public
================

.. py:module:: jaxns.public

.. rubric:: :code:`jaxns.public`


.. rubric:: Module Contents

.. py:class:: NestedSampler

   A static nested sampler that uses 1-dimensional slice sampler for the sampling step.
   Uses the phantom-powered algorithm. A robust default choice is provided for all parameters.

   s,k,c are defined in the paper: https://arxiv.org/abs/2312.11330

   :param model: a model to perform nested sampling on
   :param max_samples: maximum number of samples to take
   :param num_live_points: approximate number of live points to use. Defaults is c * (k + 1).
   :param s: number of slices to use per dimension. Defaults to 4.
   :param k: number of phantom samples to use. Defaults to 0.
   :param c: number of parallel Markov-chains to use. Defaults to 20 * D.
   :param devices: devices to use. Defaults to all available devices.
   :param difficult_model: if True, uses more robust default settings. Defaults to False.
   :param parameter_estimation: if True, uses more robust default settings for parameter estimation. Defaults to False.
   :param shell_fraction: fraction of the shell to use for the slice sampler. Defaults to 0.5.
   :param gradient_guided: if True, uses gradient guided sampling. Defaults to False.
   :param init_efficiency_threshold: if > 0 then use uniform sampling first down to this acceptance efficiency.
                                     0 turns it off.
   :param verbose: whether to log progress.


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


   .. py:attribute:: max_samples
      :type:  Optional[Union[int, float]]
      :value: None



   .. py:attribute:: num_live_points
      :type:  Optional[int]
      :value: None



   .. py:attribute:: num_slices
      :type:  Optional[int]
      :value: None



   .. py:attribute:: s
      :type:  Optional[Union[int, float]]
      :value: None



   .. py:attribute:: k
      :type:  Optional[int]
      :value: None



   .. py:attribute:: c
      :type:  Optional[int]
      :value: None



   .. py:attribute:: devices
      :type:  Optional[List[jaxlib.xla_client.Device]]
      :value: None



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



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



   .. py:attribute:: shell_fraction
      :type:  float
      :value: 0.5



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



   .. py:attribute:: init_efficiency_threshold
      :type:  float
      :value: 0.1



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



   .. py:method:: __post_init__()


   .. py:property:: nested_sampler
      :type: jaxns.nested_samplers.abc.AbstractNestedSampler



   .. py:method:: __call__(key, term_cond = None)

      Performs nested sampling with the given termination conditions.

      :param key: PRNGKey
      :param term_cond: termination conditions. If not given, see `TerminationCondition` for defaults.

      :returns: termination reason, state



   .. py:method:: to_results(termination_reason, state, trim = True)

      Convert the state to results.

      Note: Requires static context.

      :param termination_reason: termination reason
      :param state: state to convert
      :param trim: if True, trims the results to the number of samples taken, requires static context.

      :returns: results



   .. py:method:: trim_results(results)
      :staticmethod:


      Trims the results to the number of samples taken. Requires static context.

      :param results: results to trim

      :returns: trimmed results