Canonical: https://knowlpedia.com/measure-theory/convergence-in-measure/

# Convergence in measure

A mode of convergence where the set of large errors has measure tending to zero.

A **convergence in measure** is the following notion of convergence for a [sequence](https://knowlpedia.com/shared-foundations/sequence/) \((f_n)\) of [measurable functions](https://knowlpedia.com/measure-theory/measurable-function/) on a [measure space](https://knowlpedia.com/measure-theory/measure-space/) \((X,\mathcal{A},\mu)\): we say that \(f_n\) **converges in measure** to a measurable function \(f\) if for every \(\varepsilon>0\),
\[
\mu\big(\{x\in X:\ |f_n(x)-f(x)|>\varepsilon\}\big)\to 0
\quad\text{as }n\to\infty.
\]

This definition treats two functions as close whenever they differ by more than \(\varepsilon\) only on a set of small measure. If \(\mu(X)<\infty\), then [convergence almost everywhere](https://knowlpedia.com/measure-theory/convergence-almost-everywhere/) implies convergence in measure, but the converse can fail.

## Examples

- On \(([0,1],\mathcal{B},\lambda)\) with [Lebesgue measure](https://knowlpedia.com/measure-theory/lebesgue-measure/) \(\lambda\), the functions \(f_n=\mathbf{1}_{[0,1/n]}\) satisfy \(f_n\to 0\) in measure because \(\lambda(\{|f_n|>1/2\})=\lambda([0,1/n])=1/n\to 0\).
- On \((\mathbb{R},\mathcal{B},\lambda)\), the functions \(f_n=\mathbf{1}_{[n,n+1]}\) do **not** converge in measure to \(0\) because for \(\varepsilon=\tfrac12\) one has \(\lambda(\{|f_n|>\varepsilon\})=\lambda([n,n+1])=1\) for all \(n\).
