Avyukth Nilajagi discovers Fibonacci sequence hidden in harmonic series

In a fascinating LessWrong post titled 'Fibonacci Structure in Harmonic Series Partitions', independent researcher Avyukth Nilajagi presents a striking mathematical observation: when you greedily group consecutive terms of the harmonic series (1 + 1/2 + 1/3 + …) so that each group's sum just exceeds a fixed threshold τ, the number of terms in each group appears to follow the Fibonacci sequence exactly. Nilajagi discovered that setting τ = ln(φ), where φ is the golden ratio (~1.618), yields group sizes of 1, 1, 2, 3, 5, 8, 13, 21, … for the first 25 groups. The asymptotic growth rate of the group sizes is provably φ, but what is startling is the exactness of the Fibonacci numbers themselves — a phenomenon that remains unproven and invites further research.

The derivation uses the approximation H(n) ≈ ln(n) + γ (Euler-Mascheroni constant) and the Binet formula for Fibonacci numbers. By setting the sum of a group of k terms starting at n approximately equal to ln(1 + k/n) and requiring this to equal the threshold, the recurrence k_{m+1}/k_m → φ emerges when threshold = ln(φ). Empirically, the exact match holds for at least 25 groups. Stanislav Krym offered a comment providing a proof sketch showing that the Fibonacci property is preserved under the greedy algorithm when the threshold is ln(φ), leveraging the fact that φ^{k} ~ F_{k+1} + F_{k-1}φ and that cumulative sums of reciprocals honor Fibonacci identities.

This result elegantly connects two foundational mathematical sequences — the harmonic series (divergent, slowly growing) and the Fibonacci sequence (exponential growth). Beyond pure aesthetic pleasure, it suggests deeper combinatorial structures that may inform partition theory, continued fractions, or even computational sequence generation. The open question of whether the exact correspondence holds for all n is a tantalizing challenge for number theorists and recreational mathematicians alike.