0
$\begingroup$

Let $\mu_n$ be a sequence of probability measures, and $\mu$ some other probability measure (all on the same space). Denote $C((\mu_n), \mu)$ the space of functions $f$ such that $\int f d\mu_n\to_{n\to\infty}\int f d\mu$. Can anything be said a priori about this set? Is it necessarily closed under monotone convergence, for example?

More generally, let $E$ be a subspace of functions for which all of the required integrals are defined. When does there exist a $(\mu_n)$ and $\mu$ such that $E=C((\mu_n), \mu)$ ?

$\endgroup$

1 Answer 1

2
$\begingroup$

This set is not closed with respect to monotone limits in general: if $\mu_n = \delta_{1/n}$, $\mu = \delta_0$, $f_k = \mathbb{1}_{[1/k,1]}$, $f = \mathbb{1}_{(0,1]}$ then $\int f_k d\mu_n = 0 = \int f_k d\mu$ for $n$ large enough, but $\int f d\mu_n = 1 \ne 0 = \int f d\mu$ for all $n$.

It is closed under uniform convergence, though: if $f_k$ converge to $f$ uniformly, then the sequence $(\int f_k d\mu_n : n > 0)$ converge to the sequence $(\int f d\mu_n : n > 0)$ uniformly, and also $\int f_k d\mu$ converges to $\int f d\mu$ (as $k \to \infty$).

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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