1
$\begingroup$

Let $(\mu_n)_{n \in \mathbb{N}}$ be a sequence of a measures. We know, by the Portmanteau Theorem, that: $$\int f d \mu_n \to \int f d\mu, \quad \forall \, f \in C_b \hbox{(class of continuous and bounded function)}$$ is equivalent to $\mu_n(E) \to \mu(E)$, for all $E \in \mathcal{C}_\mu$, class of continuity sets of $\mu$.

Now, I want show that

\begin{equation}\label{I}\tag{I} \int f d \mu_n \to \int f d\mu, \quad \forall \, f \in C_{b}, \hbox{ vanishing on a neighborhood of } 0 \end{equation}

is equivalent to

\begin{equation}\label{II}\tag{II} \mu_n(E) \to \mu(E), \quad E \in \mathcal{C}_{\mu}, \,\, 0 \notin \bar{E} \end{equation} where $\bar{E}$ denotes the closure of $E$.

Is this equivalence true?

Update

Let's try to give a stretch of proof of (\ref{II}) implies (\ref{I}). The converse was given in Jochen Wengenroth's answer.

Assume (\ref{II}) is valid and let any fixed $f \in \mathcal{C}_b$ vanishing on a neighborhood of $0$. Denote such neighborhood as $V_f$. Define for all $E \in \mathcal{C}_\mu$: $$\nu(E) := \mu(E \cap V_f^c),\quad \nu_n(E) := \mu_n(E \cap V_f^c)$$ Since $0 \notin \overline{E \cap V_f^c}$ and $E \cap V_f^c \in \mathcal{C}_\mu$, using (\ref{II}), we have that $\nu_n(E) \to \nu(E)$, as $n \to \infty$ for all $E \in \mathcal C_\nu = \mathcal C_\mu$. By the Portmanteau Theorem, we have that $$\int \bar f \chi_{V_f^c} d \mu_n = \int \bar f d \nu_n \to \int \bar f d \nu = \int \bar f \chi_{V_f^c} d \mu , \quad \bar f \in \mathcal C_b$$

So taking any $\bar f \in \mathcal C_b$ such that $\bar f|_{V_f^c} \equiv f|_{V_f^c}$, we have that $\bar f \chi_{V_f^c}= f$. This shows (\ref{I}) for $f$.

$\endgroup$
2
  • 1
    $\begingroup$ Do you assume probabibilty measures? $\endgroup$ Nov 13, 2022 at 10:51
  • $\begingroup$ Not necessarily. $\endgroup$
    – PSE
    Nov 14, 2022 at 2:38

1 Answer 1

4
$\begingroup$

It seems that this can be deduced from the Portmanteau theorem: Assume the convergence of the intergals $\int fd\mu_n$ for all $f\in C_b$ vanishing in a neighbourhood of $0$ and fix $E\in C_\mu$ with $0\notin \overline E$. You may then choose a continuous function $g$ with values in $[0,1]$ which is $1$ in a neighbourhood of $E$ and $0$ in a neighbourhood of $0$. Then $E$ is also a continuity set of the measure $g\cdot \mu (A)=\int_Agd\mu$ and the measures $g\cdot\mu_n$ satisfy $\int fd(g\cdot \mu_n) = \int fgd\mu_n \to \int fgd\mu=\int fd(g\cdot \mu)$ for every $f\in C_b$ since $fg$ vanishes in a neighbourhood of $0$. Hence, $\mu_n(E)=(g\cdot \mu_n)(E)\to (g\cdot\mu)(E)=\mu(E)$.

The other implication is very similar.

$\endgroup$
1
  • 1
    $\begingroup$ Dear, thank you. I updated the question just to write my attempt to show the converse. If there are any grotesque errors, please let me know. $\endgroup$
    – PSE
    Nov 14, 2022 at 2:37

Your Answer

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

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