Expand description
General Markov Random Fields (graph + Ising), Gibbs sampler, loopy belief propagation.
Re-exports§
pub use belief_prop::BpConfig;pub use belief_prop::BpResult;pub use belief_prop::loopy_bp_map;pub use belief_prop::loopy_bp_marginals;pub use gibbs::GibbsConfig;pub use gibbs::ising_gibbs;pub use junction_tree::Clique;pub use junction_tree::JunctionTree;pub use junction_tree::JunctionTreeConfig;pub use mrf::IsingModel;pub use mrf::Mrf;pub use swendsen_wang::SwendsenWang;pub use swendsen_wang::SwendsenWangConfig;pub use wolff::Wolff;pub use wolff::WolffConfig;
Modules§
- belief_
prop - Loopy belief propagation on a pairwise MRF (sum-product + max-product).
- gibbs
- Gibbs sampling on the Ising MRF with optional simulated annealing.
- junction_
tree - Junction-tree (clique-tree) exact inference for discrete pairwise/higher-order Markov Random Fields.
- mrf
- Pairwise MRF on a generic graph plus an Ising-model specialisation.
- swendsen_
wang - Swendsen-Wang (1987) full-lattice cluster Monte-Carlo update for the Ising model.
- wolff
- Wolff (1989) single-cluster Monte-Carlo update for the ferromagnetic Ising model.