Is it true that $\lim_{x \to a}\frac{f(x)-f(a)}{x-a}=\pm\infty \implies \lim_{x \to a}f'(x) = \pm\infty$?

Here, $f$ is a function defined on some open interval $I$, and $a\in I$. Assume $f$ is continuous at $a$ and differentiable around $a$.

I can't for the life of me see how to prove\disprove this implication, but my gut feeling is that it's false. Any guidance is greatly appreciated.

  • $\begingroup$ isn't that definition of limit? the (ε, δ) should provide that the function is strictly increasing and must go unbounded to $ \infty $ $\endgroup$ – user29418 Jan 7 at 1:28
  • $\begingroup$ Derivative of a differential function need not be continuous $\endgroup$ – Sorfosh Jan 7 at 6:00

The claim is false. Consider $$ g(x)=\sqrt{x}\sin\frac{1}{x}+x^{1/4},\qquad x>0, $$ and define $f$ on $\mathbb R$ by $$ f(x)=\begin{cases}g(x) & \text{for }x>0, \\ 0 & \text{for }x=0, \\ -g(-x) & \text{for }x<0.\end{cases} $$ Then $f$ is continuous everywhere, differentiable in $\mathbb R\backslash\{0\}$ and satisfies $f(-x)=-f(x)$. Moreover, for $x>0$, \begin{align*} \frac{f(x)}{x}=\frac{1}{\sqrt{x}}\sin\frac{1}{x}+\frac{1}{x^{3/4}}\geq-\frac{1}{x^{1/2}}+\frac{1}{x^{3/4}}\longrightarrow+\infty,\quad\text{as }x\to0+. \end{align*} Due to the symmetry, the same is true for $x<0$ and $x\to0-$. Thus, $f(x)/x\to+\infty$ as $x\to0$.

Now, for $x>0$, $$ f'(x)=\frac{x^{3/4}+2 x \sin \left(\frac{1}{x}\right)-4 \cos \left(\frac{1}{x}\right)}{4 x^{3/2}}. $$ However, the limit $\lim_{x\to0+}f'(x)$ does not even exist. For $x_n:=1/(n\pi)$ we have \begin{align*} f'(x_n)=\frac{\pi ^{3/4}}{4 \left(\frac{1}{n}\right)^{3/4}}-\frac{\pi ^{3/2} (-1)^n}{\left(\frac{1}{n}\right)^{3/2}}, \end{align*} and so $f'(x_{2n})\to-\infty$, while $f'(x_{2n+1})\to+\infty$.


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.