3

I'm interested in proving a possible generalization of Dirichlet's approximation theorem, where all integers involved are odd. The approximation theorem states that, for any irrational number $\alpha$, there are infinitely many integer pairs $(p,q)$ obeying: $$\left|p-q\alpha\right|<\frac{1}{q}$$

This is equivalent to saying $|p/q-\alpha| < 1/q^2$. In particular, the convergents $p/q$ of $\alpha$ satisfy this inequality, and moreover, at least every other convergent has odd denominator, so there are infinitely many pairs where $q$ is odd. Is it possible to prove a similar result where $p,q$ are restricted to be odd? I would accept any result which approximates $|p/q-\alpha|$ within $\mathcal{O}(1/q^2)$. Equivalently, my question is the following.

For every irrational $\alpha$, can we find infinitely many pairs of odd integers $p_n, q_n$ such that $|p_n-q_n\alpha| = \mathcal{O}(1/q_n)$?

I've been studying various proofs of the approximation theorem to try to get a generalization, but no luck yet. There might be some insight in the equidistribution theorem, but I haven't been able to figure anything out yet. A positive answer to this question would complete this partial answer I posted yesterday. For my purposes, I'm interested in the specific case where $\alpha=\frac{\pi}{2}$, hence $\alpha$ is transcendental, if that helps. I might also accept a slightly weaker result, like finding odd integers where $|p_n-q_n\alpha| = \mathcal{O}(q_n^{-1+\varepsilon})$ for some sufficiently small $\varepsilon>0$. It seems intuitively obvious that something like this should work, but I couldn't figure it out.

J. W. Tanner
  • 63,683
  • 4
  • 43
  • 88
Jade Vanadium
  • 5,046
  • 10
  • 25
  • Yes, it is possible to find infinitely many pairs of odd integers ( p_n ) and ( q_n ) such that ( |p_n - q_n \alpha| = \mathcal{O}(1/q_n) ), depending on the arithmetic properties of the real number ( \alpha ). – zeraoulia rafik Dec 02 '24 at 21:48
  • @zeraouliarafik Hi, thanks! What sort of properties are relevant, and do you have any links or references to results like that? – Jade Vanadium Dec 02 '24 at 21:51
  • It's really difficult to control the parity of $p$, I don't know how – Chris Sanders Dec 03 '24 at 00:37
  • 1
    See e.g. W.T. Scott, "Approximation to real irrationals by certain classes of rational fractions", Bull. Amer. Math. Soc. 46, 124-129 https://projecteuclid.org/journals/bulletin-of-the-american-mathematical-society/volume-46/issue-2/Approximation-to-real-irrationals-by-certain-classes-of-rational-fractions/bams/1183502433.full – Robert Israel Dec 03 '24 at 03:20
  • @RobertIsrael Oh, this is exactly what I needed! If you post an answer summarizing the results (just the introduction, if you want), I would accept it. – Jade Vanadium Dec 03 '24 at 03:32

1 Answers1

1

[Posted as an answer, at OP's request]

W. T. Scott, "Approximation to real irrationals by certain classes of rational fractions", Bull. Amer. Math. Soc. 46, 124-129, shows that if $\alpha$ is irrational there are infinitely many rationals $p/q$ in each of the three cases $(p,q) = (odd, odd)$, $(even,odd)$ and $(odd,even)$ such that $$ \left|\alpha - \frac{p}{q} \right| \le \frac{1}{q^2}$$

Robert Israel
  • 470,583