# Uncertainty principle (really for Mellin, but never mind that!)

Is there a smooth funtion $f:\mathbb{R}\to \mathbb{C}$ such that

(a) $f(x)$ decreases faster than $e^{-e^x}$ when $x\to \infty$,

(b) $\widehat{f}(t)$ decreases faster than $e^{-|t|}$ when $t\to \pm\infty$?

Note there is no restriction on $f(x)$ for $x\to -\infty$ except that it decay fast enough for the Fourier transform $\widehat{f}$ to be well defined. The function $f_\epsilon(x) = e^{\epsilon x - e^x}$, $\epsilon>0$, decays almost a fast as $e^{-e^x}$ for $x\to \infty$ and $\widehat{f_\epsilon}(t)$ decays roughly as $e^{-|t|}$ for $t\to \pm \infty$ -- so I am really asking whether one can defeat $f_\epsilon$ on both the physical and Fourier aspect.

PS. Yes, this does come from a Mellin transform, as the double exponential probably gives away.

• I can, by the way, do much better than (a) while doing not uch worse than (b): take $f(x) = e^{-x^2}$. Nov 8 '12 at 16:58
• How is $e^{-x^2}$ better than (a)? Also, $\hat f(t)$ decaying exponentially means that $f(x)$ extends to an analytic function on a horizontal strip and is $L^2$ on horizontal lines. Then the question comes down to finding a function analytic on a strip such that it decays very quickly on the positive real axis, and Phragmen Lindelof gives you bounds on how quickly it can decay (and how wide the strip may be) Nov 8 '12 at 19:55
• Sorry, forgot to change variables. Meant $f(x)=e^{e^{−2x}}$. Agreed, it's not much better than (a), and the Fourier transform is slightly worse than (b). – H A Helfgott 0 secs ago Nov 9 '12 at 2:21
• I know Phragmen-Lindelof only as a tool to get subconvexity, so I am not familiar with the argument you allude to - could you give us a reference (or a one-line explanation)? Nov 9 '12 at 11:15
• There are some nice consequences of Phragmen-Lindelof (and its proof) along these lines in Titchmarsh's Theory of Functions. For example there's the following result of Carlson. Suppose $f(z)$ is holomorphic in some sector of interior angle $\theta$, is exponentially bounded $|f(z)|\ltlt e^{k|z|}$ and exponentially decays on the boundary. $\exp$ gives an example for $\theta<\pi$ and the result is that if $\theta=\pi$ then $f(z)\equiv0$. In fact, having such a non-trivial function would give you a stronger Phragmen-Lindelof principle that could prove that the exponential function does not e Nov 12 '12 at 1:36

How about $$\frac{e^{-e^x}}{1+\epsilon x^2}$$ If you compute the Fourier transform you can shift the contour to height $\pm\pi/2$ to get an $e^{-|t|}$ times something decaying to 1, by Riemann-Lebesgue lemma
Or you can look at a shift of your original example: $f(x+\log A)$ to get something on the lines of $$e^{-A e^x} e^{-B x^2} e^{C x}$$
• You can also replace the $1+x^2$ with any function growing in the strip, the best example being along the lines of $e^{A e^x}$. In fact, if you translate your original example you get something decaying much more quickly, without changing the modulus of the Fourier transform: $$e^{-A e^x}e^{B x}$$ Nov 9 '12 at 18:48
• Thanks for this, but it's like the example I gave ($e^{e^{Cx}}$) - these variations due to rescaling improve one of the two aspects, but not the other (though one should be happy if they don't degrade it). I was wondering whether one can improve both... Nov 10 '12 at 16:01
• $e^{-e^{C x}}$ ends up hurting one of the bounds, but I think with $e^{-A e^x}$ you help the first super-exponentially without hurting the second at all, and when you add in the $e^{-x^2}$ you end up beating the second bound by something like $e^{-(\log t)^2}$ Nov 10 '12 at 20:03