Canonical: https://knowlpedia.com/probability/cumulant/

# Cumulant

A numerical summary of a distribution given by derivatives of the cumulant generating function at zero.

A **cumulant** (of order \(n\)) of a [random variable](https://knowlpedia.com/probability/random-variable/) \(X\) is the number
\[
\kappa_n(X) \;=\; K_X^{(n)}(0),
\]

provided the [cumulant generating function](https://knowlpedia.com/probability/cumulant-generating-function/) \(K_X\) exists in a neighborhood of \(0\) and is \(n\) times differentiable at \(0\).

## Remarks

The first two cumulants are \(\kappa_1(X)=\mathbb{E}[X]\) and \(\kappa_2(X)=\mathrm{Var}(X)\), linking cumulants to [expectation](https://knowlpedia.com/probability/expectation/) and [variance](https://knowlpedia.com/probability/variance/). Cumulants are additive over sums of [independent random variables](https://knowlpedia.com/probability/independence-random-variables/) (when defined), which makes them useful for studying distributional limits and approximations.

## Examples

- If \(X\sim \mathcal{N}(\mu,\sigma^2)\), then \(\kappa_1(X)=\mu\), \(\kappa_2(X)=\sigma^2\), and \(\kappa_n(X)=0\) for all \(n\ge 3\).
- If \(X\sim \mathrm{Bernoulli}(p)\), then \(\kappa_1(X)=p\), \(\kappa_2(X)=p(1-p)\), and \(\kappa_3(X)=p(1-p)(1-2p)\).
