trureturingGitHub MATHEMATICAL ATLAS

RESEARCH / Focused target

Kurkov's full lower-bit recurrence for A284005

Literature status: not recheckedRepository record: proved

RELEASED FOUNDATIONS / PROPOSED CONNECTION

Research connections

1 released anchors
All source anchors (1)

Problem

The question

This dossier anchors exactly the November 10, 2019 FORMULA in OEIS A284005, attributed to Mikhail Kurkov:

Conjecture: a(n) = 2a(f(n)) + Sum_{k=0..floor(log_2(n))-1} a(f(n) + 2^k(1 - T(n,k))) for n > 1 with a(0) = 1, a(1) = 2, f(n) = A053645(n), T(n,k) = floor(n/2^k) mod 2. - Mikhail Kurkov, Nov 10 2019

The NAME defines a(n)=(1+A000120(n))*a(floor(n/2)) for n>1. The initial data begins 1,2,4,6,8,12,18,24, including a(1)=2. Here wt(n)=A000120(n) is binary digit sum, L=floor(log_2(n)), and removing the leading binary one gives f(n)=n-2^L. All operations in the formal statement are natural-number operations. Every index 0 <= k < L occurs: an existing one bit contributes a(f(n)), and these repeated contributions are retained.

The exact native result is:

theorem result :
    a 0 = 1 ∧ a 1 = 2 ∧
    (∀ n : ℕ, 1 < n → a n = (1 + wt n) * a (n / 2)) ∧
    (∀ n : ℕ, 1 < n →
      a n = 2 * a (n - 2 ^ Nat.log 2 n) +
        ∑ k ∈ Finset.range (Nat.log 2 n),
          a (n - 2 ^ Nat.log 2 n + 2 ^ k * (1 - n / 2 ^ k % 2)))

This includes n=2 and every power of two. The lower-bit identity asserts nothing at zero or one. The April 24, 2023 two-parameter binomial conjecture and the A329369 modulo-two binomial transform are distinct questions. The divisor-count representation in the entry's NAME is outside this result.

Motivation

Our foothold

The defining recurrence follows successive binary prefixes. The question relates their product to setting each lower bit, while preserving the multiplicity of contributions from bits already equal to one. Its quantifier ranges over every natural index greater than one.

Gap

Missing bridges

The original-entry and history reading, attribution, repository issue/comment reading, and bounded resolution searches described in Library/Digit/kurkov2019a284005.md identified no exact prior proof in their inspected scope. The OEIS text endpoint reading on 2026-10-28 independently confirms the quoted line and its "Conjecture" label. Neither observation establishes global openness or priority.

The sequence definition and binary-weight lemmas already exist in the repository. The sealed result addresses the complete 2019 question; the imported sequence supplier's 2023 binomial theorem does not supply this identity. Preregistration and outstanding delivery obligations are associated with issue #23151; this dossier adds no mathematical claim.

Route

Proposed approach

The result reuses KurkovWeightProductBinomial.a and DyadicRowPolynomialRecurrence.wt, together with the retained binary-weight zero, even and odd lemmas. For arbitrary depth L and m<2^L, prefix products satisfy a(m)=P(L,1,m) and a(2^L+m)=2*P(L,2,m). A zero-bit change increases the relevant prefix weights by one. Induction on depth proves the product identity with the sum over zero positions. The one positions supply wt(m)*P(L,1,m), restoring the complete sum. Exact logarithmic bounds then identify m=n-2^L and preserve the source's literal n-based bit.

The proof is native repository mathematics. The source supplies the question; no external proof is imported. The symbolic unbounded proof has utility: none. The canonical result Scribe carries the typed OpenProblemResolutionClaim for this dossier.

Falsifier

What would falsify this route

An n>1 where the stated whole sum differs from a(n) refutes the identity. A discrepancy in either initial value, the defining recurrence, the interpretation of the leading-bit deletion, or any omitted existing-one summand invalidates the claimed correspondence with the source question.

Evidence

Evidence to collect

The retained Lean declaration is D5/S1/Digit/KurkovLowerBitRecurrence.result; its canonical exposition is Blueprint/D5/S1/Digit/KurkovLowerBitRecurrence.scribe.cs. The result's registration is retained in Reg/D5/S1/Digit/KurkovLowerBitRecurrence.lean, alongside the three helper registrations in the mirrored DyadicRowPolynomialRecurrence module. Current compilation and canonical validation determine their evidence status; this dossier is not a build receipt or proof authority.

Source quotations and the adapted statement are attributed to Mikhail Kurkov, entry author Antti Karttunen, and OEIS contributors under CC-BY-SA-4.0. The entry history is https://oeis.org/history?seq=A284005.

Triage

Scope assessment

theorem; exactly the full 2019 lower-bit identity, with the source's initial values, recurrence and domain. Resolution membership is derived from the typed Scribe claim and current frozen declaration.

ASSUMED-UNVERIFIED

Unverified assumptions

Global historical openness and priority remain unverified. OpenAlex rate limiting, unavailable StackExchange advanced-search responses, and unavailable GitHub code-search authorization limit the inherited literature screen. This batch performs no new broad literature search. OEIS's label alone does not establish worldwide openness. Independent review, required CI and publication remain separate delivery obligations.

Dossier & anchor history

Loading release observations...