slug: oeis-a363104-bilateral-quarter-shift-theta-mod-four bibkey: hanna2023a363104 doi: null url: https://oeis.org/A363104 triage: theorem motivation_gids:
- D5/S1/Recurrence/Bilateral/BilateralQuarterShiftThetaModFour
The A363104 bilateral generating function modulo four
Problem
OEIS A363104 (Paul D. Hanna, May 21 2023) gives the following NAME and
COMMENT, quoted verbatim from Library/Recurrence/hanna2023a363104.md:
Expansion of g.f. A(x) satisfying 4 = Sum_{n=-oo..+oo} (-x)^n * (4*A(x) + x^(n-1))^(n+1).
Conjecture: g.f. A(x) == theta_3(x) (mod 4); a(n) == 2 (mod 4) iff n is a nonzero square and a(n) == 0 (mod 4) iff n is nonsquare.
The NAME is formalised coefficientwise with exact finite windows after the
two Laurent terms cancel. mod_four_identity proves the generating-function
clause; hanna_conjecture proves both coefficient biconditionals. This dossier
has one resolution claim, attached only to hanna_conjecture.
Motivation
This is a first-tier OEIS conjecture from the 2023 entry. The KPI is open problems resolved; the target is the universal conjecture, not a finite prefix.
Gap
The search seat reported no proof found after reading the OEIS entry and its revision history on 2026-09-09 and searching the identifier on arXiv, MathOverflow, and GitHub. This Stage-B seat had no network access and did not independently repeat those searches. The literature conclusion is limited to that reported scope.
Route
Read the bilateral NAME formally using bilateralTerm A n for integer n.
At n = 0 it is 4*A; at n = -1 it is zero, because the Laurent parts
X⁻¹ and -X⁻¹ at these two indices cancel. At n >= 1 it is
(-X)^n * (4*A + X^(n-1))^(n+1). At n <= -2 it is
(-1)^n * X^(n²+n-1) * invOfUnit(1 + 4*A*X^(1-n))^(-(n+1)),
with the inverse taken at the unit constant coefficient one and all remaining
exponents nonnegative. The Lean definitions below reindex these cases by
natural m.
bilateralTerm_coeff_eq_zero proves every coefficient outside
Finset.Icc (-(N : ℤ) - 2) N vanishes. generating_equation gives constant
coefficient one and equality of the coefficient of four with the coefficient
of that exact finite sum at every degree N. As in the landed
BilateralThetaReversionModSixteen, the honesty boundary is explicit: this is
not a Mathlib bilateral infinite sum of formal series, and no such summation
operation or Laurent-series equality is asserted.
Pairing m+1 with -m-2 cancels the terms at A = 0. Each paired remainder
is divisible by four over the integers. Integral division by four gives a
degree-contracting iteration; stabilized coefficients construct the series,
and generating_unique proves uniqueness for the finite-window equation.
Modulo-sixteen linearisation and weighted telescoping use the unsigned frozen
ThetaSelfCompositionModFour.thetaSeries directly: multiplication by eight
removes the alternating signs modulo sixteen. cancellation_identity gives
∃ Q : ℤ⟦X⟧, 1 = A*θ₃ + 4*Q. Cancellation of four takes place over the
integers BEFORE reduction; cancellation inside ZMod 4 would be invalid.
Since θ̄₃ = 1 + 2*S̄ is self-inverse over ZMod 4, mod_four_identity
gives Ā = θ̄₃. Coefficient comparison proves hanna_conjecture: for
n >= 1, a n % 4 = 2 ↔ IsSquare n and
a n % 4 = 0 ↔ ¬ IsSquare n. a_zero gives a 0 = 1.
Target generality is I because the proof imports the frozen
D5/S1/Recurrence/Parity/ThetaSelfCompositionModFour, itself generality G.
Falsifier
A positive counterexample index at which either coefficient biconditional fails would falsify the COMMENT. The orchestrator’s exact arithmetic check is supporting evidence only; the formal theorem quantifies over every positive index.
Evidence
- Lean module:
D5/S1/Recurrence/Bilateral/BilateralQuarterShiftThetaModFour.lean. - Main theorem and sole resolution claim:
hanna_conjecture. - Companions:
bilateralTerm_coeff_eq_zero,generating_equation,generating_unique,a_zero,cancellation_identity,mod_four_identity. - Axioms: std3, namely
propext,Classical.choice,Quot.sound, reported for each of the seven public theorems by the implementation seat. - Frozen dependency statement_id:
sha256:0aec24647d5df63513f2c164c40043a38afc3ac9d0a739e594942fb5328c725b.
The exact Lean cases, including the private definitions that expand them, are:
private noncomputable def positiveTerm (A : PowerSeries R) (m : ℕ) : PowerSeries R :=
(-1) ^ (m + 1) * X ^ (m + 1) * (4 * A + X ^ m) ^ (m + 2)
private noncomputable def reciprocal (A : PowerSeries R) (m : ℕ) : PowerSeries R :=
invOfUnit (1 + 4 * A * X ^ (m + 3)) 1
private noncomputable def negativeTerm (A : PowerSeries R) (m : ℕ) : PowerSeries R :=
(-1) ^ (m + 2) * X ^ (m ^ 2 + 3 * m + 1) * reciprocal A m ^ (m + 1)
noncomputable def bilateralTerm (A : PowerSeries ℤ) (n : ℤ) : PowerSeries ℤ :=
if n = 0 then 4 * A else if n = -1 then 0 else
if 0 < n then positiveTerm A (n.toNat - 1) else negativeTerm A (-n - 2).toNat
The generating equation statement is quoted with its proof body omitted:
theorem generating_equation : constantCoeff generatingSeries = 1 ∧
∀ N, coeff N (4 : PowerSeries ℤ) =
coeff N (∑ n ∈ Finset.Icc (-(N : ℤ) - 2) N, bilateralTerm generatingSeries n) := by
These declarations express the finite-window interpretation of the NAME, not an independently formalised analytic or bilateral infinite-sum identity.
Triage
theorem. The formal proof closes both clauses of the COMMENT for the unique
integer series satisfying the stated coefficientwise interpretation of the NAME.
ASSUMED-UNVERIFIED
The OEIS quotations and entry attribution were supplied by the orchestrator. The OEIS entry and revision history were read by the search seat on 2026-09-09, not by this network-disabled Stage-B seat. The literature search scope was identifier searches on arXiv, MathOverflow, and GitHub; no exhaustive claim about other indexes or unpublished proofs is made. First-publication priority and identification of the OEIS NAME with the finite-window formal reading are not kernel-checked facts. The orchestrator’s exact check and external gate results were supplied evidence, not executions by this seat.