BayesianWorkflows.jl

Bayesian inference workflows for (mainly) Discrete-state-space Partially Observed Markov Process (DPOMP) models in Julia

This package contains tools for Bayesian inference and simulation of DPOMP models. See the docs.

Features

• Simulation and
• Bayesian parameter inference for,
• Discrete-state-space Partially Observed Markov Processes, in Julia.
• Includes automated tools for convergence diagnosis and analysis.

Applications

• Epidemiological modelling (e.g. SEIR models)
• Ecology (e.g. predator-prey dynamics)
• Many other potential use cases, e.g. physics; chemical reactions; social media.

Algorithms

The package implements several different customisable algorithms for Bayesian parameter inference, including:

• Data-augmented MCMC
• Particle filters (i.e. Sequential Monte Carlo)
• Iterative-batch-importance sampling (e.g. 'SMC^2')

Getting started

Package installation

The package is not registered and must be added via the package manager Pkg. From the Julia REPL type ] to enter the Pkg mode, and run:

pkg> add https://github.com/mjb3/BayesianWorkflows.jl

Usage

See the package documentation for instructions and examples.