Source code for pyro.distributions.transforms.discrete_cosine

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

import torch
from torch.distributions.transforms import Transform

from pyro.ops.tensor_utils import dct, idct

from .. import constraints


[docs]class DiscreteCosineTransform(Transform): """ Discrete Cosine Transform of type-II. This uses :func:`~pyro.ops.tensor_utils.dct` and :func:`~pyro.ops.tensor_utils.idct` to compute orthonormal DCT and inverse DCT transforms. The jacobian is 1. :param int dim: Dimension along which to transform. Must be negative. This is an absolute dim counting from the right. :param float smooth: Smoothing parameter. When 0, this transforms white noise to white noise; when 1 this transforms Brownian noise to to white noise; when -1 this transforms violet noise to white noise; etc. Any real number is allowed. https://en.wikipedia.org/wiki/Colors_of_noise. """ bijective = True def __init__(self, dim=-1, smooth=0.0, cache_size=0): assert isinstance(dim, int) and dim < 0 self.dim = dim self.smooth = float(smooth) self._weight_cache = None super().__init__(cache_size=cache_size) def __hash__(self): return hash((type(self), self.dim, self.smooth)) def __eq__(self, other): return ( type(self) == type(other) and self.dim == other.dim and self.smooth == other.smooth ) @constraints.dependent_property(is_discrete=False) def domain(self): return constraints.independent(constraints.real, -self.dim) @constraints.dependent_property(is_discrete=False) def codomain(self): return constraints.independent(constraints.real, -self.dim) @torch.no_grad() def _weight(self, y): size = y.size(-1) if self._weight_cache is None or self._weight_cache.size(-1) != size: # Weight by frequency**smooth, where the DCT-II frequencies are: freq = torch.linspace(0.5, size - 0.5, size, dtype=y.dtype, device=y.device) w = freq.pow_(self.smooth) w /= w.log().mean().exp() # Ensure |jacobian| = 1. self._weight_cache = w return self._weight_cache def _call(self, x): dim = self.dim if dim != -1: x = x.transpose(dim, -1) y = dct(x) if self.smooth: y = y * self._weight(y) if dim != -1: y = y.transpose(dim, -1) return y def _inverse(self, y): dim = self.dim if dim != -1: y = y.transpose(dim, -1) if self.smooth: y = y / self._weight(y) x = idct(y) if dim != -1: x = x.transpose(dim, -1) return x
[docs] def log_abs_det_jacobian(self, x, y): return x.new_zeros(x.shape[: self.dim])
[docs] def with_cache(self, cache_size=1): if self._cache_size == cache_size: return self return DiscreteCosineTransform(self.dim, self.smooth, cache_size=cache_size)
[docs] def forward_shape(self, shape): if len(shape) < self.event_dim: raise ValueError("Too few dimensions on input") return shape
[docs] def inverse_shape(self, shape): if len(shape) < self.event_dim: raise ValueError("Too few dimensions on input") return shape