1

For any integer $n$, consider its sequence of residues, $n\bmod1, n\bmod2, n\bmod3, \ldots$, as a function $f:\mathbb N^+\to\mathbb N$. Such a function has the following properties: $$\begin{gather} 0\le f(m)<m, \tag{1} \\ f(km)\bmod m=f(m). \tag{2} \end{gather}$$

Does there exist a function that has properties $(1,2)$ but is not identical to $m\mapsto n\bmod m$ for any fixed $n$? If so, are there uncountably many such functions?

(This may be an obvious "number theory 101" problem, but I know zero number theory. It came up when I was trying to answer whether there are functions that satisfy the tangent addition rule $f(a+b)=\frac{f(a)+f(b)}{1-f(a)f(b)}$ that are not linear transformations of the tangent function.)

1 Answers1

1

It is easy to construct a counterexample by induction and a diagonal argument (in fact, a little more work shows that there are uncountably many counterexamples).

In the sequel, $f$ is always a map $A\to {\mathbb N}$ (where $A$ is a subset of ${\mathbb N}^{+}$). I say that $f$ is Rahul iff it satisfies $0\leq f(m) < m$ and $f(km) \equiv f(m) \ ({\sf mod}\ m)$ whenever $m\in A$ and $km\in A$.

Trivially, we have :

Remark 1. A map $f:{\mathbb N}^{+} \to {\mathbb N}$ is Rahul iff the restriction $f_n$ of $f$ to $\lbrace 1,2,\ldots, n\rbrace$ is Rahul for every $n\geq 1$.

This will allow us to construct Rahul maps step by step. Indeed, by the Chinese remainder theorem we have

Remark 2.1. Let $f: \lbrace 1,2,\ldots, n\rbrace \to {\mathbb N}$ be a Rahul map. Then $f$ admits at least one Rahul extension $g:\lbrace 1,2,\ldots, n+1\rbrace \to {\mathbb N}$.

And trivially, we have

Remark 2.2. If $n+1$ is prime, any extension of $f$ to $\lbrace 1,2,\ldots, n+1\rbrace$ such that $f(n+1)\in \lbrace 0,1,2,\ldots, n\rbrace$ is still Rahul (so that $f$ has $n+1$ Rahul extensions to $\lbrace 1,2,\ldots, n+1\rbrace$).

Let $(a_k)_{k\geq 1}$ be any enumeration of $\mathbb Z$ (for example, $a_n=-\frac{k}{2}$ when $n$ is even and $a_k=\frac{k-1}{2}$). Denote by $(p_k)_{k\geq 1}$ the sequence of primes, so that $p_1=2,p_2=3\ldots$.

Iterating remark 2, we can construct a Rahul map on ${\mathbb N}^{+}$ such that $f(p_k) \neq (a_k \ {\sf mod}\ p_k)$ for every $k\geq 1$ : when extending from $\lbrace 1,2,\ldots, n\rbrace$ to $\lbrace 1,2,\ldots, n+1\rbrace$, if $n+1$ is not prime we extend by Remark 2.1, and if $n+1$ is prime, $n+1=p_k$, we use Remark 2.2 and set for $f(n+1)$ any value $v$ such that $v \neq (a_k \ {\sf mod}\ p_k)$ (there are $n$ such values by Remark 2.2.).

Then by construction $f$ is distinct from the map $m \mapsto (a_k \ {\sf mod}\ m)$ for every $k\geq 1$, so $f$ is a counterexample.

Ewan Delanoy
  • 63,436