Let $f: \RR \longrightarrow \RR$ be a continuous function such that
$f(x+y)=f(x)f(y)$ and $f$ is not the zero function. Prove that $f(x) = a^x$ for
some $a \in \RR$ and any $x \in \RR$.
Proposed by Carlos Torres Pastor in the Telegram group "Retos Matemáticos".
Notice first that from $f$ being a nontrivial solution there exists $x \in \RR$
such that $f(x) \neq 0$, so $f(x) = f(0+x) = f(0)f(x)$ and thus $f(0)=1$. Next,
one can exploit the functional equation inductively to characterize $f$ when
$x$ is an integer:
$$
\begin{alignat*}2
f(x) &= f(1 + (x - 1)) = f(1)f(x-1) \stackrel{\text{IH}}{=} f(1)^x
\qquad&&\text{if $x \in \ZZ^+$,}\\
f(x) &= f(-1 + (x + 1)) = f(-1)f(x+1) \stackrel{\text{IH}}{=} f(1)^x
\qquad&&\text{if $x \in \ZZ^-$,}
\end{alignat*}
$$
where the last step in the second equation follows from the fact that
$$
1 = f(0) = f(x - x) = f(x)f(-x) \implies f(-x) = \frac1{f(x)}.
$$
Clearly $f$ cannot have any roots, since assuming $f(\alpha) = 0$ for some
$\alpha \in \RR$ yields $f(x) = f(\alpha+(x-\alpha)) = f(\alpha)f(x-\alpha)=0$,
which contradicts the assumption that $f$ is not the zero function; likewise,
it is always positive since $f(x)=f(x/2+x/2)=f(x/2)^2 > 0$. Hence, taking
$p/q \in \QQ$ shows that
$$
f(1)^p
= f(p)
= f\underbrace{\left(\frac pq+\cdots+\frac pq\right)}_{q \text{ times}}
= f\left(\frac pq\right)^q
\implies f\left(\frac pq\right) = f(1)^{p/q}
$$
Finally, the definition of the continuity of $f$ states that, for any $x_0 \in
\RR$, $\lim_{x\to x_0} f(x) = f(x_0)$; expanding the previous identity in terms
of $\varepsilon$-$\delta$ and invoking the fact that $\QQ$ is dense in $\RR$ we
see that for any $\varepsilon > 0$ there is a $p'/q' \in \QQ$ such that
$|x_0 - p'/q'| < \delta$ and $|f(x_0) - f(1)^{p'/q'}| < \varepsilon$. Since
$\delta$ gets arbitrarily small, it can be concluded that $f(x_0) = f(1)^{x_0}$
and thus $f(x) = f(1)^x$ for any $x \in \RR$, as it was to be proven.
(From Antonio Roberto Martínez Fernández) The last part of the proof is a
corollary of a much more general result in topology, namely that for continuous
functions $f,g: X\longrightarrow Y$ (where $X, Y$ are topological spaces, and
$Y$ is a Hausdorff space) such that $f(x)=g(x)$ for all $x\in D$, with
$D\subseteq X$ dense in $X$, it holds that $f(x)=g(x)$ for all $x\in X$. In
this particular case, $X=Y=\RR$ and $D=\QQ$.
As far as I am aware, both from Antonio Roberto's comment and from a quick search on the internet, this theorem does not have a standard name in the literature, but rather is part of the mathematical folklore.
As far as I am aware, both from Antonio Roberto's comment and from a quick search on the internet, this theorem does not have a standard name in the literature, but rather is part of the mathematical folklore.