A uniform family of reduced-coordinate cube recursions

Matijasevick, Rodriguez and Salazar, Perfect squares in reciprocal-sum sequences and primes that are inert in quadratic fields, arXiv:2610.00765v1, Section 7, question 2, asks:

For the recursions of this paper the exponent m=2 is forced. Is there a natural family of rational recursions for which c a_n+b_n is always a cube, other than those conjugate to z -> z^3?

The reduced-fraction convention is inherited from Problem 1.1: a_n,b_n are coprime positive integers, starts are positive integers, c is positive rational, and indices are n>=2. “Natural” is undefined. The explicit minimal reading here is a uniform autonomous rational family over Q of fixed degree, parameterized by a nonnegative integer, with one formula and a single c independent of the parameter, initial value and index. An unstated stricter author intent is not resolved by this reading.

For every t in N, including zero, put

$$F_t(x)=\frac{x+t}{(x+t+1)^3-(x+t)},\qquad x_n=F_t^{\circ n}(x_0).$$

The proved endpoint is: for every positive rational x_0, the intrinsic reduced numerator a_n and denominator b_n are positive and coprime for every n>=0; for every n>=1,

$$a_n+b_n=\bigl(a_{n-1}+(t+1)b_{n-1}\bigr)^3.$$

Thus c=1 works uniformly and the inherited integer starts and n>=2 follow. Each map has degree exactly 3 and no M in PGL_2(C) has M F_t M^{-1}(z)=z^3 on the entire projective line. No assertion at n=0 of a cube sum, or classification of all cube recursions, is made.

The question asks whether its square phenomenon has a cubic analogue beyond a coordinate copy of the power map. The exact question, quantifiers and Tier 1 classification were preregistered in issue #18920. The tier is a recent closing question with a complete elementary constructive route, rather than a central longstanding conjecture.

The supplied ordinary qualification read the complete primary source and relevant scope passages of Jones--Manes arXiv:1101.4339v3 and Cahn--Jones--Spear arXiv:1512.03085v2. The former's displayed families have degree two; the latter studies rational values that are powers. Neither inspected scope supplies the whole assertion about reduced numerator plus denominator. In particular F_0(1)=1/7 has sum 8 while its value is not a rational cube. The primary source's reciprocal-sum argument forcing squares does not range over this cubic family.

No equivalent whole answer was identified in the supplied bounded comparison. The primary source's two companion preprints were unlocated. Global priority is UNDETERMINED. This is not an exhaustive literature finding. An attributable equivalent whole answer, if located, removes new-resolution eligibility and routes to reuse. The repository and pinned-Mathlib searches did not identify an existing whole theorem. Existing rational reduction, polynomial gcd, projectivization and PGL action APIs are reused directly.

For a,b positive coprime integers let u=a+tb and w=u+b. Substitution gives

$$F_t(a/b)=N/D,\quad N=ub^2,\quad D=w^3-ub^2=u^3+3u^2b+2ub^2+b^3.$$

Both entries are positive. Since gcd(u,b)=1, w is coprime to u and b, so N is coprime to w^3 and hence to D. Mathlib's canonical rational numerator and denominator lemmas identify N,D with the actual reduced pair. Their sum is w^3. Induction applies this exact reduction to every positive rational iterate without a finite-orbit assumption.

The complex polynomials x+t and (x+t+1)^3-(x+t) are coprime, with monic cubic denominator. The true map degree is the maximum of their reduced degrees, 1 and 3; RatFunc.intDegree, their difference, is not used. The homogeneous pair is

$$P_t(X,Y)=(X+tY)Y^2,\quad Q_t(X,Y)=(X+(t+1)Y)^3-(X+tY)Y^2.$$

It has no common zero at any nonzero pair (X,Y). The coordinate theorem identifies its projectivization with the full OnePoint(C) map, including infinity and poles. The rational-function evaluation theorem identifies the affine chart; the rational orbit theorem identifies the rational iterates with that same projective map. The PGL action is the actual Mathlib action transported through the canonical projective-line equivalence. Every PGL class has a GL representative, and the coordinate theorem proves the usual fractional-linear formulas at finite points, poles and infinity.

The zero fibre of F_t has exactly two points, -t and infinity. In contrast, if two distinct points have equal images under the cubic power map, a third point has the same image. For finite u,v the third root is -u-v; the factorization of u^3-v^3 proves its image and distinctness. Infinity has a singleton power-map fibre. Bijections preserve these fibre properties, so no element of the full PGL_2(C) conjugates F_t to the power map. This proof uses no analytic critical-point formalization or unproved normal form.

The admission basis is open-problem-resolution under #18920. The result is the single source settlement; the remaining public theorems are consumed reduction and projective bridges. Their proof shape is conservatively bind-only: existing arithmetic and projective APIs, algebraic normalization, logical combination and induction supply the chain. No escape-witness claim is used to admit an unconsumed wrapper. Utility is none: the proofs are unbounded symbolic results, not a finite enumeration, checker, numeric reduction or certified finite instance.

A nonnegative t and positive rational start with a nonpositive iterate or an incorrect intrinsic reduced-pair identity at any n>=1 would refute the endpoint. A degree below 3 after true polynomial reduction, a common homogeneous zero, a failure of chart correspondence, or an actual complex PGL conjugacy to the power map would also refute it. Testing finitely many orbits cannot establish these universal claims.

The native source endpoint is D5/S3/Arith/CubeRecursionFamily.result with claim definition originalClaim. It includes intrinsic positive reduced pairs, all positive rational starts, every n>=1, degree 3, full rational-map and homogeneous-coordinate semantics, every PGL representative, the fractional-linear action and nonconjugacy. The projective bridges live in D5/S3/Arith/CubeRecursionProjective. No numerical experiment is a premise. Kernel verification, axiom closure, audit status and admission are distinct obligations; mathematical compilation does not certify priority or completion of independent reviews, freezing or required CI.

The information-escape audit is unfinished. All eight same-author compiled registration attempts remain declared_unresolved, with certificate: null and IE-C050 ClosedTruthReadout reason=unclassified_form. The actual published issue #18969 discloses the missing source bindings; publication is not validation.

| Original theorem | Missing binding | | --- | --- | | D5.S3.Arith.CubeRecursionFamily.reduced_step | source.selection_missing | | D5.S3.Arith.CubeRecursionFamily.positive_orbit_cube | source.selection_missing | | D5.S3.Arith.CubeRecursionFamily.result | source.residual_requires_open | | D5.S3.Arith.CubeRecursionProjective.map_degree_three | source.selection_missing | | D5.S3.Arith.CubeRecursionProjective.homogeneous_coordinates | source.selection_missing | | D5.S3.Arith.CubeRecursionProjective.rational_map_semantics | source.selection_missing | | D5.S3.Arith.CubeRecursionProjective.not_conjugate | source.selection_missing | | D5.S3.Arith.CubeRecursionProjective.mobius_coordinates | source.selection_missing |

The seven absent selections still require faithful compiler readout-occurrence selectors, source-coordinate selection, full-statement reconstruction, state/escape identity and current binding evidence. The aggregate result's residual/reconstruction binding remains unresolved. An explicitly unknown continuation supplies neither those bindings nor a validation certificate. The mathematical proof terms and the author's Reg attempts are retained. No native acceptance or solve credit is claimed.

theorem; mathematical resolution proved under the stated interpretation. Independent source/judgement review and canonical repository admission remain separate from the construction and proof.

What the settlement shows

• Proved, D5/S3/Arith/CubeRecursionFamily: the no-cancellation mechanism gives a cube already at the first step from every positive rational start, strengthening the inherited positive-integer/n>=2 domain.
• Proved, D5/S3/Arith/CubeRecursionProjective: the two-point fibre is incompatible with the cubic power map under every projective change of coordinates. Affine nonconjugacy alone is not the conclusion.
• Open: any stricter intended meaning of “natural,” classification of all such recursions, and the other closing questions. The source's square theorems concern a different class and remain unchanged. No conclusion depending on a hypothetical negative answer to question 2 is used.

The ordinary qualification supplies bounded source and literature readings; worldwide originality and the unlocated companion works are unverified. No claim of global priority, accepted solve credit, full repository checks, completed four-slot source binding or frozen status follows from the prose.