0
$\begingroup$

Statement:

If $f: \mathbb{R}^n\rightarrow \mathbb{R}^m$ is differentiable at $\underline{x}$, it's continuous at $\underline{x}$.

My proof:

Since $f$ is differentiable at $\underline{x}$, all directional derivatives exist.

Note also that for $\underline{h}=(h_1,\cdots, h_n),$ $$||f(\underline{x}+\underline{h})-f(\underline{x})|| = ||\sum\limits_{i=1}^n f(\underline{x}+\underline{v_i})-f(\underline{x}+\underline{v_{i-1}})||$$ where $v_i = (h_1, h_2, \cdots, h_i, 0, \cdots, 0)$ when $i\neq 0$ and $(0,\cdots,0)$ when $i=0$.

Thus, on an arbitrary direction $i$, since the derivative exists, the function reduced to that direction is continuous

Therefore $\forall \epsilon>0, \exists \delta_i>0$ such that $\forall ||(\underline{x}+\underline{v_i})-(\underline{x}+\underline{v_i})||=||(0,\cdots, 0, h_i, 0, \cdots,0)||=|h_i|<\delta_i$, we have: $$||f(\underline{x}+\underline{v_i})-f(\underline{x}+\underline{v_{i-1}})||<\frac{\epsilon}{n}$$

And finally, let $\delta = \min\{\delta_i\}$ and $\forall \underline{h}\in B_\delta(\underline{x})$, $$||f(\underline{x}+\underline{h})-f(\underline{x})|| = ||\sum\limits_{i=1}^n f(\underline{x}+\underline{v_i})-f(\underline{x}+\underline{v_{i-1}})||<\frac{\epsilon}{n}\cdot n=\epsilon$$

Question:

I'm not sure if my proof is correct, it is unnecessarily more complicated than the textbook proof but I would be grateful if someone could tell me if this was a right way, and if not, what are the problems about it.

Thanks in advance!

$\endgroup$
  • $\begingroup$ You may want to try to mimic the 1D proof. $\endgroup$ – Calvin Khor Feb 23 at 18:06
  • $\begingroup$ Recall that, by sheer definition of differentiability at $x,$ there exists a linear function $u$ such that $f(x + h) = f(x) + u(h) + e(h),$ with $\|e(h)\|/\|h\| \to 0$ as $\|h\| \to 0$ (with $h \neq 0$). Then, make $h \to 0$ and you'll get right away $f(x + h) \to f(x).$ $\endgroup$ – Will M. Feb 23 at 21:45
1
$\begingroup$

No, it is not correct. The restriction of $f$ to every straight line being continuous at $(0,0,\ldots,0)$ does not imply that $f$ is continuous there. Consider, for instance,$$\begin{array}{rccc}f\colon&\mathbb{R}^2&\longrightarrow&\mathbb R\\&(x,y)&\mapsto&\begin{cases}1&\text{ if }y=x^2\text{ and }(x,y)\neq(0,0)\\0&\text{ otherwise.}\end{cases}\end{array}$$

$\endgroup$

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.