Truncated Normal Distribution

class TruncatedNormal

Namescope for collection of helpers providing different useful properties of truncated normal distributions. Used by WeakRegime_AcceptanceInterval and IndicatorImplication_AcceptanceInterval.

static mills_ratio(beta: float) float

Compute the mills-ratio for \(\beta\).

Parameters:

beta (float) – argument \(\beta\)

Returns:

mills-ratio

Return type:

float

static approx_mills_lower_bound(beta: float) float

Lower bound for mills-ratio.

Parameters:

beta (float) – argument \(\beta > 0\)

Returns:

lower bound for mills-ratio

Return type:

float

static approx_mills_upper_bound(beta)

Upper bound for mills-ratio.

Parameters:

beta (float) – argument \(\beta > 0\)

Returns:

upper bound for mills-ratio

Return type:

float

classmethod inv_mills_ratio(beta: float) float

Reciprocal value of the mills-ratio with improved numerical stability.

Parameters:

beta (float) – argument \(\beta > 0\)

Returns:

reciprocal value of mills-ratio

Return type:

float

classmethod mean_cutoff_below(true_mean: float, true_sigma: float, cutoff: float) float

Mean-value of a truncated-below normal distribution.

Parameters:
  • true_mean (float) – the normal-distribution’s original mean-value parameter \(\mu\)

  • true_sigma (float) – the normal-distribution’s original standard-deviation parameter \(\sigma\)

  • cutoff (float) – the cutoff location \(c\)

Returns:

\(E[X|X\geq c]\), where \(X \sim \mathcal{N}(\mu, \sigma^2)\).

Return type:

float

classmethod mean_cutoff_above(true_mean: float, true_sigma: float, cutoff: float) float

Mean-value of a truncated-above normal distribution.

Parameters:
  • true_mean (float) – the normal-distribution’s original mean-value parameter \(\mu\)

  • true_sigma (float) – the normal-distribution’s original standard-deviation parameter \(\sigma\)

  • cutoff (float) – the cutoff location \(c\)

Returns:

\(E[X|X\leq c]\), where \(X \sim \mathcal{N}(\mu, \sigma^2)\).

Return type:

float