nnpd.nre.distributions
======================

.. py:module:: nnpd.nre.distributions

.. autoapi-nested-parse::

   Distribution interfaces, independent bounded priors, and exact samplers.



Classes
-------

.. autoapisummary::

   nnpd.nre.distributions.Parameter
   nnpd.nre.distributions.Simulator
   nnpd.nre.distributions.Prior
   nnpd.nre.distributions.IndependentPrior


Functions
---------

.. autoapisummary::

   nnpd.nre.distributions.grid_axis


Module Contents
---------------

.. py:class:: Parameter

   .. py:attribute:: name
      :type:  str


   .. py:attribute:: low
      :type:  float


   .. py:attribute:: high
      :type:  float


   .. py:attribute:: options
      :type:  dict


.. py:function:: grid_axis(low: float, high: float, step: float) -> numpy.ndarray

.. py:class:: Simulator

   Bases: :py:obj:`abc.ABC`


   Helper class that provides a standard way to create an ABC using
   inheritance.


   .. py:attribute:: parameters
      :type:  tuple[Parameter, ...]


   .. py:attribute:: observation_dim
      :type:  int


   .. py:method:: sample(theta: numpy.ndarray, rng: numpy.random.Generator) -> numpy.ndarray
      :abstractmethod:


      One observation per parameter row; result shape (..., observation_dim).



   .. py:method:: log_likelihood(x: numpy.ndarray, theta: numpy.ndarray) -> numpy.ndarray
      :abstractmethod:


      Optional exact reference, not used by the NRE training objective.



.. py:class:: Prior

   Bases: :py:obj:`abc.ABC`


   Helper class that provides a standard way to create an ABC using
   inheritance.


   .. py:attribute:: parameters
      :type:  tuple[Parameter, ...]


   .. py:method:: sample(size: int, rng: numpy.random.Generator) -> numpy.ndarray
      :abstractmethod:


      Draw shape (size, number_of_inferred_parameters).



   .. py:method:: log_prob(theta: numpy.ndarray) -> numpy.ndarray
      :abstractmethod:


      Log density or log mass with respect to the declared support measure.



   .. py:property:: support_axes
      :type: tuple[numpy.ndarray | None, ...]


      None means continuous; an array means discrete counting measure.



   .. py:property:: measure
      :type: str



.. py:class:: IndependentPrior(parameters: tuple[Parameter, ...], definition: dict)

   Bases: :py:obj:`Prior`


   Uniform, truncated normal, bounded signed exponential, or grid per axis.

   A negative exponential rate is valid on a finite box; it is not an unbounded
   exponential distribution. Grid log_prob returns a mass, never a density.
   'by_parameter' may override the default family for particular coordinates.


   .. py:attribute:: families


   .. py:attribute:: parameters


   .. py:attribute:: kinds


   .. py:property:: support_axes

      None means continuous; an array means discrete counting measure.



   .. py:method:: sample(size: int, rng: numpy.random.Generator) -> numpy.ndarray

      Draw shape (size, number_of_inferred_parameters).



   .. py:method:: log_prob(theta: numpy.ndarray) -> numpy.ndarray

      Log density or log mass with respect to the declared support measure.