProduct<T>

SettableToProduct<T, U>

SettableToRatio<T>

SettableToRatio<T, U>

SettableToPower<T>

SettableToWeightedSumExact<T>

SettableToWeightedSum<T>

SettableToPartialUniform<TDist>

CanGetLogAverageOf<T>

CanGetLogAverageOfPower<T>

CanGetAverageLog<T>

Sampleable<T>