7
$\begingroup$

There is a certain number, say $v$. I can prove it is irrational. That would be more interesting if it is expressible in terms of known values ... zeta functions, Catalan's number, L-functions, etc. The number is defined precisely below. To $60$ decimals it is \begin{align} v &= 2.4953225371\ 3460865878\ 0441528421\ 4735785377\ 0845907036\ 757056931 \\ \frac{1}{v} &= 0.4007497969\ 1735762467\ 103472853\ 2244486515\ 1108820032\ 4884686488 \end{align}

Question. What number is that?


A recent post by Tito Piezas III proposes values of continued fractions arising from certain three-term recurrences. It reminded me of this, which I did around 2014. Following Apéry's proofs of irrationality of $\zeta(2), \zeta(3)$, I worked out many similar examples. There were many continued fractions. In some cases (a) conjectured values were clear; in some cases (b) the Liouville proof that the value is irrational worked. But cases (which would be most interesting) where both (a) and (b) hold were very few. This is one of the (b) cases.

Continued fraction \begin{align} v &= \frac{5}{2} +{\Large{K}}_{n=1}^\infty\frac{-n^2(n-\frac12)^2}{\;34n^2+17n+\frac52\;} \\ 2v &= 5 + {\Large{K}}_{n=1}^\infty\frac{-n^2(2n-1)^2}{\;68n^2+34n+5\;} = 5 - \frac{1}{\displaystyle 107 -\frac{36}{\displaystyle345 - \frac{225}{\displaystyle719 - \ddots}}} \end{align}

Recurrences
The "pre-Apery" sequence $(a_n)$ is defined by the recurrence $$ 0 = (n+1)^2 a_{n+1}-\left(34n\left(n+\frac12\right) +\frac52\right)a_n+\left(n-\frac12\right)^2a_{n-1} \tag1$$ with initial values $a_0 = 1, a_1 = \frac52$. I called it that because of the generating function equation $$ \left(\sum_{n=0}^\infty a_n x^n\right)^2 = \sum_{n=0}^\infty A_n x^n $$ where $(A_n)$ is the usual Apéry sequence. [Note $(4^n a_n) =(1,10,534,40900,\cdots)$ is an integer sequence not listed in OEIS.]
A "complementary sequence" $(b_n)$ satisfies the same recurrence $(1)$ with initial values $b_0 = 0, b_1 = 1$. \begin{align} (a_n) &= \left(1, \frac52, \frac{267}{8}, \frac{10225}{16}, \frac{1836275}{128}, \frac{90191115}{256}, \frac{9386644575}{1024}, \frac{508540579305}{2048},\cdots\right), \\ (b_n) &= \left(0, 1, \frac{107}{8}, \frac{12293}{48}, \frac{4415321}{768}, \frac{1807203551}{12800}, \frac{22570175443}{6144}, \frac{1497911874023249}{15052800},\cdots\right) . \end{align} Our value is: $$ v = \lim_{n\to\infty}\frac{a_n}{b_n} . $$

Rapidly converging series
$$ \frac{1}{v} = \sum_{k=0}^\infty \frac{(2k)!^2}{2^{4k} k!^2 (k+1)!^2 a_k a_{k+1}} $$

$\endgroup$
5
  • 1
    $\begingroup$ Wonder why there are brackets around the left-hand side of the generating function equation - did you mean to square it, or something along those lines? $\endgroup$ Commented May 24, 2023 at 15:05
  • $\begingroup$ @MaxMuller a square was missing. Fixed. $\endgroup$ Commented May 24, 2023 at 15:08
  • 1
    $\begingroup$ As usual, I tried the Inverse Symbolic Calculator. No luck. So I used the Integer Relations algorithm of Mathematica and tried to find a rational relation between 8 vectors: your number, $\zeta(2), \zeta(3), \pi^3$, log(2), log(3), Catalan, and Gieseking's constant. No luck either. :( $\endgroup$ Commented May 24, 2023 at 18:06
  • 1
    $\begingroup$ Yes but what about some of those numbers with square-root of integers for some coefficients? For example, one of the ones in your post has value $$\frac{2}{2\sqrt3}\kappa$$ $\endgroup$ Commented May 25, 2023 at 0:27
  • $\begingroup$ @GeraldEdgar Yes, admittedly that is a weakness of most integer relations algorithms. I have a continued fraction of my own in another post that I thought would be easy, but is resisting a closed-form even after H. Cohen provided 100-digits. But do inform me if you finally nail this real number. $\endgroup$ Commented May 27, 2023 at 17:03

0

You must log in to answer this question.