liesel.goose

Goose MCMC framework.

Functions

history_to_df(history)	Turns a OptimResult.history dictionary into a pandas.DataFrame.
plot_cor(results[, params, param_indices, ...])	Visualizes autocorrelations of posterior samples.
plot_density(results[, params, ...])	Visualizes posterior distributions with a density plot.
plot_param(results, param[, param_index, ...])	Visualizes trace plot, density plot and autocorrelation plot of a single subparameter.
plot_trace(results[, params, param_indices, ...])	Visualizes posterior samples over time with a trace plot.
plot_scatter(results, params, param_indices)	Produces a scatterplot of two parameters.
plot_pairs(results[, params, param_indices, ...])	Produces a pairplot panel.
stan_epochs([warmup_duration, ...])	Sets up a list of EpochConfig's.
optim_flat(model_train, params[, optimizer, ...])	Optimize the parameters of a Liesel Model.

Classes

DictInterface(log_prob_fn)	A model interface for a model state represented by a dict[str, Array] and a corresponding log-probability function.
DataclassInterface(log_prob_fn)	A model interface for a model state represented by a dataclass and a corresponding log-probability function.
LieselInterface(model)	A ModelInterface for a Liesel Model.
Engine(seeds, model_states, kernel_sequence, ...)	MCMC engine capable of combining multiple transition kernels.
EngineBuilder(seed, num_chains)	The EngineBuilder is used to construct an MCMC Engine.
MCMCSpec(kernel[, kernel_kwargs, ...])	Specification for the MCMC kernel and optional jitter distribution associated with a model variable.
LieselMCMC(model[, which])	Manages the setup of MCMC specifications for a Liesel model.
EpochConfig(type, duration, thinning, optional)	Defines an Epoch in an MCMC algorithm.
EpochType(*values)	Indicates which MCMC phase the epoch is part of.
GibbsKernel(position_keys, transition_fn[, ...])	A Gibbs kernel implementing the Kernel protocol.
HMCKernel(position_keys[, ...])	A HMC kernel with dual averaging and an inverse mass matrix tuner, implementing the Kernel protocol.
IWLSKernel(position_keys[, chol_info_fn, ...])	An IWLS kernel with dual averaging and an (optional) user-defined function for computing the Cholesky decomposition of the Fisher information matrix, implementing the liesel.goose.types.Kernel protocol.
MHKernel(position_keys, proposal_fn[, ...])	A Metropolis-Hastings kernel implementing the Kernel protocol.
MHProposal(position, log_correction)
ModelInterface(*args, **kwargs)	Defines a standardized way for Goose to communicate with a statistical model.
NamedTupleInterface(log_prob_fn)	A model interface for a model state represented by a NamedTuple and a corresponding log-probability function.
NUTSKernel(position_keys[, ...])	A NUTS kernel with dual averaging and an inverse mass matrix tuner, implementing the Kernel protocol.
Stopper(max_iter, patience[, atol, rtol])	Handles (early) stopping for optim_flat().
RWKernel(position_keys[, initial_step_size, ...])	A random walk kernel.
Summary(results[, additional_chain, ...])	A summary object.
SamplesSummary(samples[, quantiles, ...])	A summary object based on a dictionary of samples.
SamplingResults(positions, transition_infos, ...)	Contains the results of the MCMC engine.
OptimResult(model_state, position, ...)	Holds the results of model optimization with optim_flat().