Canonical: https://knowlpedia.com/measure-theory/probability-kernel/

# Probability kernel

A measurable family of probability measures indexed by another measurable space.

A **probability kernel** from \((Y,\mathcal T)\) to \((X,\Sigma)\) assigns a [probability measure](https://knowlpedia.com/probability/probability-measure/) \(\mu_y\) on \(X\) to each \(y\in Y\), such that \(y\mapsto\mu_y(E)\) is measurable for every \(E\in\Sigma\).

## Integrating a kernel

Given probability \(\nu\) on \(Y\), the formula \(\mu(E)=\int_Y\mu_y(E)\,d\nu(y)\) defines a probability measure on \(X\). A [disintegration](https://knowlpedia.com/measure-theory/disintegration-theorem/) uses such a kernel to describe conditional probabilities on fibers of a measurable map.
