1
$\begingroup$

If $(X_n: n\in \mathbb{N}), X$ are a sequence of random variables in $\mathbb{R}$, I wish to show that $X_n \to X$ weakly if and only if $X_n \to X$ in distribution. By 'converging weakly' I mean that $\mu_{X_n}(f) \to \mu_X(f)$ for all continuous bounded functions $f$.

Secondly I want to show that if $(\mu_n: n\in \mathbb{N}), \mu$ are Borel probability measures on $\mathbb{R}^d$, and $\mu_n(f)\to\mu(f)$ for all $C^\infty$ functions on $\mathbb{R}^d$ of compact support, then $\mu_n$ converges weakly to $\mu$ on $\mathbb{R}^d$.

Any help with these questions would be really appreciated. Thanks!

$\endgroup$
  • $\begingroup$ And what do you mean by "converge in distribution"? $\endgroup$ – Nate Eldredge Jan 26 '14 at 17:32
  • $\begingroup$ By converge in distribution I mean that the distribution functions $F_n$ of $X_n$ converge to the distribution function $F$ of $X$ at all points where $F$ is continuous. $\endgroup$ – user123998 Jan 26 '14 at 17:37
  • $\begingroup$ You assume that your random variables all have an absolutely continuous distribution? I.e. no concentration? $\endgroup$ – Emanuele Paolini Jan 26 '14 at 18:06
  • $\begingroup$ I do not think there is any concentration in the distributions of the random variables. $\endgroup$ – user123998 Jan 26 '14 at 18:46
0
$\begingroup$

Assume $X_{n}$ converges weakly to $X$. Let $a,b$ be points of continuity of the CDF for $X$. Define continuous functions $f_{\delta}$ and $g_{\delta}$ so that $f_{\delta}$ is 1 on $[a+\delta,b-\delta]$ and tapers linearly to $0$ on $\mathbb{R}\setminus[a,b]$, and so that $g_{\delta}$ is $1$ on $[a,b]$ and tapers linearly to $0$ on $\mathbb{R}\setminus[a-\delta,b+\delta]$. By the continuity of the distribution of $X$ at $a$ and $b$, for each $\epsilon > 0$, there exists $\delta > 0$ such that $$ \mu_{X}[a,b]-\epsilon < \mu_{X}(f_{\delta}) \le \mu_{X}(g_{\delta}) < \mu_{X}[a,b]+\epsilon. $$ You should be able to take it from there.

For the opposite direction, use behavior distributional behavior of $\mu_{X}$ at $\pm\infty$ and distributional convergence to reduce the problem to a finite interval where you can uniformly approximate a continuous function $f$ by a linear combination of characteristic functions $\chi_{[a,b)}$.

$\endgroup$

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.