Let $f(x)=e^x \sin x$. Prove that the function is not uniformly continuous in the interval $[0,\infty)$.

What I tried:

Let $x=2\pi k, y=2\pi k + \frac{\pi}{2}$. I have to prove that there is some $\varepsilon>0$ such that $|f(y)-f(x)|\geq \varepsilon$ where $|y-x|<\delta$.

$$|f(y)-f(x)|=\left|e^{2\pi k+\frac{\pi}{2}}\cdot \sin\left(2\pi k+\frac{\pi}{2}\right) - e^{2\pi k}\cdot \sin(2\pi k )\right|.$$

I know that $\sin (2\pi k)=0, \sin\left(2\pi k + \frac{\pi}{2}\right)=1$, therefore: $$|f(y)-f(x)|=|e^{2\pi k+\frac{\pi}{2}}\cdot 1 - e^{2\pi k}\cdot 0|=|e^{2\pi k+\frac{\pi}{2}}|.$$

Now it seems like all I have to do is select some $\varepsilon>0$, for instance $\varepsilon=1$ and I'm finished.

Am I?

Is my solution correct?

  • $\begingroup$ Not quite, because $|x_k-y_k|=\pi/2$, which is not small enough. $\endgroup$ – uniquesolution Apr 11 '17 at 13:13

Not quite. You need to prove that for some $\epsilon$, no matter how small you make $\delta$, you can find $x, y$ such that $|y - x| < \delta$ and $|f(y) - f(x)| \geq \epsilon$. One approach that could work: note that for $\delta > 0$ small enough, $\sin (2\pi k + \delta) > \delta/2$, so let's choose, say, $x = 2\pi k$ for some very large integer $k$ and $y = 2\pi k + \delta/2$. Obviously $|y - x| < \delta$, and $$|f(y) - f(x)| = \sin (\delta/2) e^{2\pi k + \delta/2} > \frac{\delta}{4} e^{2\pi k + \delta/2} $$ which may be made arbitrarily large, no matter how small $\delta$ is, by choosing $k$ large enough.

A reminder: in logical symbols, the criterion for uniform continuity on an interval $I$ is $$(\forall \epsilon > 0) (\exists \delta > 0) (\forall x \in I)(\forall y \in I) (|y-x| < \delta \implies |f(y) - f(x)| < \epsilon)$$ (unlike ordinary continuity, where $\forall x$ and $\exists \delta$ are reversed); the negation of this, by de Morgan's laws, is $$(\exists \epsilon > 0) (\forall \delta > 0) (\exists x, y \in I) (|y - x| < \delta \wedge |f(y) - f(x)| \geq \epsilon)$$ (recall that $A \implies B$ is equivalent to $(\neg A) \vee B$).

  • $\begingroup$ Thanks! how do we know for sure that $$|f(2\pi k + \delta) - f(2\pi k)| > \frac{\delta}{2} e^{2\pi k + \delta} $$? $\endgroup$ – Alan Apr 11 '17 at 13:35
  • 1
    $\begingroup$ $f(2\pi k) = 0$ and $f(2\pi k + \delta) = \sin(\delta) e^{2\pi k + \delta}$, and $\sin(\delta) > \delta/2$ for $0 < \delta \ll 1$. $\endgroup$ – Connor Harris Apr 11 '17 at 13:36
  • $\begingroup$ Is there a way of solving this using derivative? $\endgroup$ – Alan Apr 11 '17 at 13:52
  • 1
    $\begingroup$ @Alan it is sufficient to know that the derivative is unbounded. $\endgroup$ – Umberto P. Apr 11 '17 at 13:55
  • 1
    $\begingroup$ @Alan The difference between strict and non-strict inequality is immaterial, but I'll edit my answer above to adjust to your version. $\endgroup$ – Connor Harris Apr 11 '17 at 15:03

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.