Decay of the Fourier transform once again

190 Views Asked by At

I've seen many questions and answers about the relation between the smoothness of some function and the rate at which its Fourier transform decays. However, I haven't found the answer to the following question:

If $f \in C^1(\mathbb{R}) \cap L^1(\mathbb{R})$, i. e. $f$ is continuously differentiable and integrable on $\mathbb{R}$, is it true that its Fourier transform is $\mathcal{O}(t^{-1})$ at infinity?

Note that we don't require $f'$ to be integrable.

If $f'$ were integrable this could be shown by the relation $\hat{f}(t) = it \widehat{f'}(t)$ which in turn follows from integration by parts. In my situation however the fourier transform of $f'$ does not even exist (or does is it in some sense?).
I've thought about approximating the original function $f$ by functions with compact support or Schwartz-functions but the best I can get by this method is that ${f}(t) \leq C \cdot t^{-1} + \epsilon$ for every $\epsilon > 0$ and for some constant $C > 0$ which depends on $\epsilon$.