In an article it is used the fact that $$q_n\left|q_n\alpha-p_n\right|(a_{n+1}+1)>1$$ where $\alpha=[a_0;a_1,\ldots]$ is an irrational number and $q_i$ is the series of the best approximation denominator of $\alpha$, ie it is the sequence of the denominators of the sequence of the convergents of $\alpha$. I am not able to prove it, any idea?

  • $\begingroup$ Potentially related inequality: $q_{n+1}=a_{n+1}q_n+q_{n-1}< q_n(a_{n+1}+1)$ $\endgroup$ – Thomas Andrews Aug 16 '14 at 13:39
  • $\begingroup$ I am not able to obtain the thesis using this :( $\endgroup$ – Nisba Aug 16 '14 at 13:48
  • 2
    $\begingroup$ It may can be useful the fact (wikipedia) $|q_n\alpha-p_n|>\frac{1}{q_{n+1}+q_n}$ $\endgroup$ – Nisba Aug 16 '14 at 14:21
  • $\begingroup$ Yeah, that's an essentially weaker inequality than the one in the problem, since $$\frac{1}{q_{n+1} +q_n}<\frac{1}{(a_{n+1}+1)q_n}$$ $\endgroup$ – Thomas Andrews Aug 16 '14 at 14:43
  • $\begingroup$ With the inequality you gave and using wikipedia's one, I have proved a weaker version of what I want to prove: instead of $(a_{n+1}+1)$ I have $(a_{n+1}+2)$ $\endgroup$ – Nisba Aug 16 '14 at 14:46


$$\alpha=\frac{1+\sqrt{5}}{2}, p_n=3,q_n=2, a_{n+1}=1$$

Another counterexample:

$$\alpha=\frac{1+\sqrt{3}}{2}, p_n=4, q_n=3, a_{n+1}=2$$

I can show it is true when $a_n\geq 2$.

Your inequality is equivalent to:


Now assume the opposite, so that:


From $$\frac{1}{q_{n+1}q_n}=\left|\frac{p_{n+1}}{q_{n+1}}-\frac{p_n}{q_n}\right|= \left|\frac{p_{n+1}}{q_{n+1}}-\alpha\right|+\left|\alpha-\frac{p_n}{q_n}\right| $$ we see:


But $q_{n+1}=a_{n+1}q_n+q_{n-1}$, so:

$$\left|\frac{p_{n+1}}{q_{n+1}}-\alpha\right|> \frac{q_n-q_{n-1}}{q_{n+1}q_n^2}$$

When we can show that $\frac{q_n-q_{n-1}}{q_n}\geq\frac{1}{2}$, we'd be done, because this would mean that:


But $\frac{q_n-q_{n-1}}{q_n}>\frac 12$ is true when $a_n\geq 1$, since:

$$\frac{q_n-q_{n-1}}{q_n} = \frac{(a_n-1)q_{n-1}+q_{n-2}}{a_nq_{n-1}+q_{n-2}}\geq\frac{a_{n}-1}{a_n}\geq \frac{1}{2}$$

  • $\begingroup$ Thank you very much! For the case $a_n=1$ I have proved (it is true in general) a weaker version: $q_n\left|q_n\alpha-p_n\right|(a_{n+1 + \frac{1}{a_n}}+1)>1$. The proof follows easily from $|q_n-\alpha p_n|>\frac{1}{q_n+q_{n+1}}$ and using the fact that $q_n>a_n q_{n-1}$ for $n>1$. $\endgroup$ – Nisba Aug 16 '14 at 16:20
  • $\begingroup$ I doubt you mean $a_{n+1+1/a_n}+1$ which only makes sense when $a_n=1$, but perhaps $a_{n+1}+\frac{1}{a_n}+1$? $\endgroup$ – Thomas Andrews Aug 16 '14 at 16:25
  • $\begingroup$ It's also easy to prove that $$q_n\left|q_n\alpha-p_n\right|(a_{n+1}+1)>\frac{1}{2}$$ $\endgroup$ – Thomas Andrews Aug 16 '14 at 16:26
  • $\begingroup$ Yes I meant exactly what you wrote! $\endgroup$ – Nisba Aug 18 '14 at 16:52

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

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