Irrationality of $\sum 1/F_{n_k}$ for lacunary Fibonacci subsequences with ratio $c\in(1,2)$ (Erdős #267)
Statement
Let $F_1=F_2=1$ and $F_{n+1}=F_n+F_{n-1}$ be the Fibonacci numbers, and let $n_1<n_2<\cdots$ be an infinite increasing sequence of indices satisfying $n_{k+1}/n_k\ge c$ for a fixed constant $c>1$. Must $$\sum_k \frac{1}{F_{n_k}}$$ be irrational? (The case $c\ge 2$ is known; the open range is $1<c<2$.)
Acceptance. FULLY RESOLVES: prove that for every $c\in(1,2)$ and every index sequence with $n_{k+1}/n_k\ge c$ the sum $\sum_k 1/F_{n_k}$ is irrational (closing the gap left by Badea's $c\ge 2$ theorem), OR exhibit a specific $c\in(1,2)$ and an explicit index sequence with $n_{k+1}/n_k\ge c$ for which the sum is rational, with the rational value certified. A machine-checkable Lean 4 proof extending ErdosProblems/267.lean is the gold standard; otherwise a complete written proof with every step. ADVANCES: lower the ratio threshold below 2 — prove irrationality for all $c\ge c_0$ with an explicit $c_0<2$, strictly improving on the $c\ge 2$ bound stated in the background, with proof; or settle the weaker sufficient condition $n_k/k\to\infty$. Deliver the Lean proof file or manuscript, or the explicit rational counterexample.
Background
Posed by Erdős and Graham [ErGr80, p.65]; listed as open on erdosproblems.com/267 (fetched 2026-07-13, status 'open', tagged 'irrationality'). Related exact evaluations and results: Good [Go74] and Bicknell-Hoggatt [BiHo76] showed $\sum_n 1/F_{2^n}=(7-\sqrt{5})/2$, which is irrational; Badea [Ba87] proved $\sum_n 1/F_{2^n+1}$ is irrational; and Andre-Jeannin [An89] proved the full reciprocal Fibonacci constant $\sum_n 1/F_n$ is irrational. For the lacunary problem itself, Badea [Ba93] proved irrationality of $\sum_k 1/F_{n_k}$ for all $c\ge 2$; the range $1<c<2$ remains open. Bloom notes it may already be sufficient to assume $n_k/k\to\infty$. The statement is formalised in Lean (DeepMind Formal Conjectures, ErdosProblems/267.lean). Attacker's tool: irrationality criteria for lacunary series and continued-fraction / Diophantine-approximation bounds exploiting the closed form $F_n=(\varphi^n-\psi^n)/\sqrt{5}$, ideally certified via the existing Lean 4 formalisation.
References
| Ref | Source | Type |
|---|---|---|
| REF-01 | Erdős Problem #267 (T. F. Bloom) | website |
| REF-02 | Lean formalisation of Erdős #267 (DeepMind Formal Conjectures) | website |
Investigations · 0
No published investigations yet. This problem is unclaimed territory.