nessai.reparameterisations.discrete
===================================

.. py:module:: nessai.reparameterisations.discrete

.. autoapi-nested-parse::

   Reparameterisations for discrete variables

   ..
       !! processed by numpydoc !!


Classes
-------

.. autoapisummary::

   nessai.reparameterisations.discrete.Dequantise


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

.. py:class:: Dequantise(parameters=None, prior_bounds=None, rescale_bounds=None, update_bounds=False, post_rescaling=None, rng=None)

   Bases: :py:obj:`nessai.reparameterisations.rescale.RescaleToBounds`


   
   Reparameterisation that adds noise to discrete variables and then
   rescales to the specified bounds.

   Can also optionally apply a sigmoid/logit transform after rescaling.
   See :py:class:`~nessai.reparameterisations.rescale.RescaleToBounds` for
   more details.

   Note that :code:`update_bounds` is disabled by default and its use is not
   recommended with this reparameterisation.

   :Parameters:

       **parameters** : list[str]
           List of parameters to apply the reparameterisation to

       **prior_bounds** : dict
           Dictionary of prior bounds

       **rescale_bounds** : Optional[list]
           Bounds to rescale to. Defaults to [-1, 1].

       **update_bounds** : bool
           Update the bounds for rescaling during sampling. Can be enabled but
           not recommended.

       **post_rescaling** : Optional[str]
           Name of the rescaling to apply after rescaling to the specified bounds.














   ..
       !! processed by numpydoc !!

   .. py:method:: set_bounds(prior_bounds)

      
      Set the initial bounds for rescaling
















      ..
          !! processed by numpydoc !!


   .. py:method:: pre_rescaling(x)

      
      Function applied before rescaling to bounds
















      ..
          !! processed by numpydoc !!


   .. py:method:: pre_rescaling_inv(x)

      
      Inverse of function applied before rescaling to bounds
















      ..
          !! processed by numpydoc !!