def get_sample_means(
self,
sample_size: int,
num_samples: int = 1,
t: float = -1,
) -> np.ndarray:
"""
Calculate the sample means.
Each mean is from a sample of ``sample_size`` agents.
The number of means is the same as ``num_samples``.
Means are taken either from ``opinions`` argument or from ``self.result.y`` at time point ``t``.
see https://en.wikipedia.org/wiki/Central_limit_theorem
:param sample_size: Pick this many agents' opinions (i.e. a sample).
:param num_samples: Number of sample to perform.
:param t: Time at which to conduct the sampling (-1 for last time point).
:return: Array of means (size equal to ``num_samples``).
"""
t_idx = np.argmin(np.abs(t -
self.result.t)) if isinstance(t, float) else t
opinions = self.result.y[:, t_idx]
return sample_means(opinions, sample_size, num_samples, rng=self.rn)
def _subsample_clt_sample(sn, y: np.ndarray, n: int, num_samples: int):
"""
method to re-assign self._sample_means
:math:`(X_1 - \\bar{X_n})`
where :math:`X` is a random sample and :math:`\\bar{X}_{n}` is the sample mean for :math:`n` random samples.
:param sn: SocialNetwork object to store sample means (and to use it's random number generator)
:type sn: opdynamics.socialnetworks.SocialNetwork
:param y: opinions
:param n: sample size
:param num_samples: number of samples (N)
"""
sn._sample_means = sample_means(
y, 1, num_samples=num_samples, rng=sn.rn) - sample_means(
y, n, num_samples=num_samples, rng=sn.rn)
def _basic_clt_sample(sn, y: np.ndarray, n: int, num_samples: int):
"""
method to re-assign self._sample_means
:math:`\\sqrt {n}\\left(\\bar{X}_{n}-\\mu \\right) \\rightarrow \\mathcal{N}\\left(0,\\sigma ^{2}\\right)`
where :math:`X` is a random sample and :math:`\\bar{X}_{n}` is the sample mean for :math:`n` random samples.
:param sn: SocialNetwork object to store sample means (and to use it's random number generator)
:type sn: opdynamics.socialnetworks.SocialNetwork
:param y: opinions
:param n: sample size
:param num_samples: number of samples (N)
"""
sn._sample_means = np.sqrt(n) * (
sample_means(y, n, num_samples=num_samples, rng=sn.rn) - np.mean(y))