I am working on some physics problem and got stuck with the following equation: Let $a$ be a very small positive number. Is there a bounded function $F$, $0 \leq F \leq 1$, such that for all $x \in \mathbb{R}$, $$ F(x - a) e^{x a} + (1 - F(x+a)) e^{-x a} = e^{a^2/2}. $$ I had never seen anything like that before. Any references are most welcome.

  • $\begingroup$ You don't want measurability or anything, so the question comes down to whether one can specify values of $F$ on $[0,2a)$ (all between $0$ and $1$) so that $0 \le F \le 1$ always holds when using the functional equation to extend $F$ to all of $\mathbb{R}$. $\endgroup$ Apr 23 at 3:39

1 Answer 1


Denote $F(x)=G(x)e^{x^2/2}$, this multiple is chosen in order to get an equation of the form $G(x+a)-G(x-a)=$ a given function. Indeed, you have $G(x)\in [0,e^{-x^2/2}]$ and $$G(x-a)e^{x^2/2+a^2/2}+e^{-xa}-G(x+a)e^{x^2/2+a^2/2}=e^{a^2/2}\\ G(x+a)-G(x-a)=e^{-(x+a)^2/2}-e^{-x^2/2}.$$ One such function $G=:G_0$ is straightforward: $$G_0(t)=\sum_{k=0}^\infty (-1)^ke^{-(t-ka)^2/2},$$ then we are given that $G-G_0$ is $2a$-periodic. On the other hand, $G(x)$ goes to 0 when $x$ goes to $-\infty$, and so does $G_0$. Thus $G-G_0$ is a periodic function which goes to 0 when $x$ goes to $-\infty$. The only such function is identical 0. Therefore $G\equiv G_0$, and the question reduces to whether $G_0(x)\in [0,e^{-x^2/2}]$ for all $x$. But $$\lim_{n\to +\infty}G_0(x+2na)=\sum_{k\in \mathbb{Z}}(-1)^k e^{-(x-ka)^2/2}$$ depends on $x$ (it is closely related to Jacobi theta-function) and is not in general equal to 0, so alas, your $F$ does not exist.

  • $\begingroup$ Sorry, I made an erroneous edit to your answer, which is now "under review". I don't know how to revert it. $\endgroup$ Apr 23 at 5:41
  • 1
    $\begingroup$ @HjalmarRosengren never mind, I rejected this edit $\endgroup$ Apr 23 at 5:59
  • $\begingroup$ @FedorPetrov Well done! $\endgroup$ Apr 23 at 17:07
  • 2
    $\begingroup$ @mathworker21 medium $\endgroup$ Apr 23 at 20:10

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.