0
$\begingroup$

Hurwitz's Theorem in Number Theory states that for every irrational number $\xi$, there are infinitely many relatively prime natural numbers $(p,q)$ satisfying the equation: $$ | \xi−\frac{p}{q}| < \frac{1}{\sqrt{5}q^2} $$

I'm interested in unilateral approximations to $\xi.$ Specifically, letting

$$ L = \left\{(p,q): \; p \text{ and } q \text{ are relatively prime positive integers such that } \frac{p}{q} < \xi \text{ and } \left|\xi - \frac{p}{q}\right| < \frac{1}{\sqrt{5}q^2} \right\} $$

and

$$ U = \left\{(p,q): \; p \text{ and } q \text{ are relatively prime positive integers such that } \frac{p}{q} > \xi \text{ and } \left|\xi - \frac{p}{q}\right| < \frac{1}{\sqrt{5}q^2} \right\}, $$

then for each irrational $\xi$ we can conclude that AT LEAST ONE of the sets $L$ and $U$ is infinite. Do we know whether, for each irrational $\xi,$ BOTH of the sets are infinite? If not, then do we at least know whether, for each irrational $\xi,$ BOTH of the sets are not empty?

$\endgroup$
10
  • 1
    $\begingroup$ any $p>\xi$, and $q=1$? $\endgroup$
    – Albert
    Nov 26, 2021 at 10:08
  • 2
    $\begingroup$ The simple continued fraction gives approximations alternating too big and too small , hence there are in fact inifnite many fractions (and more important : arbitary good fractions both from above and from below) doing the job. $\endgroup$
    – Peter
    Nov 26, 2021 at 10:10
  • $\begingroup$ (I retracted my vote to close because I just understood what op meant; still I think the question should be reformulated to be clearer) $\endgroup$
    – Albert
    Nov 26, 2021 at 10:19
  • 1
    $\begingroup$ It might be that if all the even partial quotients (in the continued fraction for $\xi$) are small, while all the odd partial quotients are big, then all the Hurwitz-good approximations will be on the same side of $\xi$. $\endgroup$ Nov 26, 2021 at 12:20
  • 1
    $\begingroup$ @DaveL.Renfro thanks for reformulating. That's indeed what I meant. $\endgroup$
    – dd689
    Nov 27, 2021 at 11:09

1 Answer 1

0
$\begingroup$

Theorem 1 of the Eggan & Niven paper mentioned in the comments says,

For any $c>1$ there are uncountably many irrationals $\xi$ for which $$0<{a\over b}-\xi<{1\over cb^2}$$ has no rational solutions $a/b$. Analogously, for any $c>1$ there are uncountably many irrationals $\xi$ for which $0<\xi-(a/b)<1/(cb^2)$ has no rational solutions.

$\endgroup$
1
  • 1
    $\begingroup$ Thanks for the clear answer! $\endgroup$
    – dd689
    Dec 1, 2021 at 11:02

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .