跳转至

预备内容

如果\(X = \ln(Y)\)服从正态分布,则\(Y\)服从对数正态分布。假定\(X \in N(\mu, \sigma^2)\),则\(Y\)服从的对数正态分布参数为:

\[ E(Y) = \exp(\mu + \frac{\sigma^2}{2}), \quad \operatorname{Var}(Y) = [\exp (\sigma^2) - 1]\exp(2\mu + \sigma^2) \]

反之,如果\(Y \sim LN(\mu_y, \sigma_y^2)\),则\(X\)服从的正态分布为:

\[ E(X) = \ln \left(\frac{\mu_y}{\sqrt{1+\sigma_y^2 / \mu_y^2}}\right), \quad \operatorname{Var}(X) = \ln (1+\frac{\sigma_y^2}{\mu_y^2}) \]