Erdős Problem #267 — Let F1=F2=1F_1=F_2=1 and Fn+1=Fn+Fn−1F_{n+1}=F_n+F_{n-1} be the Fibonacci sequence. Let n1<n2<⋯n_1<n_2<\cdots be an infinite sequence with nk+1/nk≥c>1n_{k+1}/n_k \geq c>1. Must ∑k1Fnk\sum_k\frac{1}{F_{n_k}} be irrational?

About 46 years old · traced to

Let F1=F2=1F_1=F_2=1 and Fn+1=Fn+Fn−1F_{n+1}=F_n+F_{n-1} be the Fibonacci sequence. Let n1<n2<⋯n_1<n_2<\cdots be an infinite sequence with nk+1/nk≥c>1n_{k+1}/n_k \geq c>1. Must ∑k1Fnk\sum_k\frac{1}{F_{n_k}} be irrational?

References

Progress summary

Refreshed
Claimed progress

A formal proof is claimed but not publicly verifiable; the problem remains open in the range below the known threshold.

Erdős and Graham posed the question in 19801980: whether every sufficiently lacunary reciprocal Fibonacci sum is irrational.

Known results

  • Badea proved irrationality for c≥2c\geq 2 in 19931993.
  • Good (19741974) and Bicknell–Hoggatt (19761976) found ∑k≥01/F2k=(7−5)/2\sum_{k\geq0}1/F_{2^k}=(7-\sqrt{5})/2.
  • André-Jeannin proved ∑1/Fn\sum 1/F_n irrational in 19891989.
  • Transcendence holds for arbitrary sequences when c>2c>2; this strengthens earlier bounds.

AlphaProof claim

A formalization comment says “Formal proof provided by AlphaProof,” but the theorem remains marked open and contains sorry; no completed proof or independent mathematical artifact is available.

Current status (as of March 2026): Irrationality is settled for c≥2c\geq2 and transcendence for c>2c>2, while the original range 1<c<21<c<2 remains open; the AlphaProof solution claim is unverified.

Sources

Solutions 0

No solutions have been posted yet.