0
$\begingroup$

I came across the question: Evaluate$f(x)=\lim_{m \to \infty}\lim_{n \to \infty}[\cos(n!\pi x)]^{2m}$. I simplified this to: $$f(x)= \begin{cases} 1 & \text{if $x \in Q$} \\ 0 & \text{if $x\notin Q$} \end{cases}$$ Is this function continuous?

As far as I know, if the limit of the function at a point $x=a$ equals $f(a)$, then the function is continuous at $x=a$. In this case however, how does one decide whether the limit is 1, 0, or does not exist?

$\endgroup$
8
  • $\begingroup$ Your distinction should focus on whether $x$ is integer or not, not on it being rational/irrational. In the case of an integer $x$, this is just $1$, yes. In the case of almost all irrational $x$, the inner limit $n\rightarrow\infty$ does not exist I am pretty sure. Are you sure about the order of the limits? Things would be different if $m \rightarrow\infty$ was on the inside. Your last paragraph is very fishy too. $\endgroup$
    – GDumphart
    Apr 21, 2015 at 11:12
  • 2
    $\begingroup$ @GDumphart: If $x$ is rational, then for $n$ sufficiently large, $n!$ will contain all prime factors of the denominator of $x$, and hence $n! x \in \mathbb N_0$. $\endgroup$
    – cfh
    Apr 21, 2015 at 11:15
  • 1
    $\begingroup$ @GDumphart Well for rational $x$'s eventually the denominator is canceled out by the $n!$. So I think the OP is right there. The issue is irrational $x$'s where very probably the inner limit does not exist. $\endgroup$
    – DRF
    Apr 21, 2015 at 11:15
  • $\begingroup$ @cfh Oh, you guys are correct of course. $\endgroup$
    – GDumphart
    Apr 21, 2015 at 11:16
  • $\begingroup$ I think @DRF is right that the inner limit doesn't exist for irrational $x$; it basically looks like some pseudo-random generator. With the limits in $m$ and $n$ swapped, things would be different. $\endgroup$
    – cfh
    Apr 21, 2015 at 11:22

1 Answer 1

1
$\begingroup$

In the comments to your question, we have discussed that your first $f$ cannot be your second $f$. However, if we just look at $$ f(x)= \begin{cases} 1 & \text{if $x \in Q$} \\ 0 & \text{if $x\notin Q$} \end{cases}, $$ this is a well-defined function and we can think about whether it's continuous in some point $x$ or not.

First, assume that $x \in \mathbb Q$, so $f(x) = 1$. Then, for any $\epsilon>0$, you can find an irrational number $y$ with $|x-y| < \epsilon$ (for instance by choosing $y = x + \frac1n \sqrt2$ with a large enough natural number $n$). But then $|f(x) - f(y)| = 1$ even though $|x-y|$ can be arbitrarily small.

On the other hand, if $x \notin \mathbb Q$, we can use that the rational numbers are dense in the real numbers to find a $y \in \mathbb Q$ with $|x-y| < \epsilon$ for any small $\epsilon$ we like. Again, $|f(x) - f(y)| = 1$.

In both cases, if you refer to the $\epsilon$-$\delta$ definition of continuity, this shows that $f$ is not continuous in $x$. Thus, $f$ is a function which is continuous nowhere.

$\endgroup$
2
  • $\begingroup$ I'd like to add that this is known as the Dirichlet function (usually shown as an intuitive example for a non-Riemann-integrable function, but Lebesque integration gives $0$). en.wikipedia.org/wiki/… $\endgroup$
    – GDumphart
    Apr 21, 2015 at 12:37
  • $\begingroup$ Oh look, my Wiki link even contains the problem of the question and, like I suggested, the limits are taken the other way around. $\endgroup$
    – GDumphart
    Apr 21, 2015 at 12:40

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .