fglib2.probabilistic_circuits

Module Contents

Classes

SumUnitWrapper
SumUnitFactor	A sum unite (mixture model) that can be used as factor for variables in a factor graph.

class fglib2.probabilistic_circuits.SumUnitWrapper

Bases: probabilistic_model.probabilistic_circuit.DeterministicSumUnit

static from_sum_unit(unit: probabilistic_model.probabilistic_circuit.SmoothSumUnit) → probabilistic_model.probabilistic_circuit.DeterministicSumUnit

_conditional(event: random_events.events.EncodedEvent) → Tuple[typing_extensions.Self | None, float]

class fglib2.probabilistic_circuits.SumUnitFactor(distribution: probabilistic_model.probabilistic_circuit.SmoothSumUnit)

Bases: fglib2.graphs.FactorNode

A sum unite (mixture model) that can be used as factor for variables in a factor graph.

Example use-case:

Imagine you have a set of variables that expand over some template, e.g. time. You learn a mixture for each time step and then use the latent variable interpretation of the mixture model to create a factor graph. The factors for the transition model are multinomial distributions over the latent variables. The factors for the emission model are the joint probability trees.

property variables: List[random_events.variables.Discrete]

latent_variable: random_events.variables.Symbolic

marginal(variables: List[random_events.variables.Variable]) → fglib2.distributions.Multinomial | typing_extensions.Self

latent_distribution()

sum_product(messages: List[fglib2.distributions.Multinomial]) → fglib2.distributions.Multinomial

Apply the sum product algorithm of a factor node.

The product of all incoming messages is calculated and multiplied with the distribution of this factor node.

Parameters:

messages – The incoming messages.

Returns:

The resulting, multivariate distribution.