1

There were two previous threads (first and second) that I found which asked the same question but offered different methods of proof. I want to know if at least the logic behind my proof is valid.

My reasoning behind the proof is that if I can show that a periodic continuous function is bounded and uniformly continuous on the compact interval $\displaystyle I=[ 0,k]$, where k is the period, then it is bounded and continuous on $\displaystyle \mathbb{R}$. Because essentially the interval $\displaystyle [ 0,k]$ would contain all the points of f, so these points would just repeat over all of $\displaystyle \mathbb{R}$. Does this work? For thoroughness I've included the proof here, and I apologize for the specific theorem references, but they are very well-known theorems and it would be my application of them that would be at fault since their validity I assume won't be disputed.

Problem: Assume that $\displaystyle f:\mathbb{R}\rightarrow \mathbb{R}$ is periodic and continuous. Prove that $\displaystyle f$ is bounded and uniformly continuous on $\displaystyle \mathbb{R}$.

Solution:

Let $\displaystyle f:\mathbb{R}\rightarrow \mathbb{R}$ be a periodic continuous function. Then there exists some real number $\displaystyle k >0$ such that for all $\displaystyle x\in \mathbb{R}$, $\displaystyle f( x+k) =f( x)$, where $\displaystyle k$ is the period of $\displaystyle f$. Now consider the compact interval $\displaystyle I=[ 0,k]$. Since $\displaystyle I$ is compact, $\displaystyle f( I)$ is also compact by Theorem 5.3.10. Hence $\displaystyle f$ is closed and bounded on $\displaystyle I$ (by Heine-Borel Theorem). Let $\displaystyle m_{1}$ be the minimum and $\displaystyle m_{2}$ be the maximum of $\displaystyle f$ on $\displaystyle I$. Thus $\displaystyle f([ 0,k]) =[ m_{1} ,m_{2}]$. Since $\displaystyle f$ is periodic with period $\displaystyle k$, it follows that $\displaystyle m_{1} \leq f( x) \leq m_{2}$ for all $\displaystyle x\in \mathbb{R}$. Hence $\displaystyle f$ is bounded on $\displaystyle \mathbb{R}$.

Now since $\displaystyle I$ is compact, then by Theorem 5.4.6, $\displaystyle f$ is uniformly continuous on $\displaystyle I$. That is, for every $\displaystyle \varepsilon >0$ there exists $\displaystyle \delta >0$ such that $\displaystyle | f( x) -f( y)| < \varepsilon $ whenever $\displaystyle | x-y| < \delta $ and $\displaystyle x,y\in I$. Since $\displaystyle f$ has period $\displaystyle k$, then for any integer $\displaystyle n$, $\displaystyle f( x+nk) =f( x)$ and $\displaystyle f( y+nk) =f( y)$. Thus $\displaystyle | f( x+nk) -f( y+nk)| < \varepsilon $ whenever $\displaystyle | ( x+nk) -( y+nk)| < \delta $. Hence $\displaystyle f$ is uniformly continuous on $\displaystyle \mathbb{R}$. $\displaystyle \blacksquare $

  • 2
    The only concern one could possibly have is that you never examine a $\delta$-neighborhood so close to an end of a period that is spans into the next copy of the period. That is, you haven't shown that something "bad" could not happen crossing from one copy of the period to the next. (To be clear: I recognize that it can't, but your argument doesn't address the point.) – Eric Towers Dec 12 '21 at 01:19
  • @EricTowers but haven't I shown that f is bounded on all of R and likewise uniformly continuous on all of R and hence obviously holds for any delta neighborhood on R? Since the interval I chose is compact, any integer multiple of that interval (e.g. x+nk, y+nk) would include all real numbers, right? – jinks908 Dec 13 '21 at 02:33
  • "and likewise uniformly continuous on all of $\Bbb{R}$ ... ?" No. You chose $|x-y| < \delta$ and $x,y \in I$. So you never considered the delta neighborhoods $(-\delta / 2, \delta / 2)$ and $(k - \delta / 2, k + \delta / 2)$. You definitely showed uniform continuity on $[0,k]$, but you did not show uniform continuity for $\delta$-neighborhoods that straddle your period boundaries. – Eric Towers Dec 13 '21 at 13:33
  • Another way to say this: Consider the collection of all $\delta$-neighborhoods of $\Bbb{R}$. For uniform continuity on $\Bbb{R}$, your argument must address all of them. Your specification of $x,y$ and the periodicity you observed, means that you only addressed $\delta$-neighborhoods that fit entirely in one of the $[nk, (n+1)k], n \in \Bbb{Z}$ intervals, but did not speak to all the $\delta$-neighborhoods that do not. – Eric Towers Dec 13 '21 at 13:39
  • Okay, so just because these conditions are met for any one complete period of f, no matter where on $\Bbb{R}$ we take it to be, this does not guarantee that the same conditions are met for any delta neighborhood on $\Bbb{R}$ which contains regions of two such periods. Thanks for the help! – jinks908 Dec 13 '21 at 19:08

0 Answers0