Skip to content
Permalink
Branch: master
Find file Copy path
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
38 lines (24 sloc) 1.08 KB
The hyperexponential (mixture exponential) distribution may be useful
for fitting fat-tailed empirical distributions.
\subsection{Hyperexponential densities}
The hyperexponential distribution is a mixture of $K$ independent
exponentials with a common location $\mu$ and different decay
constants $\lambda_k$.
The probability density function (PDF) is:
\begin{equation}
P(X=x) = \sum_k^{K} q_k \lambda_k e^{- \lambda_k (x - \mu)}
\label{eqn:hyperexp_pdf}
\end{equation}
The cumulative distribution function (CDF) is:
\begin{equation}
P(X \leq x) = \sum_k^{K} q_k (1 - e^{- \lambda_k (x - \mu)})
\label{eqn:hyperexp_cdf}
\end{equation}
Variate $x$ ranges $\mu \leq x < \infty$.
Mixture coefficients $q_k$ specify the prior probability of each
component $k$; $0 \leq q_k \leq 1$ and $\sum_k q_k = 1$.
The single location parameter $\mu$ is unconstrained, $-\infty < \mu <
\infty$. (Exponential distributions are usually represented without an
explicit location parameter, implicitly assuming $\mu = 0$.)
The scale parameters $\lambda_k$ for each component are nonnegative,
$\lambda_k > 0$.
You can’t perform that action at this time.