uni_slice_sampler
===========================

.. py:module:: jaxns.samplers.uni_slice_sampler

.. rubric:: :code:`jaxns.samplers.uni_slice_sampler`


.. rubric:: Module Contents

.. py:class:: UniDimSliceSampler

   Bases: :py:obj:`jaxns.samplers.bases.BaseAbstractMarkovSampler`\ [\ :py:obj:`jaxns.nested_samplers.common.types.SampleCollection`\ ]


   Slice sampler for a single dimension.

   :param model: AbstractModel
   :param num_slices: number of slices between acceptance. Note: some other software use units of prior dimension.
   :param midpoint_shrink: if true then contract to the midpoint of interval on rejection. Otherwise, contract to
                           rejection point. Speeds up convergence, but introduces minor auto-correlation.
   :param num_phantom_save: number of phantom samples to save. Phantom samples are samples that meeting the constraint
                            but are not accepted. They can be used for numerous things, e.g. to estimate the evidence uncertainty.
   :param perfect: if true then perform exponential shrinkage from maximal bounds, requiring no step-out procedure.
                   Otherwise, uses a doubling procedure (exponentially finding bracket).
                   Note: Perfect is a misnomer, as perfection also depends on the number of slices between acceptance.
   :param gradient_slice: if true then always slice along increasing gradient direction.
   :param adaptive_shrink: if true then shrink interval to random point in interval, rather than midpoint.
   :param gradient_guided: if true then do householder reflections at between proposals with a 50% probability.


   .. py:attribute:: model
      :type:  jaxns.framework.bases.BaseAbstractModel


   .. py:attribute:: num_slices
      :type:  int


   .. py:attribute:: num_phantom_save
      :type:  int


   .. py:attribute:: midpoint_shrink
      :type:  bool


   .. py:attribute:: perfect
      :type:  bool


   .. py:attribute:: gradient_slice
      :type:  bool
      :value: False



   .. py:attribute:: adaptive_shrink
      :type:  bool
      :value: False



   .. py:attribute:: gradient_guided
      :type:  bool
      :value: False



   .. py:method:: __post_init__()


   .. py:method:: num_phantom()


   .. py:method:: get_seed_point(key, sampler_state, log_L_constraint)


   .. py:method:: get_sample_from_seed(key, seed_point, log_L_constraint, sampler_state)