import numpy as np
import scipy.stats as stats
from popsynth.auxiliary_sampler import AuxiliaryParameter, AuxiliarySampler
[docs]class TruncatedNormalAuxSampler(AuxiliarySampler):
_auxiliary_sampler_name = "TruncatedNormalAuxSampler"
mu = AuxiliaryParameter(default=0)
tau = AuxiliaryParameter(default=1, vmin=0)
lower = AuxiliaryParameter()
upper = AuxiliaryParameter()
sigma = AuxiliaryParameter(default=1, vmin=0)
[docs] def __init__(self, name: str, observed: bool = True):
"""
A truncated normal sampler,
where property ~ N(``mu``, ``sigma``), between
``lower`` and ``upper``.
:param name: Name of the property
:type name: str
:param observed: `True` if the property is observed,
`False` if it is latent. Defaults to `True`
:type observed: bool
:param mu: Mean of the normal
:type mu: :class:`AuxiliaryParameter`
:param tau: Standard deviation of the normal
:type tau: :class:`AuxiliaryParameter`
:param lower: Lower bound of the truncation
:type lower: :class:`AuxiliaryParameter`
:param upper: Upper bound of the truncation
:type upper: :class:`AuxiliaryParameter`
:param sigma: Standard deviation of normal distribution
from which observed values are sampled, if ``observed``
is `True`
:type sigma: :class:`AuxiliaryParameter`
"""
super(TruncatedNormalAuxSampler, self).__init__(
name=name, observed=observed
)
[docs] def true_sampler(self, size):
lower = (self.lower - self.mu) / self.tau
upper = (self.upper - self.mu) / self.tau
self._true_values = stats.truncnorm.rvs(
lower,
upper,
loc=self.mu,
scale=self.tau,
size=size,
)
assert np.alltrue(self._true_values >= self.lower)
assert np.alltrue(self._true_values <= self.upper)
[docs] def observation_sampler(self, size):
if self._is_observed:
self._obs_values = stats.norm.rvs(
loc=self._true_values, scale=self.sigma, size=size
)
else:
self._obs_values = self._true_values