I'm trying to find out if $$ f(x,y)=\cos\left(\sqrt{xy}\right) $$ is uniformly continuous on the set $\{(x, y)\in\mathbb{R}^2 : x\geq0, y\geq 0\}$. The theorems I have available to use for this are for showing that the function is uniformly continuous (and merely sufficient rather than necessary), but I happen to know that it is not uniformly continuous in this case. But I can't show why.

I've tried working with the definition to get: $$ |(x, y)-(x_0, y_0)|=\sqrt{(x-x_0)^2+(y-y_0)^2}<\delta $$ but then I need to show that this does not imply that $$ |f(x, y)-f(x_0, y_0)|=|\cos\sqrt{xy}-\cos\sqrt{x_0y_0}|<\epsilon. $$

But how would I do that? Some hints would be appreciated. Thanks.


Look at special points. Say points with the same $x$-coordinate and a small difference in the $y$-coordinate,

$$f(x,y+h) - f(x,y).$$

Looking at $\sqrt{xy}$, you should be able to see for what choice of $y$ you get the largest difference.

You need one $\varepsilon > 0$ so that for every $\delta > 0$ there are points less than $\delta$ apart so that $\lvert f(x_1,y_1) - f(x_2,y_2)\rvert \geqslant \varepsilon$.

$f(2/\delta,0) - f(2/\delta, \delta/2) = 1 - \cos 1$, and $\lVert (2/\delta,0) - (2/\delta,\delta/2)\rVert = \delta/2$.

| cite | improve this answer | |
  • $\begingroup$ So, please tell me if I'm misunderstanding: irrespective of the choice of $\lambda (>0)$, we can find an $\epsilon$ such that $|f(x_1, y_1)-f(x_2, y_2)|\geq \epsilon$. For example, $\epsilon=0.25$. But, shouldn't this be for all $(x, y)$ in the domain? I can see that the for all $y$ part is satisfied since we can set $y=0$ which is the maximum. But what about $x$? Hope it makes sense... "Maximum" in the sense that is when the greatest difference is achieved. $\endgroup$ – hejseb Jan 31 '14 at 13:52
  • $\begingroup$ No actually, it's only for some $(x, y)$ --- it's of course uniform continuity that should be for all points in the domain. $\endgroup$ – hejseb Jan 31 '14 at 13:57
  • $\begingroup$ We need to find one $\varepsilon > 0$, so that for all $\delta > 0$ we find points with $\lVert a-b\rVert < \delta$, but $\lvert f(a) - f(b)\rvert \geqslant \varepsilon$ to show that $f$ is not uniformly continuous. In short: Yes to your second comment. $\endgroup$ – Daniel Fischer Jan 31 '14 at 14:06

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.