coronalyze.core.statistics

Statistical functions for masked arrays and small-sample corrections.

Implements JAX-native masked statistics and the Mawet et al. (2014) small-sample penalty for high-contrast imaging SNR calculations.

Functions

masked_mean(values, mask)	Compute the mean of masked values.
masked_std(values, mask[, mean])	Compute the standard deviation of masked values.
small_sample_penalty(n)	Compute the Mawet et al. (2014) small-sample statistics correction.

Module Contents

coronalyze.core.statistics.masked_mean(values, mask)

Compute the mean of masked values.

Args:

values: 1D array of values. mask: Boolean mask (True for valid values).

Returns:

Mean of valid values, or 0 if no valid values.

Parameters:

• values (jax.numpy.ndarray)

• mask (jax.numpy.ndarray)

Return type:

float

coronalyze.core.statistics.masked_std(values, mask, mean=None)

Compute the standard deviation of masked values.

Uses Bessel’s correction (N-1 denominator) for unbiased estimation.

Args:

values: 1D array of values. mask: Boolean mask (True for valid values). mean: Pre-computed mean. If None, computed from masked values.

Returns:

Standard deviation of valid values.

Parameters:

• values (jax.numpy.ndarray)

• mask (jax.numpy.ndarray)

• mean (float | None)

Return type:

float

coronalyze.core.statistics.small_sample_penalty(n)

Compute the Mawet et al. (2014) small-sample statistics correction.

At small angular separations, fewer reference apertures are available, which inflates the noise estimate. This penalty factor accounts for the additional uncertainty.

Reference: Mawet et al. (2014) ApJ

Equation 9: sigma_corrected = sigma * sqrt(1 + 1/n)

Args:

n: Number of reference apertures.

Returns:

Correction factor sqrt(1 + 1/n).

Parameters:

n (int | jax.numpy.ndarray)

Return type:

float