SamplingResults

class liesel.goose.SamplingResults(positions, transition_infos, generated_quantities, tuning_infos, kernel_states, full_model_states, kernel_classes, kernels_by_pos_key)

Bases: object

Contains the results of the MCMC engine.

Easy access to the samples is provided via the methods get_samples() and get_posterior_samples().

Methods

get_adaptation_samples()	Returns a dictionary of adaptation samples for all parameters included in the position.
get_error_log([posterior_only])	Returns the error log that is an dict[kernel_name, KernelErrorLog]
get_kernels_by_pos_key()	Returns a dict, identifying the kernel used to sample each position.
get_pos_keys_by_kernels()	Returns a dict, identifying the position keys governed by each kernel.
get_posterior_acceptance_probabilities()	Returns dictionary of acceptance probabilities during posterior by kernel.
get_posterior_position_moved()	Returns dictionary of transition movements (0: no move, 1: move) during posterior by kernel.
get_posterior_samples()	Returns a dictionary of posterior samples for all parameters included in the position.
get_posterior_transition_infos()	Returns a dictionary of posterior transition information for all parameters included in the position.
get_samples()	Returns a dictionary of all samples for all parameters included in the position.
get_tuning_times()	Returns array of tuning times.
get_warmup_acceptance_probabilities()	Returns dictionary of acceptance probabilities during warmup by kernel.
get_warmup_kernel_states([process_state])	If available, returns a dictionary of kernel states recorded during warmup, organized by kernel.
get_warmup_position_moved()	Returns dictionary of transition movements (0: no move, 1: move) during warmup by kernel.
get_warmup_samples()	Returns a dictionary of adaptation samples for all parameters included in the position.
get_warmup_transition_infos()	Returns a dictionary of posterior transition information for all parameters included in the position.
pkl_load(path)	Loads the pickled object from path.
pkl_save(path)	Save result as a pickled object under path.

Attributes

positions	EpochChainManager giving access to monitored variables.
transition_infos	EpochChainManager storing all transition infos.
generated_quantities	Option[EpochChainManager] storing all generated_quantities.
tuning_infos	Option[Chain] storing all tuning infos.
kernel_states	Option[EpochChainManager] holds all kernel states.
full_model_states	Option[EpochChainManager] holds the full model state of each iteration.
kernel_classes	Optional map of kernel identifier to the respective kernel type.
kernels_by_pos_key	Optional map of position key to identifier of the for sampling responsible kernel.