49

Does anyone know if this number is algebraic or transcendental, and why?

$$\sum\limits_{n = 1}^\infty {10}^{ - n(n + 1)/2} = 0.1010010001000010000010000001 \ldots $$

MJD
  • 62,206
  • 36
  • 276
  • 489
Raffaele
  • 25,898
  • 2
  • 18
  • 34
  • 8
    Well, it's *almost* a [Liouville number](https://en.wikipedia.org/wiki/Liouville_number), but the obvious good approximations are not that good ... – Hagen von Eitzen May 02 '14 at 11:55
  • 1
    @AlfredYerger Consider the first three terms of the sum: $0.1 + 0.001 + 0.000001$ The big chains of zeroes for smaller terms are "hidden" by the larger terms to the left :) – Zubin Mukerjee May 05 '14 at 00:35
  • 1
    IIRC this is a challenge problem in *Tools of the Trade* by Paul Sally. I remember finding some reference a few years ago (after Sally assigned it to me among a list of problems) to this being open. – Alex Becker May 05 '14 at 05:49
  • I can't answer the question but do remember that this or similar questions were answered in the book "Aufgaben und Lehrsaetze aus der Analysis" by Polya and Szego. I no longer have this book but it is a serious place to look for the solution to this question. – berkeleychocolate May 05 '14 at 00:29
  • 2
    Yes it must be transcendental since it is the root of a geometric sequence which is not a polynomial. Or wait... is that a valid argument? – mathreadler Jun 21 '15 at 22:49
  • 2
    Wait, I was being stupid, it's not a geometric sequence even. – mathreadler Jun 21 '15 at 23:09
  • Transcendal it is. –  Oct 25 '20 at 08:30

1 Answers1

72

The number $0.1010010001000010000010000001\ldots$ is transcendental.

Consider following three Jacobi theta series defined by $$\begin{align} \theta_2(q) &= 2q^{1/4}\sum_{n\ge 0} q^{n(n+1)} = 2q^{1/4}\prod_{n=1}^\infty (1-q^{4n})(1 + q^{2n})\\ \theta_3(q) &= \sum_{n\in\mathbb{Z}} q^{n^2} = \prod_{n=1}^\infty (1-q^{2n})(1+ q^{2n-1})^2\\ \theta_4(q) &= \theta_3(-q) = \sum_{n\in\mathbb{Z}} (-1)^n q^{n^2} = \prod_{n=1}^\infty (1-q^{2n})(1- q^{2n-1})^2\\ \end{align} $$ and for any $m \in \mathbb{Z}_{+}$, $k \in \{ 2, 3, 4 \}$, use $\displaystyle D^m\theta_k(q)$ as a shorthand for $\displaystyle \left( q\frac{d}{dq} \right)^m \theta_k(q)$.

Based on Corollary 52 of a survey article Elliptic functions and Transcendence by M. Waldschmidt in 2006,

Let $i, j$ and $k \in \{ 2,3,4 \}$ with $i \ne j$. Let $q \in \mathbb{C}$ satisfy $0 < |q| < 1$. Then each of the two fields $$ \mathbb{Q}( q, \theta_i(q), \theta_j(q), D\theta_k(q)) \quad\text{ and }\quad \mathbb{Q}( q, \theta_k(q), D\theta_k(q), D^2\theta_k(q)) $$ has transcendence degree $\ge 3$ over $\mathbb{Q}$

We know for any non-zero algebraic $q$ with $|q| < 1$, the three $\theta_k(q)$, in particular $\theta_2(q)$ is transcendental. Since

$$\sum_{n=1}^\infty 10^{-n(n+1)/2} = \frac{\sqrt[8]{10}}{2} \theta_2\left(\frac{1}{\sqrt{10}}\right) - 1$$

and using the fact $\frac{1}{\sqrt{10}}$ and $\frac{\sqrt[8]{10}}{2}$ are both algebraic, we find the number at hand is transcendental.

achille hui
  • 116,756
  • 6
  • 165
  • 318