Source code for pyro.infer.reparam.structured

# Copyright Contributors to the Pyro project.
# SPDX-License-Identifier: Apache-2.0

from contextlib import ExitStack

import pyro.distributions as dist
from pyro import poutine
from pyro.infer.autoguide.structured import AutoStructured
from pyro.poutine.plate_messenger import block_plate

from .reparam import Reparam

[docs]class StructuredReparam(Reparam): """ Preconditioning reparameterizer of multiple latent variables. This uses a trained :class:`~pyro.infer.autoguide.AutoStructured` guide to alter the geometry of a model, typically for use e.g. in MCMC. Example usage:: # Step 1. Train a guide guide = AutoStructured(model, ...) svi = SVI(model, guide, ...) # ...train the guide... # Step 2. Use trained guide in preconditioned MCMC model = StructuredReparam(guide).reparam(model) nuts = NUTS(model) # use the model in HMC or NUTS... This reparameterization works only for latent variables, not likelihoods. Note that all sites must share a single common :class:`StructuredReparam` instance, and that the model must have static structure. .. note:: This can be seen as a restricted structured version of :class:`~pyro.infer.reparam.neutra.NeuTraReparam` [1] combined with ``poutine.condition`` on MAP-estimated sites (the NeuTra transform is an exact reparameterizer, but the conditioning to point estimates introduces model approximation). [1] Hoffman, M. et al. (2019) "NeuTra-lizing Bad Geometry in Hamiltonian Monte Carlo Using Neural Transport" :param ~pyro.infer.autoguide.AutoStructured guide: A trained guide. """ def __init__(self, guide: AutoStructured): if not isinstance(guide, AutoStructured): raise TypeError( f"StructuredReparam expected an AutoStructured guide, but got {type(guide)}" ) = guide.requires_grad_(False) self.deltas = {} def _reparam_config(self, site): if site["name"] in return self
[docs] def reparam(self, fn=None): return poutine.reparam(fn, config=self._reparam_config)
[docs] def apply(self, msg): name = msg["name"] fn = msg["fn"] value = msg["value"] is_observed = msg["is_observed"] if name not in return {"fn": fn, "value": value, "is_observed": is_observed} if is_observed: raise NotImplementedError( f"At pyro.sample({repr(name)},...), " "StructuredReparam does not support observe statements" ) if name not in self.deltas: # On first sample site. with ExitStack() as stack: for plate in stack.enter_context(block_plate(dim=plate.dim, strict=False)) self.deltas = new_fn = self.deltas.pop(name) value = new_fn.v if poutine.get_mask() is not False: log_density = new_fn.log_density + fn.log_prob(value) new_fn = dist.Delta(value, log_density, new_fn.event_dim) return {"fn": new_fn, "value": value, "is_observed": True}
[docs] def transform_samples(self, aux_samples, save_params=None): """ Given latent samples from the warped posterior (with a possible batch dimension), return a `dict` of samples from the latent sites in the model. :param dict aux_samples: Dict site name to tensor value for each latent auxiliary site (or if ``save_params`` is specifiec, then for only those latent auxiliary sites needed to compute requested params). :param list save_params: An optional list of site names to save. This is useful in models with large nuisance variables. Defaults to None, saving all params. :return: a `dict` of samples keyed by latent sites in the model. :rtype: dict """ with poutine.condition(data=aux_samples), poutine.mask(mask=False): deltas = return {name: delta.v for name, delta in deltas.items()}