Keyboard shortcuts

Press ← or → to navigate between chapters

Press ? to show this help

Press Esc to hide this help

Linear Exponents and the Parity of OEIS A397591

Abstract

The integer series of OEIS A397591 exists uniquely and has odd coefficients precisely at dyadic neighbors above degree three.

The source defines A over integer formal power series with constant coefficient zero. For every m>1, [X^(m-1)](1-A)^m/(1-mX)=0. The inverse of the constant-one denominator is expressed by invOfUnit, and a(n) means the coefficient [X^n]A.

Definition 1.1 (Construction over the integers).

Formalization. D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.generatingSeries (✓ std3).

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

Write F=1-A. At degree n>0, put e=n+1 and g=(1-eX) inverse. Let R=[X^n](F^e g) and T=[X^(n-1)](F^(e-1) F’ g+F^e g squared). Differentiation gives nR=eT, so N=R-T satisfies R=eN. Changing the nth coefficient of F changes N by exactly the same amount. Successive coefficient corrections stabilize, giving an integer series F with N=0 at every positive degree. The generating series is A=1-F.

Definition 1.2 (The exact source equation).

Formalization. D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.DefiningEquation (✓ std3).

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

The predicate specifies integer coefficients, the zero constant term, the linear exponent m, and all natural m>1. The power-series unit inverse implements the division in the source NAME.

Theorem 1.3 (Every defining row vanishes).

Proof. Machine-checked in Lean as D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.generating_equation (✓ std3). ∎

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

Stabilization gives N=0 at each positive index. The identity R=eN and the equality of the geometric series with the denominator inverse give the original defining equation.

Theorem 1.4 (Integer existence and uniqueness).

Proof. Machine-checked in Lean as D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.integer_exists_unique (✓ std3). ∎

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

The stabilized construction supplies an integer solution. If two solutions agree below degree n, their normalized residual difference equals their nth coefficient difference. Induction forces equality in every degree.

Definition 1.5 (Coefficient indexing).

Formalization. D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.a (✓ std3).

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

The source coefficient indexing is extended by a(0)=0.

Theorem 1.6 (Odd coefficients at dyadic neighbors).

Proof. Machine-checked in Lean as D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.hanna_conjecture (✓ std3). ∎

Source. Repository-derived.

Acknowledgement. Paul D. Hanna (2026). OEIS A397591: linear-exponent coefficient equations and dyadic neighbors. URL: https://oeis.org/A397591.

Commentary.

Over ZMod 2, let U be the Catalan unit, so U=1+X U squared. Put E=1+X, O=EU, B=E squared+X O squared, and V=1+XU. Then B=E squared U=EO and UV=1. For each m>0, [X^m](B^m V)=0: an odd m selects an odd coefficient of a square, while an even m reduces to m/2 by the Catalan equation. This eliminates the even normalized rows of B; EO=B eliminates the odd rows. Uniqueness identifies B with the reduction of 1-A. For n>3 its coefficient is [X^n]U+[X^(n-2)]U. The frozen binary_catalan theorem says these summands are one when n+1 or n-1 is a power of two. They cannot both be one here, since two powers of two greater than two cannot differ by two. This proves the stated equivalence for every solution.

References

  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.DefiningEquation
  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.a
  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.generatingSeries
  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.generating_equation
  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.hanna_conjecture
  • Truth anchor: D5/S1/Recurrence/LinearRows/LinearExponentDyadicSupport.integer_exists_unique
  • Dependency: D5/S1/Recurrence/Invariants/CatalanCompositionSquareParity