• Paper 111 (draft): Topological and Ergodic Approaches to the Collatz Conjecture — Rei-AIOS vs. Santana (2026)
  • Abstract
  • 1. Introduction
  • 2. Santana’s Approach — Summary
    • 2.1 The key topology 𝒯
    • 2.2 Thermodynamic formalism
    • 2.3 Main result (Santana Theorem B)
    • 2.4 Conditional status — explicit
  • 3. Rei-AIOS Ergodic Tooling — Summary
    • 3.1 Component engines (2026-04-17 snapshot)
    • 3.2 Honest gap
  • 4. Five-Axis Comparison
    • 4.1 Equivalence of conditional structure
  • 5. Empirical Complementarity — STEP 870 MANDALA Integration
  • 6. Hybrid Framework — Proposal
  • 7. Open Problems
  • 8. Conclusion
  • 9. Reproducibility
  • 10. Acknowledgements
  • References

Canonical DOI: https://doi.org/10.5281/zenodo.19637784
Author: Nobuki Fujimoto (ORCID 0009-0004-6019-9258)
License: CC-BY-4.0

§Paper 111 (draft): Topological and Ergodic Approaches to the Collatz Conjecture — Rei-AIOS vs. Santana (2026)

Authors: Nobuki Fujimoto (ORCID 0009-0004-6019-9258), Claude Code (verification) Date: 2026-04-18 Status: DRAFT — not yet peer-reviewed. Comparative analysis paper. License: CC-BY-4.0

§Abstract

Santana (arXiv:2601.03297v4, 2026) proposes a coarse topology on ℕ and applies thermodynamic formalism to establish that Collatz-family maps have at most finitely many periodic orbits, conditional on the existence of equilibrium states for every continuous potential. The Rei-AIOS project (Fujimoto 2026) independently developed five ergodic and measure-theoretic tools for the Collatz orbit — Birkhoff averaging, Koopman spectral approximation, Perron-Frobenius transfer operators on ℤ/Mℤ, a Lean 4 formalization of Tao (2019) logarithmic density, and (as of 2026-04-18) a Santana-topology MANDALA lens. We present a five-axis structural comparison and conclude: (1) the two approaches are complementary, not competing; (2) both exhibit the same conditional structure — Santana’s equilibrium-state axiom is isomorphic to Rei’s honest gap between log-density 1 and pointwise convergence; (3) a hybrid framework combining Santana’s topology with Rei’s finite modular transfer operators is a natural next step.

§1. Introduction

The Collatz conjecture — every positive integer iterates to 1 under T(n) = n/2 if even, 3n+1 if odd — has attracted diverse approaches over 80 years. Classical work (Conway 1972, Eliahou 1993, Lagarias 2010) established conditional bounds on cycle length and trajectory growth. Tao’s landmark 2019 result (arXiv:1909.03562) shifted the field by proving that almost all orbits (in logarithmic density) attain almost bounded values, though the pointwise conjecture remained out of reach.

Two recent lines of work are compared in this paper:

• Santana (2026) introduces a novel coarse topology 𝒯 on ℕ and applies thermodynamic formalism — pressure functional, equilibrium states, continuous potentials — to deduce finiteness of cycles for a class of Collatz-like maps, conditional on the existence of equilibrium states.

• Rei-AIOS (Fujimoto 2026) is a TypeScript + Lean 4 research system that incrementally developed ergodic tooling for Collatz: Birkhoff time averages, Koopman operator approximations, Perron-Frobenius transfer operators on ℤ/Mℤ, a Lean 4 formalization of Tao (2019), and as of STEP 870, a direct implementation of Santana’s topology as a MANDALA observation lens (E31).

Our thesis: these approaches are complementary. Neither subsumes the other, and a hybrid is natural.

§2. Santana’s Approach — Summary

§2.1 The key topology 𝒯

The coarsest topology on ℕ containing the family {{n, 2n} : n ∈ ℕ} (Santana Lemma 4-5). In this topology:

• Every even singleton {n} is open (since n ∈ {n/2, n} ∩ {n, 2n}).
• Every odd n has {n, 2n} as its smallest open neighborhood.
• Connected components are 2-adic rays: for odd m, the cluster of m is {m, 2m, 4m, 8m, …}.

A set U is open ⟺ for every odd n ∈ U, 2n ∈ U.

§2.2 Thermodynamic formalism

Pressure functional: P(φ) = sup_{μ} [h_μ(f) + ∫φ dμ], where the supremum runs over f-invariant Borel probabilities μ, h_μ is the Kolmogorov-Sinai entropy, and φ is a 𝒯-continuous potential (Santana Eq. 1).

For this system, ergodic measures are supported on periodic orbits and have zero entropy, reducing P(φ) to sup_μ ∫φ dμ.

§2.3 Main result (Santana Theorem B)

For Collatz-family maps f : ℕ → ℕ of the form (∗) [Santana Sec. 3], there are at most finitely many periodic orbits.

Proof structure (Santana Theorem 18):

1. Assume infinitely many cycles 𝒪₁, 𝒪₂, ….
2. For each, define ergodic δ_i supported on 𝒪_i. Form μ = Σ a_i δ_i.
3. For any continuous integrable potential φ, ∫φ dμ = Σ a_i ∫φ dδ_i.
4. If ∫φ dδ_i → ∞, choose a_i so the sum diverges — non-integrability.
5. For finitely many cycles, integrability is automatic.
6. Contradiction ⟹ finitely many cycles.

§2.4 Conditional status — explicit

The proof requires that equilibrium states exist for every continuous potential (Santana Lemma 14). Santana does NOT prove this existence independently. The logical implication is:

{equilibrium states exist ∀ continuous φ} ⟹ {finitely many cycles}

So Santana’s “finiteness of cycles” is a conditional theorem. The equilibrium-state axiom is the remaining open link.

§3. Rei-AIOS Ergodic Tooling — Summary

§3.1 Component engines (2026-04-17 snapshot)

§3.2 Honest gap

Across these engines, Rei maintains an explicit honest gap between what is proven and what is conjectured:

• Tao (2019): log-density 1, NOT pointwise ∀n.
• Transfer operator: spectral gap confirms attractor, does NOT prove ∀n.
• Tier2 axiom (Paper 106): conditional complete proof modulo C1/C2/C3 residuals.
• STEP 866 C8 LTE: elementary (RCA₀), closes one component.

No claim of unconditional global convergence.

§4. Five-Axis Comparison

§4.1 Equivalence of conditional structure

Santana’s conditional: {equilibrium states exist} ⟹ finiteness. Rei’s honest gap: {log-density 1 extrapolates pointwise} ⟹ ∀n termination.

Both are statements of the form “if a regularity / completeness axiom holds on the system, then the conjecture is true.” They are isomorphic in this meta-structural sense, even though the specific axiom differs (measure-theoretic regularity vs. density-extrapolation).

This observation is itself a contribution: the field is not divided between “proofs that work” and “proofs that don’t,” but rather a landscape of different conditional axioms that all reduce the conjecture to the same remaining analytic step.

§5. Empirical Complementarity — STEP 870 MANDALA Integration

STEP 870 (2026-04-18) implements the Santana topology as MANDALA lens E31, enabling direct comparison on the 25 atomic cores:

The two classifications agree on 23 of 25 atomic cores. The two exceptions (n=91 with K/bl² = 1.88, n=121 with K/bl² = 1.94) lie on the K/bl² threshold boundary and have potentials just under 10⁷. This is evidence that the Santana lens and the K/bl² threshold are almost-but-not-quite equivalent identifiers — an independent derivation of a close-to-identical invariant.

§6. Hybrid Framework — Proposal

We propose a hybrid incorporating the best of each approach:

Santana 𝒯 topology
    │
    ├── [generates] Clusters = 2-adic rays
    │       │
    │       └── [quotient induces] Rei Syracuse shortcut map
    │
    └── [defines] Pressure functional P(φ)
            │
            └── [numerical approximation via] Rei transfer operator
                    on ℤ/Mℤ (STEP 786), Perron-Frobenius spectrum

Concretely:

1. Santana’s 𝒯 quotient ≡ Syracuse shortcut structure in Rei. This is not coincidence — both isolate the odd-part dynamics from the doubling action.
2. Pressure functional finite-truncation: approximate Santana’s sup over invariant measures by restriction to the finite ℤ/Mℤ Markov chain (M = 192, 49152, etc., per STEP 694-696). This gives a computable proxy for the otherwise abstract equilibrium state existence.
3. Lean 4 formalization of Santana’s core lemmas using Rei’s existing padicValNat-based toolkit. Particularly, Lemma 14 (finiteness ⟺ equilibrium state) is a purely mathematical equivalence that should be formalizable.

§7. Open Problems

• OP-1: Does the Santana pressure functional P(φ) on the finite restriction ℤ/Mℤ converge as M → ∞ to the true infinite-ℕ pressure?
• OP-2: Is Santana’s equilibrium-state existence axiom provable in ZFC, or does it require stronger set-theoretic assumptions?
• OP-3: Does the Santana lens classification of Rei’s 25 atomic cores (23 INFINITY + 2 TRUE) extrapolate to n > 1000? I.e., do large-n atomic cores remain bimodal in santanaPotential?
• OP-4: Can Rei’s STEP 866 LTE identity be re-expressed in Santana’s 𝒯 topology as a morphism of certain coverings?

§8. Conclusion

Santana’s 2026 approach and Rei-AIOS’s ergodic tooling are complementary tracks, unified by a shared conditional structure. Their integration (Rei STEP 870 Santana lens, MANDALA v14) produces empirical agreement on 92% (23/25) of Rei’s atomic cores. A hybrid framework combining Santana’s topology with Rei’s finite modular transfer operators is a concrete research direction.

§9. Reproducibility

All numerical claims above are verified via the following Rei-AIOS test commands (Node.js + TypeScript via npx tsx):

§10. Acknowledgements

Santana (2026) for the original topological formulation. Tao (2019) for logarithmic-density foundation. The Lean 4 / Mathlib community for the formalization infrastructure.

§References

• Santana, E. (2026). “On the Collatz Conjecture: Topological and Ergodic Approach.” arXiv:2601.03297v4.
• Tao, T. (2019). “Almost all orbits of the Collatz map attain almost bounded values.” arXiv:1909.03562 (publ. Forum of Mathematics, Pi, 2022).
• Lagarias, J. C. (2010). “The 3x+1 problem and its generalizations.” AMS.
• Fujimoto, N. (2026). Paper 66: “Collatz 8-Component Structural Decomposition.” DOI: 10.5281/zenodo.19547521.
• Fujimoto, N. (2026). Paper 108: “Ricci-Flow 3-Category Classification.” DOI: 10.5281/zenodo.19616640.

Paper 111 is a draft. Feedback to fc2webb@gmail.com.