gwin.sampler.kombine module

This modules provides classes and functions for using the kombine sampler packages for parameter estimation.

class gwin.sampler.kombine.KombineSampler(model, nwalkers, transd=False, pool=None, model_call=None, update_interval=None)[source]

Bases: gwin.sampler.base.BaseMCMCSampler

This class is used to construct the MCMC sampler from the kombine package.

Parameters:
  • model (model) – A model from gwin.models.
  • nwalkers (int) – Number of walkers to use in sampler.
  • transd (bool) – If True, the sampler will operate across parameter spaces using a kombine.clustered_kde.TransdimensionalKDE proposal distribution. In this mode a masked array with samples in each of the possible sets of dimensions must be given for the initial ensemble distribution.
  • processes ({None, int}) – Number of processes to use with multiprocessing. If None, all available cores are used.
  • update_interval ({None, int}) – Make the sampler update the proposal densities every update_interval iterations.
acceptance_fraction

Get the fraction of steps accepted by each walker as an array.

burn_in()[source]

Use kombine’s burnin routine to advance the sampler.

If a minimum number of burn-in iterations was specified, this will run the burn-in until it has advanced at least as many steps as desired. The initial positions (p0) must be set prior to running.

For more details, see kombine.sampler.burnin.

Returns:
  • p (numpy.array) – An array of current walker positions with shape (nwalkers, ndim).
  • lnpost (numpy.array) – The list of log posterior probabilities for the walkers at positions p, with shape (nwalkers, ndim).
  • lnprop (numpy.array) – The list of log proposal densities for the walkers at positions p, with shape (nwalkers, ndim).
chain

Get all past samples as an nwalker x niterations x ndim array.

clear_chain()[source]

Clears the chain and blobs from memory.

classmethod from_cli(opts, model, pool=None, model_call=None)[source]

Create an instance of this sampler from the given command-line options.

Parameters:
  • opts (ArgumentParser options) – The options to parse.
  • model (Model) – The model to use with the sampler.
Returns:

A kombine sampler initialized based on the given arguments.
Return type:

KombineSampler
lnpost

Get the natural logarithm of the likelihood as an nwalkers x niterations array.

name = 'kombine'
run(niterations, **kwargs)[source]

Advance the sampler for a number of samples.

Parameters:niterations (int) – Number of samples to get from sampler.
Returns:
  • p (numpy.array) – An array of current walker positions with shape (nwalkers, ndim).
  • lnpost (numpy.array) – The list of log posterior probabilities for the walkers at positions p, with shape (nwalkers, ndim).
  • lnprop (numpy.array) – The list of log proposal densities for the walkers at positions p, with shape (nwalkers, ndim).
set_state_from_file(fp)[source]

Sets the state of the sampler back to the instance saved in a file.

In addition to the numpy random state, the current KDE used for the jump proposals is loaded.

Parameters:fp (InferenceFile) – File with sampler state stored.
write_state(fp)[source]

Saves the state of the sampler in a file.

In addition to the numpy random state, the current KDE used for the jump proposals is saved.

Parameters:fp (InferenceFile) – File to store sampler state.