Question: Given $f:\mathbb{R}^2 \to \mathbb{R}$ defined by $f(x, y) = \begin{cases} x, & \text{if $y=x^2$} \\ 0, & \text{otherwise} \end{cases}$, show $f$ is continuous at $(0, 0)$.

Attempt: I have proven that $f$ is not differentiable at $(0, 0)$, which means I can't imply continuity from it being differentiable.

I know that I must show $\lim_{(x,y) \to (0, 0)} f = f(0, 0)=0$, but am not experienced in proving the limits of multivariable functions (at least not for ones that you can't apply the squeeze theorem to).

Or is there a way to utilise the continuity of the directional derivatives?

I know this is a pretty weak attempt but I am really stuck, and any help would be greatly appreciated.

  • 1
    $\begingroup$ Do you really need derivatives though? Can't you apply squeeze theorem with $g(x,y)=|x|$ and $h(x,y)=-|x|$ $\endgroup$
    – Anvit
    Jun 1, 2020 at 11:46

2 Answers 2


When dealing with multivariate functions like this one, you consider $(x,y)\to (0,0)$.

This means $x\to 0$ and $y\to 0$.

  • Most of the time you want to show $|f(x,y)-\ell|<k\ |x|^\alpha|y|^\beta\to 0$ for some $\alpha,\beta > 0$, since $|x|^\alpha\to 0$ and $|y|^\beta\to 0$.

Note that it is not mandatory to have both $x$ and $y$ on the RHS (i.e. $\alpha=0$ or $\beta=0$ is ok).

You can of course also have some other combination of $|x|^\alpha$ and $|y|^\beta$ (a sum for instance).

  • In some cases using polar coordinates is helpful, you get $|f(x,y)-\ell|<\phi(\theta)\ |r|^\gamma\to 0\ \ $ provided $|\phi(\theta)|<M$ is bounded.

Since $r^2=x^2+y^2\to 0$ then $|r|\to 0$.

  • Another technique is to set $t(x,y)=\frac yx$ or $y=tx$ to get $|f(x,y)-\ell|<\phi(t)|x|^\alpha\to 0\ $ provided again that $|\phi(t)|<M$ is bounded.

In some cases, $\alpha=0$ and you need to show that $\phi$ itself tends to zero.

In our case, we are in the best conditions possible since $\ell=0$ and $|f(x,y)|\le|x|\to 0$.

So this is over and you can claim $\lim\limits_{(x,y)\to (0,0)}f(x,y)=0$, making $f$ continuous in $(0,0)$.



Let $\epsilon >0$ be given.

Need to show that for $\epsilon >0$ there is a $\delta$ s.t.

$\sqrt{x^2+y^2}<\delta$ implies


Choose $\delta=\epsilon$.


$0 \le |f(x,y)-0| \le |x|=√x^2 \le$

$\sqrt{x^2+y^2} <\delta =\epsilon$.

  • $\begingroup$ May I ask where you got the $\sqrt{x^2 + y^2}$? $\endgroup$
    – Viv4660
    Jun 1, 2020 at 12:12
  • $\begingroup$ He is using $||(x,y)||_2$ the euclidean norm, but in $\mathbb R^n$ all norms are equivalent, so it doesn't really matter, you choose $1-$norm $|x|+|y|\to 0$ or $\infty-$norm $\max(|x|,|y|)\to 0$ or $2-$norm $|r|\to 0$. $\endgroup$
    – zwim
    Jun 1, 2020 at 12:16
  • $\begingroup$ Sure. $\sqrt{x^2} \le \sqrt{x^2+y^2}$, added $y^2 \ge 0$, inequality is ok. Want to get an upper bound for $|f(x,y)|$ that relates $\delta$ and $\epsilon$.Your thoughts? $\endgroup$ Jun 1, 2020 at 12:16

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.