1

$$f\left(x,y\right)=\begin{cases}\sqrt{\left|xy\right|}\sin\left(\frac{1}{xy}\right)&xy\ne 0\\ 0&xy=0\end{cases}$$
First of all, the question:
If the partial derives at $(0,0)$ exist, can I say automatically that it is continuous at $(0,0)$? I thought of proving that way since the question before was to show the partial derives existed at that point.
If not, then I tried proving it, can you say please if it is good?
For all $\epsilon>0$, there exist $\delta>0$, such that: $0<||(x,y)-(0,0)||<\delta$ which concoludes: $|f(x,y)-f(0,0)|<\epsilon$
So from $\delta$ I know: $||x||<\delta$ and that way with $y$ also.
now:
$|f(x,y)-f(0,0)| = |f(x,y)| = \left|\sqrt{xy}\cdot \sin\left(\frac{1}{xy}\right)\right| \le |\delta \cdot \sin\left(\frac{1}{xy}\right)| \le \delta $ and thus if we choose $\delta = \epsilon$, we will receive truth.
I have another solution which the professor said, but that is my try.

We know $||(x,y)|| < \delta$, thus:
$||x|| < \delta$, so with $\sqrt {||x||}$, it is $\sqrt \delta$
$||y|| < \delta$
or basically ( but have not used ): $x^2+y^2 < \delta$

Mittens
  • 46,352
  • 1
    Use the definition of continuity at $(0,0),$ it is easy. – Viera Čerňanová Dec 13 '22 at 09:10
  • Yea, i used the definition, that is how I reached my conclusion, but nevertheless, is my proof good? – LearningToCode Dec 13 '22 at 09:11
  • It is good but you sould say how you got $\sqrt {|xy|} \leq |(x,y)|$ – Kavi Rama Murthy Dec 13 '22 at 09:13
  • 1
    Oh, I will edit since here latex is problem a little, one minute. – LearningToCode Dec 13 '22 at 09:14
  • @geetha290krm added now, hoping it is understandable. – LearningToCode Dec 13 '22 at 09:16
  • And regards to my question, about partial derives and continuity, is it true what I said? or I was just blabbering my mouth. – LearningToCode Dec 13 '22 at 09:18
  • It makes little sense. Is $x$ a real number or a vector? You should not use $x$ for the vector $(x,y)$ – Kavi Rama Murthy Dec 13 '22 at 09:19
  • Oh sorry, x and y are real number. f(x,y) is basically X and y I forgot to mention maybe,$f:R^2->R$ – LearningToCode Dec 13 '22 at 09:20
  • Existence of partial derivatives does not implie continuity. Take for example $\frac{x*y}{x^2+y^2}$, and 0 at (0,0), has partial derivative at $(0,0)$ but is not continuous, see https://math.stackexchange.com/q/87688/1123744 – linkja Dec 13 '22 at 09:21
  • @linkja ohh thanks, I didnt think it was false, there are too many ifs on partial derives, its very good, I will also take notice of the function for future refute examples. Thanks!! – LearningToCode Dec 13 '22 at 09:23
  • In general the following implications are true in $R^n$: total differential exists $\implies$ all direction derivatives exist $\implies$ partial derivatives exist, but only total differential exists $\implies$ continuity, the others are not enough – linkja Dec 13 '22 at 09:29
  • @linkja Oh I see... Thanks!! Regarding my solution, do you think its good? with the $\delta$ Because geetha said its good, but when I added something, he have not replied yet, so I want to make sure :) – LearningToCode Dec 13 '22 at 09:57

2 Answers2

1

I apply the $\delta$-$\varepsilon$ technique when I am really desperate, or the question requires using that method.

The function is continuous at each point $(x_0,y_0)$ such that $x_0y_0\neq 0,$ as it is represented as a composition, multiplication and division of continuous functions.

Concerning $x_0y_0=0$ we have $$|f(x,y)-f(x_0,y_0)|=|f(x,y)|=\begin{cases} 0 & xy=0\\ \sqrt{|xy|}\left |\sin{1\over xy}\right | & xy\neq 0 \end{cases}\ \le \sqrt{|xy|}$$ When $(x,y)\to (x_0,y_0),$ then $xy\to x_0y_0=0.$ Hence $\sqrt{|xy|}\to 0.$ This implies $|f(x,y)-f(x_0,y_0)|\to 0,$ when $(x,y)\to (x_0,y_0).$ In particular the case $x_0=y_0=0$ is covered.

  • Oh, great proof to be honest. Would have never thought of it. About the desperation, problem is, I know how to prove mostly with epsilon delta hehe, so to me numbers are desperate :) Thanks! – LearningToCode Dec 13 '22 at 10:17
0

Since $(|x|-|y|)^2\ge 0\iff |xy|\le\frac 12(|x|^2+|y|^2)\ $ we get

$$\sqrt{|xy|}\le \frac 1{\sqrt{2}}\sqrt{|x|^2+|y|^2}=\frac 1{\sqrt{2}}\sqrt{|x-0|^2+|y-0|^2}=\frac 1{\sqrt{2}}\Big\lVert(x,y)-(0,0)\Big\rVert_2$$


Also from $|\sin|\le 1$ we have $|f(x,y)|\le \sqrt{|xy|}$ it's quite straightforward to conclude.

I think this was probably the intent of the exercise to make you use the comparison between $\sqrt{|xy|}$ and the Euclidean norm.

zwim
  • 29,833
  • God lord, that proof is too advanced for me hehe. I know norm 2 only from fourier analysis, no idea what it is in calculus. About $|f(x,y)| \le \sqrt{|xy|}$, they did that also in their proof, I guess so :) ( If I could upvote, I would have, sorry :) ) – LearningToCode Dec 13 '22 at 11:40