$f'(x)$, $f''(x)$, $f^{(3)}(x)$, ..., $f^{(n)}(x)$ all exist $\forall x \in I$, an interval in $\mathbb{R}$ but $f^{(n+1)}(x)$ does not exist for any $x \in I$

Can such a function exist? I think it can't but can't think of the reason why. I can construct a function with countably infinite points in I where the function is only differentiable n times, but, it seems impossible to have every point in the interval, I, have this property.

  • $\begingroup$ Have you heard of Weierstrass' nowhere differentiable, continuous function? (Integrate it.) $\endgroup$ – David Mitra Apr 18 '18 at 11:23
  • $\begingroup$ Yes. Is there a way to smooth it out so it can be differentiated a finite number of times before failing at every point? $\endgroup$ – futurebird Apr 18 '18 at 11:24

Yes. Take the Weierstrass function $f$. It is differentiable nowhere. If $F$ is a primitive of $f$ (which exists, since $f$ is continuous), then $F'$ exists, but $F''$ exists nowhere.


Take any continuous, nowhere differentiable function (e.g. Weierstrass function). Then integrate it $n$ times. First time you can do it, as the function is continuous. Later, you are integrating differentiable, so continuous, functions.

  • $\begingroup$ How do I know the first anti-derivative of the Weierstrass function is also continuous? I understand why I can integrate it once, but am still trying to understand doing it more than once. $\endgroup$ – futurebird Apr 18 '18 at 11:35
  • 1
    $\begingroup$ The integral of a continuous function is differentiable, so is continuous and hence integrable. $\endgroup$ – B. Mehta Apr 18 '18 at 11:44
  • $\begingroup$ If $F(x) = \int_0^x f(t)\;dt$, then $F'(x) = f(x)$ provided $f$ is continuous at the point $x$. Perhaps your calculus textbook contains this fundamental theorem. $\endgroup$ – GEdgar Apr 18 '18 at 11:49

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.