Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let's consider a sequence $(\mu_n)_n$ of random probability measures on $\mathbb R$, and let $C_b$ be the Banach space of bounded continuous functions on $\mathbb R$. I am considering the following types of convergence, where $\mu$ is some non-random probability measure :

(A) $$ \forall f\in C_b, \quad \mathbb P \left(\int f(x)d\mu_n(x)\rightarrow_{n\rightarrow\infty}\int f(x)d\mu(x)\right)=1$$ and

(B) $$ \mathbb P \left(\forall f\in C_b,\quad \int f(x)d\mu_n(x)\rightarrow_{n\rightarrow\infty}\int f(x)d\mu(x)\right)=1.$$ Whereas (B) clearly implies (A), what about the other direction ?

1) Do you have a counter-example where (A) does not implies (B) ?

2) Do we have a sufficient criterion so that (A) implies (B) ? (I have in mind the use of the Borel-Cantelli lemma to improve the probability convergence to the almost sure one)

share|cite|improve this question
What measure is $\mathbb{P}$? – T. Eskin Jun 7 '12 at 15:52
@ThomasE.: $\mu_n$ are random measures. I.e. each $\mu_n$ is a measurable function from some underlying probability space $(\Omega, \mathcal{F}, \mathbb{P})$ into the space of probability measures on $\mathbb{R}$. – Nate Eldredge Jun 7 '12 at 16:21
@NateEldredge. thanks for clearing this:) – T. Eskin Jun 7 '12 at 16:23
up vote 1 down vote accepted

I believe A and B are equivalent.

First, I believe that when the underlying state space (in this case $\mathbb{R}$) is locally compact, then probability measures $\nu_n$ converge weakly to $\nu$ iff $\int f\, d\nu_n \to \int f\,d\nu$ for all continuous functions $f$ which vanish at infinity, i.e. for all $f \in C_0(\mathbb{R})$. I don't remember the proof off the top of my head, but you can find it in Billingsley's Convergence of Probability Measures.

Now $C_0(\mathbb{R})$ is separable, so one can choose a countable dense subset $\{f_1, f_2,\dots\}$. Then using the triangle inequality and the fact that the measures $\nu_n$ are uniformly bounded in total variation (since they are all probability measures), you can show that $\nu_n \to \nu$ weakly iff $\int f_k \,d\nu_n \to \int f_k \,d\nu$ for all $k$.

Now if A holds, then by countable additivity $$\mathbb{P}\left( \int f_k\,d\mu_n \to \int f_k\,d\mu \text{ for all } k\right) = 1.$$ But we have just argued that on this event, $\mu_n \to \mu$ weakly.

share|cite|improve this answer
I believe that's exactly what I was looking for ! Thx – Student Jun 7 '12 at 21:57

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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