MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

The Duffin-Schaeffer conjecture is a conjecture in metric number theory, which asks for a given function $f : \mathbb{R} \rightarrow \mathbb{R}^+$ the measure of the set of real numbers $\alpha$ such that the inequality $$\displaystyle \left | \alpha - \frac{p}{q} \right| < \frac{f(q)}{q}$$ has infinitely many solutions in integers $p,q$ with $\gcd(p,q) = 1$. The conjecture asserts that the set of solutions $\alpha$ has full measure if and only if $\displaystyle \sum_{q=1}^\infty \frac{f(q)}{q} \phi(q)$ diverges, where $\phi(q)$ is the Euler totient function.

Let the set of solutions (which depends on the function $f$) be denoted $E_f$. Then it is known (Haynes, Pollington, Velani that a sufficient condition for $m(\mathbb{R} \setminus E_f) = 0$ is for $\displaystyle \sum_{q=1}^\infty \left(\frac{f(q)}{q}\right)^{1 + \epsilon} \phi(q) = \infty$. My question concerns other possible sufficient conditions. In particular, we know (from Duffin and Schaeffer themselves) that it is not sufficient for $\displaystyle \sum_{q=1}^\infty f(q) = \infty$. Is it sufficient for the sum $\displaystyle \sum_{q=1}^\infty \frac{f(q)}{q^\epsilon}$ to diverge for any $\epsilon > 0$? What about $\displaystyle \sum_{q=1}^\infty \frac{f(q)}{\log^C(q)} = \infty$ for some $C \geq 1$?

share|cite|improve this question
up vote 3 down vote accepted

To my knowledge here is the best known sufficient condition:

Let $c$ be any positive constant and let $g: [0,\infty)\rightarrow [0,\infty)$ be defined by $g(0)=0$, $$g(x)=x \exp\left(-c (\log (- \log x))(\log \log (-\log x)) \right) \quad \text{ if } ~ 0 < x < 1,$$ and $g(x)=1$ if $x\ge 1$. If $$\sum_{q=1}^{\infty} g\left(\frac{f(q)}{q}\right) \varphi(q)=\infty$$ then $m(\mathbb{R}\setminus E_f)=0$. In particular this implies that the Duffin-Schaeffer Conjecture is true for any $f$ which satisfies $$\sum_{q=16}^{\infty} \frac{\varphi(q) f(q)}{q \exp(c(\log \log q)(\log \log \log q))}=\infty$$ for some $c>0$. Even more particularly, since $(\log \log q)(\log \log \log q)\ll \log q$ this means that if $$\sum_{q=1}^\infty\frac{f(q)}{q^\epsilon}$$ diverges for some $\epsilon >0$ then the Duffin-Schaffer Conjecture holds. However if you replace this sum with the second sum that you asked about (i.e. the one with the power of a logarithm in the denominator), then the answer is not known.

By the way the $1+\epsilon$ result that you mentioned is actually a corollary of a theorem that was originally proved by Glyn Harman (it follows from Theorem 3.7(iii) in his book Metric Number Theory). However the result in the Haynes, Pollington, Velani paper is actually a little stronger.

share|cite|improve this answer

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.