Absolute Reciprocal Cubes and Binomial Parity
Abstract
The coefficients of OEIS A380709 have Hanna’s conjectured binomial parity.
The generating equation and conjecture are recorded in hanna2025a380709. Write A for generatingSeries. The series A, B, F, and P(r) have integer coefficients, X is the indeterminate, and n and r are natural numbers. The auxiliary series Q and H used in the proof have coefficients in ZMod(2).
The operator mk forms a power series from its coefficient function, coeff(n,F) extracts its degree-n coefficient, and abs is integer absolute value. The operator invOfUnit(F,1) is Mathlib’s power-series inverse with unit parameter 1. The operator map applies a ring homomorphism coefficientwise; intCast(ZMod(2)) denotes Int.castRingHom(ZMod(2)). The operator choose is Nat.choose, with natural subtraction in its upper index. The operator ofNat denotes Int.ofNat and embeds a natural number into the integers, while zmodCast embeds it into ZMod(2). Remainders in lucas_recursion_q are natural remainders; those in hanna_conjecture are integer remainders.
Definition 1.1 (Coefficientwise absolute value).
Formalization. D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.absSeries (✓ std3).
Source. Repository-derived.
Commentary.
Absolute value is applied to each reciprocal coefficient before cubing.
Definition 1.2 (The stabilized coefficient sequence).
Formalization. D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.a (✓ std3).
Source. Repository-derived.
Commentary.
The auxiliary approximations P start at one. Inversion of series with constant coefficient one, coefficientwise absolute value, and cubing preserve coefficient agreement. Multiplication by X increases the degree of agreement, so coefficient n stabilizes by step n+1.
Definition 1.3 (The generating series).
Formalization. D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generatingSeries (✓ std3).
Source. Repository-derived.
Commentary.
The series A has the stabilized integer coefficients a(n).
Theorem 1.4 (The exact functional equation).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generating_equation (✓ std3). ∎
Source. Repository-derived.
Commentary.
Stabilization proves the functional equation with constant coefficient one. The inverse is therefore the reciprocal of A.
Theorem 1.5 (Uniqueness of the normalized solution).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generating_unique (✓ std3). ∎
Source. Repository-derived.
Commentary.
The two fixed-point equations extend agreement by one degree. Induction gives agreement in every degree and hence equality.
Theorem 1.6 (Three Lucas recursions).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.lucas_recursion_q (✓ std3). ∎
Source. Repository-derived.
Commentary.
Mathlib’s binary Lucas congruence gives the two odd-index formulas. For a positive even index, repeated halving reduces choose(4r,r) to a binomial coefficient with even upper and odd lower index, which vanishes modulo two.
Theorem 1.7 (Identification with the binomial parity series).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.mod_two_identity (✓ std3). ∎
Source. Repository-derived.
Commentary.
Absolute values disappear modulo two. If f is the reduction of A and U its reciprocal, the equation gives U=1+XU^4 and f=f^2+XU^2. The Lucas recursions and Frobenius give Q=1+XQ^4 when coefficient n of Q is choose(4n+1,n) modulo two. Factoring the difference of the quartic equations gives Q=U by cancellation of a series with unit constant coefficient. The two Lucas recursions for the displayed binomial series H give H=H^2+XQ^2. Thus (1-f-H)*(f-H)=0. The first factor has constant coefficient -1 and is a unit, proving f=H.
Theorem 1.8 (The constant coefficient).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.a_zero (✓ std3). ∎
Source. Repository-derived.
Commentary.
The exact equation gives a(0)=1. This represents the convention binomial(-1,0)=1 without using natural subtraction at index zero.
Theorem 1.9 (Hanna’s A380709 conjecture).
Proof. Machine-checked in Lean as D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.hanna_conjecture (✓ std3). ∎
Resolves. Problems/oeis-a380709-absolute-reciprocal-cube-parity (proved) by D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.hanna_conjecture.
Citation. Paul D. Hanna (2025). OEIS A380709, g.f. satisfying A(x) = 1 + xabs(1/A(x))^3*. URL: https://oeis.org/A380709.
Commentary.
At each positive index, coefficient extraction from the reduction identity gives equality in ZMod(2). Mathlib identifies this equality with equality of the two integer remainders modulo two.
References
- Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.a - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.a_zero - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.absSeries - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generatingSeries - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generating_equation - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.generating_unique - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.hanna_conjecture - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.lucas_recursion_q - Truth anchor:
D5/S1/Recurrence/Parity/AbsoluteReciprocalCubeParity.mod_two_identity