I know that if I have two differentiable functions $f, g$ then the functions $(f + g)$ and $fg$ are also differentiable. I would like to find a way how to argue about the function $h$ where \begin{equation} f(x) = (hg)(x) := h(x)g(x) \quad \text{and } f,g \text{ are differentiable} \end{equation}

For a start I can conclude $h$ is differentiable at all points where $g(x) \neq 0$ since there I can express $h$ as \begin{equation} h = \frac{f}{g} \end{equation}

But for the remaining points I am not sure, my guess is that $h$ is differentiable, any hints how I can make this into a formal argument ? Or am I probably wrong ? In that case, would it help to impose further smoothness on $f$ and $g$, say both are $C^\infty$ ?

Many thanks!


In general the answer is no: Consider $f(x) = x^3$, $h(x) = |x|$ and $g(x) = \text{sign}(x) x^2$. In this case, $f$ is differentiable at $0$ while $h$ is not.

I guess that $C^\infty$ in not enough, but analytic functions should work if $f$ is not constant zero.

  • $\begingroup$ that's a very interesting counterexample, great answer many thanks ! $\endgroup$ – harlekin Feb 1 '12 at 10:44
  • $\begingroup$ I swapped $f$ and $g$ in your answer to match the OP's notation. Please check that I didn't accidentally introduce any mistakes. Oh, and +1. :) $\endgroup$ – Ilmari Karonen Feb 1 '12 at 15:42

No. Let $f(x)=g(x)=0$ for all $x$. There are no functions more regular than constant functions. Then $f=gh$ holds for any function $h$, even if it is nowhere-differentiable.

  • $\begingroup$ of course you are right, thanks! $\endgroup$ – harlekin Feb 1 '12 at 10:44

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.