mxnet.gluon.probability.distributions.half_normal¶

Half-normal Distribution

Classes

HalfNormal([scale, validate_args])

Create a half normal object, where

class mxnet.gluon.probability.distributions.half_normal.HalfNormal(scale=1.0, validate_args=None)[source]¶

Create a half normal object, where: X ~ Normal(0, scale) Y = |X| ~ HalfNormal(scale)

Parameters:: scale (Tensor or scalar, default 1) – Scale of the full Normal distribution.

cdf(value)[source]¶: Compute the cumulative distribution function(CDF) p(Y < value)

icdf(value)[source]¶: Returns the inverse cumulative density/mass function evaluated at value.

log_prob(value)[source]¶

Compute log-likelihood of value with log_det_jacobian and log-likelihood of the base distribution according to the following conclusion:

Given that Y = T(X), log(p(y)) = log(p(x)) - log(|dy/dx|)