Wiki
Wiki

Research notes on every problem, and a library of the papers behind them. Built from the open erdos repository.

Updated


Claim. Write F(t)=∑n≥1(tn−1)−1F(t)=\sum_{n\ge1}(t^n-1)^{-1}. For coprime integers a>b≥1a>b\ge1 with

log⁡blog⁡a<12−1π2=0.39868…,\frac{\log b}{\log a}<\frac12-\frac1{\pi^2}=0.39868\ldots,

F(a/b)F(a/b) is irrational, with an irrationality measure. This is the case K=QK=\mathbb Q, v=∞v=\infty, α=−1\alpha=-1 of Theorem 2 of P. Bundschuh and K. Väänänen, Arithmetical investigations of a certain infinite product, Compositio Math. 91 (1994), no. 2, 175--199 (p. 177). The paper sets Eq(z)=∏j≥1(1+zq−j)E_q(z)=\prod_{j\ge1}(1+zq^{-j}) and, by logarithmic differentiation, Lq(z)=Eq′(z)/Eq(z)=∑j≥1(qj+z)−1L_q(z)=E_q'(z)/E_q(z)=\sum_{j\ge1}(q^j+z)^{-1}, so Lq(−1)=F(q)L_q(-1)=F(q), and the irrationality of Lq(α)L_q(\alpha) is equivalent to the linear independence of Eq(α)E_q(\alpha) and Eq′(α)E_q'(\alpha) over Q\mathbb Q (p. 177). Theorem 2 bounds ∣a0Eq(α)+a1Eq′(α)∣|a_0E_q(\alpha)+a_1E_q'(\alpha)| from below by an explicit power of the height of (a0,a1)(a_0,a_1) when λ=log⁡h(q)/log⁡∣q∣\lambda=\log h(q)/\log|q| is small enough, and in the case α=−1\alpha=-1 it allows every λ<(1/2+1/π2)−1\lambda<(1/2+1/\pi^2)^{-1}. For q=a/bq=a/b in lowest terms h(q)=ah(q)=a, so λ=log⁡a/log⁡(a/b)\lambda=\log a/\log(a/b), and the condition on λ\lambda is the condition on log⁡b/log⁡a\log b/\log a above. The proof (Section 3) works from linear forms J(n)=P0Eq(α)−αP1Eq′(α)J(n)=P_0E_q(\alpha)-\alpha P_1E_q'(\alpha) with polynomial coefficients, whose size it controls under the same condition, and the paper reads its bounds as irrationality measures for Lq(α)L_q(\alpha) (p. 178). The year of the volume is the only date the record gives, so this page is dated to its first day.

Covers. Every t=a/b>1t=a/b>1 in lowest terms with log⁡b/log⁡a<12−1π2\log b/\log a<\frac12-\frac1{\pi^2} of Problem 1049: every integer t≥2t\ge2 (b=1b=1), already settled by Erdős, and infinitely many non-integer rationals, for example 7/27/2, since log⁡2/log⁡7=0.356…\log2/\log7=0.356\ldots. It does not cover 3/23/2, where log⁡2/log⁡3=0.63…\log2/\log3=0.63\ldots.

Acceptance. Refereed: the paper is a journal publication in Compositio Mathematica, volume 91, no. 2 (1994), received 6 October 1992, the refereed evidence. The site's page does not cite it; the forum comment of 11 September 2026 (the discussion link) points to its Theorem 2 for this region. The site labels the problem OPEN, so no reviewed evidence is listed.

Depends on. Nothing in this wiki.