slug: oeis-a396844-scaled-quadratic-reversion-mod-sixteen bibkey: hanna2026a396844 doi: null url: https://oeis.org/A396844 triage: theorem motivation_gids:
- D5/S1/Recurrence/Parity/ScaledQuadraticReversionModSixteen
Refutation of the A396844 mod-sixteen conjecture
Problem
OEIS A396844, Paul D. Hanna, Jul 03 2026, gives the following NAME and
COMMENT, quoted verbatim from Library/Recurrence/hanna2026a396844.md:
G.f. A(x) satisfies A( xA(x) - 4x*A(x)^2 ) = x^2.
Conjecture: for n > 8, a(n) == 12 (mod 16) iff n is of the form 2*4^k+1 otherwise a(n) is divisible by 16.
The conjecture is refuted at n = 17. The orchestrator reports
a(17) = 113101684573952804, congruent to 4 modulo 16. Since 17 is not
of the form 2*4^k+1, the “otherwise divisible by 16” clause is false.
The supplied numeric-check locator is results/verify-a396844-mod16.out;
that file was not available at this worktree path during Stage B.
Motivation
This is a first-tier OEIS conjecture from 2026. KPI = open problems resolved. The resolution is a refutation together with a corrected classification for every natural index n >= 2.
Gap
The search seat reported no proof found after reading the OEIS entry and revision history on 2026-09-09 and searching the identifier on arXiv, MathOverflow and GitHub. This seat has no network and did not repeat those searches. This records the search scope, not a proof of absence.
Route
Write the normalized series as A = X + X^2*C. A contracting operator on C
has a fixed point constructed by stabilized coefficients.
generating_equation gives A(0) = 0, a(1) = 1 and
A.subst(X*A - 4*X*A^2) = X^2, exactly the OEIS equation.
generating_unique identifies every normalized integral solution with A.
mod_four_identity gives A = X modulo 4.
A coefficientwise substitution argument preserves annihilated differences.
It shows that four times the shifted alternating-dyadic series is the
mod-16 fixed point for C. The frozen module
D5/S1/Recurrence/Invariants/AlternatingDyadicReversionModFour supplies its
public inverseSeries, R = sum_j (-1)^j X^(2^j), and public
inverse_equation, R + R(X^2) = X. Thus A = X + 4*X*R modulo 16.
Coefficient induction gives mod_sixteen_classification: for n >= 2,
a(n) = 4 modulo 16 iff n = 2^k+1 with k even, a(n) = 12 modulo 16
iff n = 2^k+1 with k odd, and 16 | a(n) iff n is off this support.
hannaClaim is the closed Prop stating both clauses of the quoted OEIS
conjecture for n > 8. hanna_conjecture_false : not hannaClaim refutes it
symbolically at 17 = 2^4+1: exponent 4 is even, so a(17) = 4 modulo 16.
17 is NOT of the form 2*4^k+1; in particular, 2*4^2+1 = 33.
The conjecture’s divisibility clause therefore fails. Its residue-12 clause
is correct, since 2*4^k+1 = 2^(2*k+1)+1 is exactly the odd-exponent case.
The classification is the corrected statement, not a proof of the conjecture.
Target generality is I because the proof imports the frozen module above,
itself generality G, with statement_id
sha256:ebfceee0d5bda860ab48b7b166d70a8e3e1c5aeb4e969050423894b570ddd652.
Only hanna_conjecture_false carries ResolutionKind.Refuted in Scribe.
Falsifier
The counterexample index is 17: n > 8, n is outside 2*4^k+1, yet
16 does not divide a(n). The orchestrator’s exact integer check is
supporting evidence only; the kernel refutation uses the symbolic residue.
Evidence
- Lean module:
D5/S1/Recurrence/Parity/ScaledQuadraticReversionModSixteen.lean. - Main theorem:
hanna_conjecture_false : not hannaClaim(Lean¬ hannaClaim). - Companions:
generating_equation,generating_unique,mod_four_identity,mod_sixteen_classification(the corrected statement). - Claim definition:
hannaClaim; series and coefficients:generatingSeries,a. - Axioms: std3 (
propext,Classical.choice,Quot.sound), as reported by the implementation seat for all five public theorems.
Triage
theorem. The refutation and the corrected classification are kernel
theorems. The resolution kind is Refuted, attached to
hanna_conjecture_false, not to the classification or a companion.
ASSUMED-UNVERIFIED
The quotes and attribution were supplied by the orchestrator. The OEIS revision history was read by the search seat, not by this seat. Literature scope was the OEIS entry and revision history plus identifier searches on arXiv, MathOverflow and GitHub on 2026-09-09. This seat had no network; first-publication priority, source-to-Lean identification and completeness of the literature search are not kernel-checked facts. The exact numeric output at the supplied locator was not read by this seat.