gwkokab.errors
==============

.. py:module:: gwkokab.errors


Functions
---------

.. autoapisummary::

   gwkokab.errors.banana_error
   gwkokab.errors.mock_spin_error
   gwkokab.errors.truncated_normal_error


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

.. py:function:: banana_error(Mc_true: numpy.ndarray, eta_true: numpy.ndarray, size: int, key: jaxtyping.PRNGKeyArray, *, estimates: dict[str | gwkokab.parameters.Parameters, numpy.ndarray], rho: numpy.ndarray, scale_Mc: float = 1.0, scale_eta: float = 1.0) -> numpy.ndarray

   Add banana error to the given values. Section 3 of the `Model-independent
   inference on compact-binary observations <https://doi.org/10.1093/mnras/stw2883>`_
   discusses the banana error. It adds errors in the chirp mass and symmetric mass
   ratio and then converts back to masses.

   .. math::

       M_{c} = M_{c}^{T}
       \left[1+\beta\frac{12}{\rho}\left(r_{0}+r\right)\right]

       \eta = \eta^{T}
       \left[1+0.03\frac{12}{\rho}\left(r_{0}^{'}+r^{'}\right)\right]

   Mc_true : np.ndarray
       True chirp mass
   eta_true : np.ndarray
       True symmetric mass ratio
   size : int
       number of samples
   key : PRNGKeyArray
       jax random key
   scale_Mc : float
       scale of the chirp mass error, defaults to 1.0
   scale_eta : float
       scale of the symmetric mass ratio error, defaults to 1.0
   estimates : dict[str | P, np.ndarray]
       Parameter estimates performed so far
   rho : np.ndarray
       SNR of the event, used to scale the error

   :returns: array of values with added banana error
   :rtype: np.ndarray


.. py:function:: mock_spin_error(chi_eff: numpy.ndarray, eta: numpy.ndarray, size: int, key: jaxtyping.PRNGKeyArray, *, estimates: dict[str | gwkokab.parameters.Parameters, numpy.ndarray], rho: numpy.ndarray, scale_chi_eff: numpy.ndarray) -> numpy.ndarray

.. py:function:: truncated_normal_error(x: numpy.ndarray, size: int, key: jaxtyping.PRNGKeyArray, *, scale: float, estimates: dict[str | gwkokab.parameters.Parameters, numpy.ndarray], rho: numpy.ndarray, low: Optional[float] = None, high: Optional[float] = None) -> numpy.ndarray

   Adds truncated normal error to the given values.

   The error is sampled from a truncated normal distribution with the given parameters. The function will resample
   until all values are within the allowed range.

   .. note::

       If :code:`cut_low` and :code:`cut_high` are both None, the function will return
       the sampled values without any truncation.

   .. note::

       if :code:`low` or :code:`high` is None, the :code:`TruncatedNormal` distribution
       will not be truncated at those bounds.

   :param x: Given values to which the error will be added.
   :type x: np.ndarray
   :param size: Number of samples to generate.
   :type size: int
   :param key: JAX random key for sampling.
   :type key: PRNGKeyArray
   :param scale: Scale parameter for the truncated normal distribution.
   :type scale: float
   :param estimates: Parameter estimates performed so far
   :type estimates: dict[str | P, np.ndarray]
   :param rho: SNR of the event, used to scale the error
   :type rho: np.ndarray
   :param low: Lower bound for the truncation, defaults to None (no lower bound).
   :type low: Optional[float], optional
   :param high: Upper bound for the truncation, defaults to None (no upper bound).
   :type high: Optional[float], optional

   :returns: Array of values with added truncated normal error, with all values within the
             specified bounds.
   :rtype: np.ndarray