1
$\begingroup$

Suppose $(X, \Sigma_X)$ and $(Y, \Sigma_Y)$ are measurable spaces such that $\Sigma_Y$ is generated by a set $B$.

Suppose $k : X \times \Sigma_Y \to [0, 1]$ has the property that $k(x, -)$ is a (sub-)probability measure for each $x \in X$, and $k(-, E)$ is measurable for every $E \in B$, wrt the Borel $\sigma$-algebra on $[0, 1]$.

Does that make $k(-, U)$ measurable for each $U \in \Sigma_Y$? (in other words, is $k$ a Markov kernel?)

$\endgroup$

1 Answer 1

1
$\begingroup$

$\newcommand{\N}{\mathbb N} \newcommand{\R}{\mathbb R} \newcommand{\B}{\mathcal B} \newcommand{\F}{\mathcal F} \newcommand{\X}{\mathcal X} \newcommand{\ep}{\epsilon} \newcommand{\la}{\lambda} \newcommand{\si}{\sigma} \newcommand{\Si}{\Sigma} \renewcommand{\c}{\circ} \newcommand{\tr}{\operatorname{tr}} \newcommand{\E}{\operatorname{\mathsf E}}$

The answer is no in general. Indeed, suppose $B=\{E,F\}$ and $k(x,A)$ equals $1/2-f(x),f(x),1/2-f(x),f(x)$ if $A$ is $E\setminus F,E\cap F,F\setminus E,Y\setminus E\setminus F$, respectively, where $f\colon X\to[0,1/2]$ is not $\Si_X$-measurable. Then $k(\cdot, U)$ is measurable for each $U\in B$ but not for $U=E\cap F\in \Sigma_Y$.

However, your desired conclusion will hold if $B$ is an algebra. Indeed, suppose that $B$ is an algebra and consider the set $C$ of all $U\in\Si_Y$ such that $k(\cdot, U)$ is measurable. Clearly, $C$ is a monotone class. Hence, your desired conclusion follows by the monotone class theorem.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.