7

Let $(x_n)_{n\in\mathbb{N}}$ be a strictly increasing sequence of natural numbers. Does there exist an irrational number $\alpha$ such that the set $\{ \alpha x_n \pmod 1: n\in\mathbb{N} \}$ is dense in $[0,1]?$

I came up with the question myself. I don't have many great ideas to answer the question, and yes I am aware of theorems such as Dirichlet's Approximation theorem as well as related stronger results like Hurwitz's theorem. But I don't see how to apply any of those to this problem, or what other tools to use.

Of course the result is true if $(x_n)_{n\in\mathbb{N}} = \mathbb{N}$ or $=k\mathbb{N},\ k\in\mathbb{N}$ (any irrational $\alpha$ will do). Other than that, I'm not sure.

Mittens
  • 46,352
Adam Rubinson
  • 24,300

1 Answers1

3

The answer is yes. Here is a rather well known metrication theorem:

Theorem: Let $a:\mathbb{N}\rightarrow\mathbb{Z}$ be an injective function. The sequence $(x a(n):n\in\mathbb{N})$ is uniformly distributed (u.d.) mod $1$ for Lebesgue almost all $x\in\mathbb{R}$.

A proof of this result can be found in the classical book Kuipers, L. and Niederreiter, H., Uniform Distribution of Sequences, Dover Publications Inc., New York, 2006 (reprint of the 1974 edition). pp. 32-33.


More general results (proved by Koksma) that include the situation of the OP are the following:

Theorem: Suppose $u_n:[a,b]\rightarrow\mathbb{R}$ is a sequence of functions in $\mathcal{C}^1[a,b]$ such that for any $n\neq m$, $u'_n-u'_n$ is monotone and $$\|u'_m-u'_n\|_\infty\geq K$$ for some constant $K>0$. Then, for (Lebesgue) almost all $x\in[a,b]$, the sequence $(u_n(x))$ is uniformly distributed (u.d.) mod $1$.

This results yields the following well known results:

Corollary 1: Let $c>0$ and $F:\mathbb{N}\rightarrow(0,\infty)$ such that $$|F(n)-F(m)|\geq c\delta_{nm}$$ For any $a\in\mathbb{R}\setminus\{0\}$, the sequence $\big(ax^{F(n)}:n\in\mathbb{N}\big)$ is uniformly distributed mod $1$ for Lebesgue almost all $x\geq1$.

Corollary 2: For (Lebesgue) For $a\neq0$, the sequence $(ax^n:n\in\mathbb{N})$ is u.d. mod $1$ for (Lebesgue) almost all $x\geq 1$.

Corollary 3: Let $c>0$. If $(\lambda_n:n\in\mathbb{N})\subset\mathbb{R}$ is a sequence such that $|\lambda_n-\lambda_m|\geq c\delta_{nm}$, then $(x\lambda_n:n\in\mathbb{N})$ is u.d. mod $1$ for (Lebesgue) almost all $x\in\mathbb{R}$.

The well known exceptions to Corollary 1 are the powers of Pisot-Vijayaraghavan numbers (PV-numbers). An algebraic number $\xi\in\mathbb{C}$ with $|\xi|>1$ is a P.V number if there is a monomial $p(z)=z^n+a_{n-1}z^{n-1}+\ldots + a_0$ with $a_j\in\mathbb{Z}$ such that $p(\xi)=0$, $p$ is irreducible over $\mathbb{Q}$ and all other roots $\xi'$ of $p$ satisfy $|\xi'|<1$. Let $\{\;\}$ denote the fractional part function. It is known that

Theorem: If $\xi$ is a real PV-number, then the only limit points of the sequence $(\{\xi^n\}:n\in\mathbb{N})$ are $0$ and $1$.

See Kuipers (idem) pp. 34-36

Mittens
  • 46,352
  • Is it established that the ONLY exceptions to corollary 1 are PV numbers? I thought that was just a conjecture at present – QC_QAOA May 04 '25 at 00:55
  • 1
    @QC_QAOA: I just checked the classical reference. I am not saying that PV numbers are exceptions. Only that in fact case, with $a=1$, the sequences $(a\theta^n) $ is not Ud mod 1. – Mittens May 04 '25 at 01:13