Source code for pydrobert.torch._enumerate_estimator

# Copyright 2022 Sean Robertson

# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at

#    http://www.apache.org/licenses/LICENSE-2.0

# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

import torch

from ._estimators import Estimator, FunctionOnSample


[docs] class EnumerateEstimator(Estimator): r"""Calculate expectation exactly by enumerating the support of the distribution An unbiased, zero-variance "estimate" of an expectation over a discrete variable may be calculated brute force by enumerating the support and taking the product of function values with their probabilities under the distribution. .. math:: v = \mathbb{E}_{b \sim P}[f(b)] = \sum_b P(b) f(b). When called, the instance does just that. Parameters ---------- proposal The distribution over which the expectation is taken, :math:`P`. Must be able to enumerate its support through :func:`torch.distributions.Distribution.enumerate_support` (``proposal.has_enumerate_support == True``). func is_log Returns ------- v : torch.Tensor Warnings -------- The call may be both compute- and memory-intensive, depending on the size of the support. """ return_log: bool def __init__( self, proposal: torch.distributions.distribution.Distribution, func: FunctionOnSample, is_log: bool = False, ) -> None: if not proposal.has_enumerate_support: raise ValueError( "proposal must be able to enumerate its support " "(proposal.has_enumerate_support == True)" ) super().__init__(proposal, func, is_log) def __call__(self) -> torch.Tensor: b = self.proposal.enumerate_support() log_pb = self.proposal.log_prob(b) fb = self.func(b) if self.is_log: v = fb + log_pb v = v.logsumexp(0) else: v = (fb * log_pb.exp()).sum(0) return v