Vague convergence of distribution functions

78 Views Asked by At

I'm now studying Durret's probability theory and examples. In section 3.2, he discussed on 'vague convergence' of distribution functions.

He said a family of distribution functions $F_{n}(x)$ (i.e. non-decreasing, right-continuous and $F(\infty)=1$, $F(-\infty)=0$) vaguely converges to a function $F(x)$ if $F(x)$ is non-decreasing and right-continuous with $$F(x)=\lim_{n \to \infty}F_{n}(x)$$ for all continuity points $x$ of $F$.

In the above definition, I thought that the non-decreasing and right-continuous condition is necessary to define such a concept (i.e., well-definedness of the limit). However, I wonder if there is a limit of distribution functions which is not non-decreasing or right-continuous. By summarizing, the question that I want to ask is as follows.

Q. Let $F_{n}:\mathbb{R} \to [0,1]$ be a family of distribution functions and $F$ be any function. If $$\lim_{n \to \infty} F_{n}(x)=F(x)$$ for all continuity points $x$ of $F$, then is $F$ necessarily right-continuous? non-decreasing? If it is, how can I prove it? If not, is there any counter-example?

This might be a stupid question, but I got a little confused from this question.

Thank you for answering this question.

1

There are 1 best solutions below

0
On BEST ANSWER

The limit $F$ is non-decreasing : if $x\leq y$, then for all $n\in\mathbb N$, $F_n(x)\leq F_n(y)$, and by taking the limit we get $F(x) \leq F(y)$.

However, it is not necessarily right-continuous and $F(+\infty) = 1,F(-\infty) =0$ do not hold automatically.

Let $\chi$ the distribution function for the uniform law on $[0,1]$, ie $$\chi(x) =\left\{\begin{array}{cl} 0 & \text{if } x\leq 0 \\ x & \text{if }x\in[0,1]\\ 1 &\text{if } x\geq 1\end{array}\right.$$

Then, with $F_n(x) = \chi(\frac xn)$, we have: $$\lim_{n\to +\infty} F_n(x) = \left\{ \begin{array}{cl}0 & \text{if } x\leq 0 \\1 & \text{if } x>0\end{array} \right.$$ which is not right-continuous.

With $F_n(x) = \chi(x +n)$, we have $\lim_{n\to +\infty} F_n(x) =1$ so $F(-\infty) \neq 0$.

With $F_n(x) = \chi(x-n)$, we have $\lim_{n\to +\infty} F_n(x) = 0$ so $F(+\infty) \neq 1$.