Erdős Theorem on Monotone Multiplicative Functions

In number theory, a totally multiplicative function is a function $f: \mathbb{N} \rightarrow \mathbb{R}$ such that $f(mn) = f(m)f(n),$ for all $m, n \in \mathbb{N}$.

In 1946, Paul Erdős proved a lovely result1 about such functions, which we are going to prove!


Theorem. If $f: \mathbb{N} \rightarrow \mathbb{R}$ is increasing and totally multiplicative, then $f(n) = n^{\alpha}$, for some $\alpha \in \mathbb{R}$.

Proof. Suppose that $f(2) = 2^{\alpha}$, and take $n > 2$ such that $f(n) = n^{\beta}$. Then for any $\ell \in \mathbb{N}$, we can write

$$ 2^a < n^{\ell} < 2^{a + 1}, $$

for $a = \lfloor \log_2(n) \ell \rfloor$. Since $f$ is increasing, evaluating the function at each of the above must still satisfy the inequality

$$ \begin{aligned} 2^{\alpha a} < n^{\beta \ell} &< 2^{\alpha(a + 1)} \\ \implies \lfloor \log_2(n) \ell\rfloor < \frac{\beta}{\alpha} \log_2(n) \ell &< \lceil \log_2(n) \ell\rceil. \end{aligned} $$

This implies the inequality

$$ \log_2(n) \ell \left|\frac{\beta}{\alpha} - 1\right| \leq 1. $$

But this has to hold for all $\ell \in \mathbb{N}$, and thus we must have $\beta = \alpha$, and we’re done.

  1. He actually proved a slightly different result about additive functions, which you can imagine the definition of, but the results are equivalent by taking logs. See Erdos, P., 1946. On the distribution function of Additive Functions. The Annals of Mathematics, 47(1)↩︎