LieselMCMC

Contents

LieselMCMC#

class liesel.goose.LieselMCMC(model, which=None)[source]#

Bases: object

Manages the setup of MCMC specifications for a Liesel model.

Parameters:
  • model (Model) – The Liesel model object containing the variables and their inference specifications.

  • which (str | None, default: None) – A named inference configuration to use. If None, the default inference attached to each variable is used.

Examples

Liesel Workflow

For this example, we import tensorflow_probability as follows:

>>> import tensorflow_probability.substrates.jax.distributions as tfd

First, we set up a minimal model:

>>> mu = lsl.Var.new_param(0.0, name="mu", inference=gs.MCMCSpec(gs.NUTSKernel))
>>> dist = lsl.Dist(tfd.Normal, loc=mu, scale=1.0)
>>> y = lsl.Var.new_obs(jnp.array([1.0, 2.0, 3.0]), dist, name="y")
>>> model = lsl.Model([y])

Now we run MCMC:

>>> results = gs.LieselMCMC(model).run_for_epochs(
...     seed=1, num_chains=4, adaptation=250, posterior=100
... )

The function returns a SamplingResults object.

More control

For additional control, we initialize an EngineBuilder and continue from there.

>>> builder = gs.LieselMCMC(model).get_engine_builder(seed=1, num_chains=4)
>>> builder.add_adaptation(1000)
>>> builder.add_posterior(1000)

Methods

get_engine_builder(seed, num_chains[, ...])

Create and configure an EngineBuilder for MCMC sampling.

get_jitter_functions()

Collect jitter functions for model variables that define a jitter distribution.

get_kernel_groups()

Collect and organize model variables into kernel groups.

get_kernel_list()

Construct the list of MCMC kernels from kernel groups.

get_spec(var)

Retrieve the MCMC specification for a given variable.

run_for_epochs(*, seed, num_chains, ...[, ...])

Shorthand method for quickly running MCMC for a set number of epochs.

validate_inference_specs()

Logs a warning if there are any parameters in the model that have no inference specification for MCMC.

Attributes

which

Parameter names included in the MCMC setup.

model

Model whose variables provide the MCMC specifications.