Crate tensorlogic_quantrs_hooks 

TL <-> QuantrS2 hooks (PGM/message passing as reductions).

Version: 0.1.0-beta.1 | Status: Production Ready

This crate provides integration between TensorLogic and probabilistic graphical models (PGMs). It maps belief propagation and other message passing algorithms onto einsum reduction patterns.

Core Concepts

• Factor Graphs: Convert TLExpr predicates into factors
• Message Passing: Sum-product and max-product algorithms as tensor operations
• Inference: Marginalization and conditional queries via reductions

Architecture

TLExpr → FactorGraph → MessagePassing → Marginals
   ↓         ↓              ↓              ↓
Predicates Factors    Einsum Ops    Probabilities

Re-exports

pub use dbn::CoupledDBN;

pub use dbn::CouplingFactor;

pub use dbn::DBNBuilder;

pub use dbn::DynamicBayesianNetwork;

pub use dbn::TemporalVar;

pub use influence::InfluenceDiagram;

pub use influence::InfluenceDiagramBuilder;

pub use influence::InfluenceNode;

pub use influence::MultiAttributeUtility;

pub use influence::NodeType;

pub use memory::BlockSparseFactor;

pub use memory::CompressedFactor;

pub use memory::FactorPool;

pub use memory::LazyFactor;

pub use memory::MemoryEstimate;

pub use memory::PoolStats;

pub use memory::SparseFactor;

pub use memory::StreamingFactorGraph;

pub use parameter_learning::BaumWelchLearner;

pub use parameter_learning::BayesianEstimator;

pub use parameter_learning::MaximumLikelihoodEstimator;

pub use parameter_learning::SimpleHMM;

pub use quantrs_hooks::AnnealingConfig;

pub use quantrs_hooks::DistributionExport;

pub use quantrs_hooks::DistributionMetadata;

pub use quantrs_hooks::ModelExport;

pub use quantrs_hooks::ModelStatistics;

pub use quantrs_hooks::QuantRSAssignment;

pub use quantrs_hooks::QuantRSDistribution;

pub use quantrs_hooks::QuantRSInferenceQuery;

pub use quantrs_hooks::QuantRSModelExport;

pub use quantrs_hooks::QuantRSParameterLearning;

pub use quantrs_hooks::QuantRSSamplingHook;

pub use quantrs_hooks::QuantumAnnealing;

pub use quantrs_hooks::QuantumInference;

pub use quantrs_hooks::QuantumSolution;

pub use quantrs_hooks::QuantumSolutionMetadata;

pub use quantum_circuit::tlexpr_to_qaoa_circuit;

pub use quantum_circuit::IsingModel;

pub use quantum_circuit::QAOAConfig;

pub use quantum_circuit::QAOAResult;

pub use quantum_circuit::QUBOProblem;

pub use quantum_circuit::QuantumCircuitBuilder;

pub use quantum_simulation::run_qaoa;

pub use quantum_simulation::QuantumSimulationBackend;

pub use quantum_simulation::SimulatedState;

pub use quantum_simulation::SimulationConfig;

pub use tensor_network_bridge::factor_graph_to_tensor_network;

pub use tensor_network_bridge::linear_chain_to_mps;

pub use tensor_network_bridge::MatrixProductState;

pub use tensor_network_bridge::Tensor;

pub use tensor_network_bridge::TensorNetwork;

pub use tensor_network_bridge::TensorNetworkStats;

Modules

dbn

Dynamic Bayesian Networks (DBN) for temporal probabilistic models.

influence

Influence diagrams (decision networks) for decision-making under uncertainty.

memory

Memory optimization utilities for large factor graphs.

parameter_learning

Parameter learning algorithms for probabilistic graphical models.

quantrs_hooks

QuantRS2 integration hooks for probabilistic graphical models.

quantum_circuit

Quantum circuit integration for probabilistic graphical models.

quantum_simulation

Quantum simulation for probabilistic inference.

tensor_network_bridge

Tensor network bridge for quantum-classical hybrid inference.

Structs

BayesianNetwork

Bayesian Network builder.

BetheApproximation

Bethe approximation for structured variational inference.

CacheStats

Cache statistics.

CachedFactor

A cached factor that memoizes operations.

Clique

A clique in the junction tree (a maximal set of connected variables).

ConditionalQuery

Query for conditional probability P(X | Y = y).

ConditionalRandomField

Conditional Random Field builder (discriminative model for structured prediction).

ConvergenceStats

Convergence statistics for belief propagation.

EliminationOrdering

Compute elimination ordering for variable elimination.

EmissionFeature

Emission feature: fires when a specific label is paired with a specific observation.

ExpectationPropagation

Expectation Propagation algorithm for approximate inference.

Factor

A factor in a probabilistic graphical model.

FactorCache

A cache for factor operations.

FactorGraph

Factor graph representation for PGM.

FactorNode

Factor node in a factor graph.

GaussianEP

Gaussian EP for continuous variables with moment matching.

GaussianSite

Gaussian site approximation for continuous variables.

GibbsSampler

Gibbs sampling for approximate inference.

HiddenMarkovModel

Hidden Markov Model builder.

IdentityFeature

Simple identity feature: always returns 1.0

ImportanceSampler

Importance sampling for approximate inference.

InferenceEngine

Inference engine for PGM queries.

JunctionTree

Junction tree structure for exact inference.

JunctionTreeEdge

Edge in the junction tree connecting two cliques.

LikelihoodWeighting

Likelihood weighting for Bayesian networks.

LinearChainCRF

Linear-chain CRF for sequence labeling.

MarginalizationQuery

Query for marginal probability P(X).

MarkovRandomField

Markov Random Field builder (undirected graphical model).

MaxProductAlgorithm

Max-product algorithm (MAP inference).

MeanFieldInference

Mean-field variational inference.

ParallelMaxProduct

Parallel max-product algorithm for MAP inference.

ParallelSumProduct

Parallel sum-product belief propagation.

Particle

Particle for particle filtering.

ParticleFilter

Particle filter (Sequential Monte Carlo) for temporal inference.

Separator

A separator between two cliques (their shared variables).

Site

Site approximation for a single factor.

SumProductAlgorithm

Sum-product algorithm (belief propagation).

TransitionFeature

Transition feature: fires when transitioning from one state to another.

TreeReweightedBP

Tree-reweighted belief propagation (TRW-BP).

VariableElimination

Variable elimination algorithm for exact inference.

VariableNode

Variable node in a factor graph.

WeightedSample

Weighted sample for importance sampling.

Enums

EliminationStrategy

Strategy for computing variable elimination ordering.

FactorOp

Operations on factors.

PgmError

Errors that can occur in PGM operations.

ProposalDistribution

Proposal distribution types for importance sampling.

Traits

FeatureFunction

Feature function for linear-chain CRF.

MessagePassingAlgorithm

Trait for message passing algorithms.

Functions

condition

Compute conditional probability P(X | Y = y).

expr_to_factor_graph

Convert a TensorLogic expression to a factor graph.

marginalize

Compute marginal probability for a variable.

message_passing_reduce

Perform message passing inference on a factor graph.

Type Aliases

Assignment

Assignment of values to variables.

Result

Result type for PGM operations.