GitHub

Public Documentation

Documentation for EpiInference.jl's public interface.

See the Internals section of the manual for internal package docs covering all submodules.

Contents

Index

Public API

EpiAware.EpiInference.ManyPathfinder — Type
struct ManyPathfinder <: AbstractEpiOptMethod

A variational inference method that runs manypathfinder.

Fields

  • ndraws::Int64: Number of draws per pathfinder run.

  • nruns::Int64: Number of many pathfinder runs.

  • maxiters::Int64: Maximum number of optimization iterations for each run.

  • max_tries::Int64: Maximum number of tries if all runs fail.

source
EpiAware.EpiInference.NUTSampler — Type
struct NUTSampler{A<:ADTypes.AbstractADType, E<:AbstractMCMC.AbstractMCMCEnsemble, M} <: AbstractEpiSamplingMethod

A NUTS method for sampling from a DynamicPPL.Model object.

The NUTSampler struct represents using the No-U-Turn Sampler (NUTS) to sample from the distribution defined by a DynamicPPL.Model.

Fields

  • target_acceptance::Float64: The target acceptance rate for the sampler.

  • adtype::ADTypes.AbstractADType: The automatic differentiation type used for computing gradients.

  • mcmc_parallel::AbstractMCMC.AbstractMCMCEnsemble: The parallelization strategy for the MCMC sampler.

  • nchains::Int64: The number of MCMC chains to run.

  • max_depth::Int64: Tree depth limit for the NUTS sampler.

  • Δ_max::Float64: Divergence threshold for the NUTS sampler.

  • init_ϵ::Float64: The initial step size for the NUTS sampler.

  • ndraws::Int64: The number of samples to draw from each chain.

  • metricT::Any: The metric type to use for the HMC sampler.

  • nadapts::Int64: number of adaptation steps

source
EpiAware.EpiInference.manypathfinder — Method
manypathfinder(
    mdl::DynamicPPL.Model,
    ndraws;
    nruns,
    maxiters,
    max_tries,
    kwargs...
) -> Any

Run multiple instances of the pathfinder algorithm and returns the pathfinder run with the largest ELBO estimate.

Arguments

  • mdl::DynamicPPL.Model: The model to perform inference on.
  • nruns::Int: The number of pathfinder runs to perform.
  • ndraws::Int: The number of draws per pathfinder run, readjusted to be at least as large as the number of chains.
  • nchains::Int: The number of chains that will be initialised by pathfinder draws.
  • maxiters::Int: The maximum number of optimizer iterations per pathfinder run.
  • max_tries::Int: The maximum number of extra tries to find a valid pathfinder result.
  • kwargs...: Additional keyword arguments passed to pathfinder.

Returns

  • best_pfs::PathfinderResult: Best pathfinder result by estimated ELBO.
source