Section 1

Introduction

1.1 The Collatz Conjecture

Let T: ℕ → ℕ be defined by:

T(n) = n / 2          if n is even
T(n) = (3n + 1) / 2    if n is odd

The Collatz conjecture asserts that for every n ≥ 1, there exists k such that Tᵏ(n) = 1, with all trajectories eventually entering the cycle 4 → 2 → 1.

Despite decades of intensive study, the conjecture remains open. Partial results include:

  • Terras (1976, 1979): The set of integers with finite stopping time has natural density 1 [1]
  • Allouche (1979): 2-adic equidistribution results [2]
  • Wirsching (1998): Symbolic dynamics and ergodic approaches [3]
  • Krasikov & Lagarias (2003): At least n^0.84 integers below n eventually reach 1 [4]
  • Tao (2019): Almost all orbits (in logarithmic density) attain almost bounded values [5]

The conjecture is widely believed true based on overwhelming computational evidence up to 2⁶⁸ [Oliveira e Silva, 2010] [6], but a rigorous proof remains elusive.

1.2 Multiplicative Drift Theory and the Random Walk Heuristic

A persistent heuristic in Collatz research is that the map behaves like a biased random walk in logarithmic space. For odd n, the next iterate is approximately (3n)/2^k(n), where k(n) = v₂(3n+1) is the exponent of the largest power of 2 dividing 3n+1.

Heuristic Observation 1.1 — Geometric Valuation Model

For large odd n, the distribution of k(n) = v₂(3n+1) is approximately geometric:

P( k(n) = r ) ≈ 2^(−r),     r = 1, 2, 3, …

giving expected value E[k(n)] = 2. This approximation is consistent with rigorous results by Terras [1976] and Lagarias [1985] and is verified computationally up to 2⁶⁸ [6].

Taking logarithms, each step produces a change:

ΔV = log( 3 + 1/n ) − k(n) log 2 ≈ log 3 − k(n) log 2

The expected drift is therefore:

E[ΔV] ≈ log 3 − 2 log 2 = log(3/4) ≈ −0.2877

This negative multiplicative drift suggests that trajectories tend downward on average, with occasional upward jumps when k(n) = 1. This multiplicative perspective — viewing Collatz as a stochastic process on the positive reals under multiplication — is central to understanding why logarithmic coordinates reveal the underlying structure.

1.3 The Critical Balance and Near-Criticality

A deeper insight emerges when we ignore the +1 term and consider the approximate multiplier M = 3 / 2ᵏ. Under the geometric model, the expected multiplier is:

E[M] = 3 · Σ(r=1..∞) 2^(−r) / 2ʳ = 3 · Σ(r=1..∞) 4^(−r) = 3 · (1/4) / (1 − 1/4) = 1

Critical insight: On the linear scale, the Collatz map is exactly critical — the expected multiplicative factor is 1, placing it at the boundary between contraction and expansion. This criticality explains why trajectories exhibit large fluctuations resembling a critical branching process, while the logarithmic scale reveals a small negative drift that eventually drives the system downward.

This tension between local randomness (critical branching) and global rigidity (negative log drift) is the key to understanding Collatz behavior. Locally, the map appears unpredictable, with large upward and downward excursions. Globally, the orbit is constrained by the arithmetic structure of 3n+1 and cannot sustain indefinite growth.

1.4 Heuristic Phase Diagram for Accelerated (an+b) Maps

The Collatz map is a special case of a broader family of accelerated maps:

T_{a,b}(n) = (an + b) / 2^v₂(an+b)

where v₂ denotes the 2-adic valuation. Under the heuristic geometric model for v₂(an+b), the expected logarithmic drift is approximately:

E[ΔV] ≈ log a − 2 log 2 = log( a/4 )

This yields a simple phase diagram:

Regime Condition Expected Drift E[Δ log n] Behavior
Contractive a < 4 < 0 Negative drift, eventual descent
Critical a = 4 = 0 Zero drift, maximal fluctuations
Expansive a > 4 > 0 Positive drift, likely divergence
Collatz (a = 3) a = 3 log(3/4) ≈ −0.288 Contractive, near-critical boundary
Remark 1.2 — Structural Interpretation via Architecture of Infinity

The Collatz map exhibits the toroidal structure characteristic of stable dynamical systems: an outward expansion channel (multiplication by 3), an inward compression channel (division by 2ᵏ), and a feedback loop (iteration). The critical parameter a = 4 marks the boundary where outward and inward flows balance exactly. For a < 4, compression dominates, forcing eventual descent. This geometric perspective — viewing Collatz as an incomplete system until both flows are modeled — aligns with the Architecture of Infinity framework [7], which identifies many classical problems as arising from analyzing only half of a circulatory structure.

1.5 Contributions of This Paper

This paper develops a probabilistic drift model that makes the heuristic precise under explicit statistical assumptions. The main contributions are:

  1. Drift structure (Section 2): The logarithmic potential V(n) = log n transforms multiplicative dynamics into additive drift, with ΔV = log 3 − k(n) log 2 + O(1/n). Rewriting as V_{m+1} = V_m + μ + ξ_m with μ = log(3/4) reveals Collatz as a linear stochastic system plus a small correction.
  2. Supermartingale formulation (Section 3): Under exponential decay of correlations in the valuation sequence {k(Tᵐ(n))}, we show that {V(Tᵐ(n))} forms a supermartingale with negative drift, implying almost sure boundedness and convergence to a bounded set.
  3. Critical threshold insight (Section 4): Embedding Collatz in the (an+b) family reveals a = 4 as the critical boundary, explaining why a = 3 produces large fluctuations despite eventual descent. The constant log(3/4) appears consistently in the drift calculation, critical branching interpretation, and phase diagram.
  4. 2-adic symbolic dynamics (Section 5): The valuation sequence k(Tᵐ(n)) records block lengths of zeros in the parity vector following each odd step. In the 2-adic setting, iteration resembles a shift map on binary sequences, explaining why geometric and mixing heuristics arise naturally.
  5. The trivial cycle as arithmetic attractor (Section 3.6): The cycle 4 → 2 → 1 is the unique stable orbit under the supermartingale dynamics. Its basin of attraction, under the mixing assumptions, is conjectured to include all .
  6. Explicit falsifiable conditions (Section 6): The required mixing properties are stated precisely and are empirically testable. They provide concrete targets for future analytic work, turning the Collatz conjecture into a structured research program.

Scope boundary: The result is not a proof of the Collatz conjecture, but a conditional probabilistic framework that reduces the conjecture to proving natural mixing properties of the valuation sequence along individual orbits.

Section 2

Logarithmic Potential and Drift Analysis

2.1 The Potential Function

Definition 2.1

Define the logarithmic potential V(n) = log n for n ∈ ℕ.

For odd n, write T(n) = (3n+1)/2^k(n) where k(n) = v₂(3n+1). Then:

ΔV(n) = V(T(n)) − V(n) = log(3n+1) − log( 2^k(n) · n )

Simplifying (Equation 2.1):

Equation 2.1
ΔV(n) = log( 3 + 1/n ) − k(n) log 2

For large n, log(3 + 1/n) = log 3 + O(1/n). Thus (Equation 2.2):

Equation 2.2
ΔV(n) = log 3 − k(n) log 2 + O(1/n)

2.2 Empirical Distribution of k(n)

The valuation k(n) has been studied extensively. Terras [1976] proved that the proportion of integers with k(n) = r approaches 2^(−r) in the limit. Lagarias [1985] showed that the 2-adic valuation of 3n+1 is asymptotically geometric [8].

Heuristic Observation 2.2 — Geometric Valuation Model

For large odd n, the distribution of k(n) = v₂(3n+1) is approximately geometric:

P( k(n) = r ) ≈ 2^(−r),     r = 1, 2, 3, …

This approximation is consistent with computations verifying Collatz convergence up to 2⁶⁸ [6]. The mean and variance are:

E[k(n)] ≈ Σ(r=1..∞) r · 2^(−r) = 2      Var(k(n)) ≈ 2

2.3 Expected Drift and Linear Form

Taking expectations in (2.2) using Heuristic Observation 2.2:

E[ΔV(n)] ≈ log 3 − 2 log 2 = log(3/4) ≈ −0.2876820724
Definition 2.3

Define μ = log(3/4) as the drift constant.

Then (2.2) can be rewritten (Equation 2.3):

Equation 2.3
ΔV(n) = μ + (2 − k(n)) log 2 + O(1/n)

For the iterated process X_m = V(Tᵐ(n)), this becomes (Equation 2.4):

Equation 2.4
X_{m+1} = X_m + μ + (2 − k_m) log 2 + O(e^(−X_m))

where k_m = k(Tᵐ(n)). This has the form of a linear stochastic difference equation:

X_{m+1} = X_m + μ + ξ_m + ε_m

with μ < 0, ξ_m = (2−k_m) log 2 a zero-mean fluctuation term (under Heuristic 2.2), and ε_m an exponentially small correction. This representation is central to the probabilistic analysis that follows.

Section 3

Supermartingale Formulation

3.1 The Stochastic Process

Consider the sequence X_m = V(Tᵐ(n)) for a fixed initial n. From (2.4):

Equation 3.1
X_{m+1} = X_m + μ + (2 − k_m) log 2 + O(e^(−X_m))

The error term decays exponentially with X_m, so for X_m sufficiently large it is negligible.

3.2 Mixing Assumptions

To treat {X_m} probabilistically, we require assumptions on the dependence structure of the valuation sequence along a fixed orbit.

Assumption A — Exponential Decay of Correlations

The sequence {k_m} exhibits exponential decay of correlations: there exist constants C < ∞ and ρ < 1 such that for all m, t ≥ 0:

| Cov(k_m, k_{m+t}) | ≤ C · ρ^t
Assumption B — Mean Stationarity

For every initial n, the Cesàro mean of the valuation converges to 2:

lim_{M→∞} (1/M) · Σ_{m=1}^{M} k_m = 2
Remark 3.1 — Status of the Assumptions

These assumptions are not proven, but are consistent with all known computational evidence up to 2⁶⁸ [6]. Terras [1976] and Allouche [1979] provide partial results showing equidistribution in residue classes and exponential decay of correlations in 2-adic models [1,2]. Tao's [2019] logarithmic density theorem also supports the heuristic that orbits behave quasi-randomly on large scales [5].

3.3 Why the Mixing Assumptions Are Plausible: Symbolic Dynamics and Entropy Heuristics

The plausibility of Assumptions A and B rests on a 2-adic symbolic representation of the Collatz map.

Construction 3.2 — Parity Encoding

For any n, define the infinite parity sequence:

ω(n) = (b₀, b₁, b₂, …) ∈ {0,1}^ℕ

where b_i = Tⁱ(n) mod 2. The sequence ω(n) encodes the trajectory via the rule: if b_i = 0 (even), apply T(n) = n/2; if b_i = 1 (odd), apply T(n) = (3n+1)/2. The valuation k_m then counts the number of consecutive zeros following the m-th occurrence of a 1 in ω(n).

Observation 3.3 — Shift Map Structure

In 2-adic coordinates, the Collatz map acts as a shift-like map on binary sequences. If n = 2ᵃ · m with m odd, then:

T(n) = 2ᵃ · (3m + 1) / 2^k(m)

This multiplication-and-shift structure resembles symbolic dynamics on {0,1}^ℕ under the left-shift σ.

Heuristic 3.4 — Entropy and Mixing

If ω(n) behaves like a typical sequence with respect to a shift-invariant measure on {0,1}^ℕ, then: (1) block frequencies stabilize (ergodicity); (2) correlations decay exponentially (mixing); (3) rare events occur with geometric frequency. Under this heuristic, Assumption A follows from mixing and Assumption B follows from ergodicity with mean E[k] = 2.

Evidence 3.5 — Computational Support

Extensive computational studies [6,9,10] show: the empirical distribution of k(n) converges to geometric 2^(−r) as trajectories lengthen; no long-range correlations are observed in residue class sequences; parity sequences exhibit statistical properties consistent with random binary strings. These observations, while not rigorous, strongly suggest that Assumptions A and B capture the true behavior of Collatz orbits.

3.4 Filtration and Conditional Expectation

Definition 3.6

Define the natural filtration F_m = σ(k₀, k₁, …, k_{m−1}), the σ-algebra generated by the first m valuations along the orbit.

Under Assumptions A and B, for large m:

E[ k_m | F_m ] ≈ E[ k_m ] = 2 + o(1)

due to decay of correlations. Thus:

E[ ξ_m | F_m ] = E[ (2 − k_m) log 2 | F_m ] ≈ 0

3.5 Supermartingale Property

Theorem 3.7 — Conditional Supermartingale

Under Assumptions A and B, there exists M₀ such that for all m ≥ M₀ and X_m ≥ X₀ (a sufficiently large threshold):

E[ X_{m+1} | F_m ] ≤ X_m + μ + o(1)

where μ = log(3/4) < 0.

Proof

From (3.1): X_{m+1} = X_m + μ + (2 − k_m) log 2 + O(e^(−X_m))

Taking conditional expectation:

E[X_{m+1} | F_m] = X_m + μ + E[(2 − k_m) log 2 | F_m] + O(e^(−X_m))

By Assumption A (decay of correlations), for large m:

E[(2 − k_m) | F_m] = 2 − E[k_m | F_m] ≈ 2 − 2 = o(1)

Thus: E[X_{m+1} | F_m] ≤ X_m + μ + o(1)

For m ≥ M₀ chosen so that o(1) + O(e^(−X_m)) < −μ/2, we have:

E[X_{m+1} | F_m] ≤ X_m + μ/2 < X_m

establishing the supermartingale property.

Corollary 3.8 — Almost Sure Boundedness

Under Assumptions A and B, for almost every orbit satisfying these assumptions, {X_m} is almost surely bounded.

Proof

By the supermartingale convergence theorem, if E[X_{m+1} | F_m] ≤ X_m − c for some c > 0 and all m ≥ M₀, then {X_m} converges almost surely. Since μ < 0, the negative drift prevents unbounded growth.

3.6 Convergence to the Trivial Cycle

Theorem 3.9 — Conditional Convergence

Under Assumptions A and B, every orbit satisfying these assumptions must eventually enter the cycle 4 → 2 → 1.

Proof Sketch

The supermartingale property (Theorem 3.7) implies that X_m = log Tᵐ(n) is almost surely bounded. Since Tᵐ(n) ∈ ℕ and is decreasing on average, it must eventually enter a finite cycle.

Arithmetic constraint: Any cycle {c₁, c₂, …, cₖ} must satisfy:

∏_{i=1}^{k} 3^{a_i} / 2^{b_i} = 1

where a_i = 1 if c_i odd, 0 if even, and b_i = v₂(3c_i+1) if odd, 1 if even. This constraint severely limits possible cycles [11].

Global rigidity: The 2-adic structure of 3n+1 imposes strong arithmetic constraints. Under the mixing assumptions, the valuation sequence cannot sustain patterns supporting non-trivial cycles [12].

Thus, the trivial cycle 4 → 2 → 1 emerges as the unique arithmetic attractor compatible with negative log drift and global rigidity of the 2-adic dynamics.

Remark 3.10 — Local Randomness vs. Global Rigidity

The Collatz map exhibits a delicate balance: locally, the valuation sequence {k_m} appears random (supporting mixing assumptions), yet globally, the arithmetic structure of 3n+1 restricts admissible symbolic patterns. This tension — between statistical irregularity (which drives negative drift) and deterministic constraint (which eliminates spurious cycles) — is characteristic of near-critical dynamical systems.

Section 4

The Phase Diagram and Critical Threshold

4.1 The Generalized Family T_{a,b}

Definition 4.1

For integers a ≥ 1, b ≥ 0, define the accelerated (a,b)-map:

T_{a,b}(n) = (an + b) / 2^v₂(an+b)

where v₂(m) is the largest k such that 2ᵏ divides m. The standard Collatz map is T_{3,1}. The map T_{5,1} is the "5x+1" problem.

4.2 Expected Drift Under Geometric Valuation

Under the heuristic that v₂(an+b) has geometric distribution 2^(−r) for r ≥ 1, the expected logarithmic drift is:

E[Δ log n] ≈ log a − 2 log 2 = log( a/4 )

This is independent of b (to leading order), depending only on the coefficient a.

4.3 Classification of Regimes

Regime Condition E[Δ log n] Behavior
Contractive a < 4 < 0 Orbits descend on average
Critical a = 4 = 0 Marginal stability, large fluctuations
Expansive a > 4 > 0 Orbits likely diverge

4.4 Collatz as a Near-Critical System

Observation 4.2 — Near-Criticality

The constant log(3/4) appearing in the drift analysis (Section 2) and the critical threshold a = 4 are intimately related:

μ = log 3 − 2 log 2 = log(3/4) = log 3 − log 4

Thus, Collatz sits at a = 3, one unit below criticality. This proximity explains:

  • Large fluctuations: Near a = 4, the variance of ΔV is large (controlled by Var(k) = 2)
  • Slow descent: |μ| = 0.2877 is small compared to log 2 ≈ 0.693, so descent is gradual
  • Computational difficulty: Trajectories can grow to very large values before eventually descending
Remark 4.3

For a = 4, the map T_{4,b} would exhibit zero drift under the geometric model, placing it exactly at criticality. Such maps are conjectured to have divergent trajectories for most n, analogous to critical branching processes.

Section 5

2-Adic Symbolic Dynamics

5.1 Parity Sequences and Symbolic Representation

Construction 5.1 — Parity Encoding

For n ∈ ℕ, define the infinite parity sequence:

ω(n) = (b₀, b₁, b₂, …) ∈ {0,1}^ℕ

where b_m = Tᵐ(n) mod 2. The sequence ω(n) uniquely determines the Collatz orbit.

Observation 5.2

The valuation k_m equals the length of the run of zeros in ω(n) immediately following the m-th occurrence of a 1.

Example 5.3 — n = 7

For n = 7: T(7) = 11 (odd), T²(7) = 17 (odd), T³(7) = 13 (odd), T⁴(7) = 5 (odd), T⁵(7) = 1 (odd), … Parity sequence: ω(7) = (1,1,1,1,1,…). This particular trajectory has k_m = 1 for all steps shown, producing maximal upward drift before eventual descent.

5.2 The Shift Map and 2-Adic Iteration

In the 2-adic integers ℤ₂, the map T extends naturally. The action on binary expansions resembles a shift map with feedback.

Proposition 5.4 — Shift-Like Structure

If n = 2ᵃ · m with m odd, then:

T(n) = 2ᵃ · T(m)

Thus, iteration separates into: (1) stripping trailing zeros (dividing by 2ᵃ); (2) applying the odd map T(m) = (3m+1)/2^v₂(3m+1); (3) reintroducing a new power of 2. This resembles the shift map σ: {0,1}^ℕ → {0,1}^ℕ, but with arithmetic feedback through the 3m+1 operation.

5.3 Entropy and Mixing in Symbolic Dynamics

Heuristic 5.5 — Entropy Model

If ω(n) behaves like a sequence sampled from a shift-invariant measure μ on {0,1}^ℕ with positive entropy, then: ergodicity (time averages equal ensemble averages); mixing (correlations decay exponentially); geometric block lengths (runs of zeros have geometric distribution). Under this model, Assumptions A and B are natural consequences of ergodic-theoretic principles.

Evidence 5.6 — Computational Studies

Studies of parity sequences [6,9,10] show: no detectable long-range correlations; block frequencies consistent with independent coin flips (after appropriate normalization); run lengths of zeros consistent with geometric distribution 2^(−r). These observations support the entropy heuristic, though they do not constitute a proof.

Section 6

Discussion and Open Problems

6.1 Summary of Results

This paper has presented a conditional probabilistic framework for the Collatz conjecture. The main results are:

  1. Drift structure (Section 2): The logarithmic potential V(n) = log n exhibits expected drift μ = log(3/4) < 0 under geometric valuation statistics.
  2. Supermartingale formulation (Section 3): Under Assumptions A and B, the sequence {V(Tᵐ(n))} forms a supermartingale, implying almost sure boundedness.
  3. Critical threshold (Section 4): The Collatz map sits at a = 3 in a family of maps with critical point a = 4, explaining its large fluctuations and slow descent.
  4. Symbolic dynamics (Section 5): The valuation sequence {k_m} encodes block lengths in a 2-adic parity representation, motivating the mixing assumptions via entropy heuristics.
  5. Convergence to trivial cycle (Section 3.6): Under the stated assumptions, every orbit must eventually enter 4 → 2 → 1, the unique arithmetic attractor compatible with negative log drift.

6.2 Relationship to Existing Work

  • Terras [1976]: Showed that the set of finite-time orbits has natural density 1. Our framework explains this via negative drift — almost all trajectories eventually descend because E[Δ log n] < 0.
  • Tao [2019]: Proved that almost all orbits attain almost bounded values. Consistent with our supermartingale result (Corollary 3.8), which asserts boundedness under mixing assumptions.
  • Lagarias [1985]: Studied 2-adic properties and established partial equidistribution results. Our symbolic dynamics (Section 5) builds on this foundation, interpreting Collatz as a shift-like map in ℤ₂.
  • Wirsching [1998]: Developed ergodic-theoretic approaches. Our entropy heuristic (Heuristic 5.5) aligns with Wirsching's framework but makes explicit the connection to valuation statistics.

6.3 Empirical Tests of the Mixing Assumptions

The mixing assumptions (A and B) are empirically testable:

  • Test 1 (Correlation decay): Compute Cov(k_m, k_{m+t}) for large orbits and verify exponential decay ∝ ρ^t
  • Test 2 (Mean stationarity): Verify that the Cesàro mean (1/M) Σ k_m converges to 2 as M → ∞ for random initial n
  • Test 3 (Block frequency): Analyze parity sequences ω(n) and test for consistency with an ergodic measure on {0,1}^ℕ

Computational evidence to date [6,9,10] strongly supports these properties up to 2⁶⁸, but analytic proof remains open.

6.4 Future Research Directions

  • Direction 1 — Rigorous mixing results: Prove that {k_m} satisfies Assumption A (exponential decay of correlations) for all n, or for a set of n with positive density
  • Direction 2 — Symbolic constraints: Characterize which parity sequences ω ∈ {0,1}^ℕ can arise from Collatz orbits; use 2-adic algebra to constrain admissible patterns
  • Direction 3 — Quantitative stopping time bounds: Under Assumptions A and B, derive explicit bounds on the expected stopping time E[τ(n)], where τ(n) = min{m : Tᵐ(n) = 1}
  • Direction 4 — Generalizations: Extend the supermartingale framework to other maps in the (an+b) family, particularly near-critical cases (a close to 4)

6.5 Limitations and Conditional Nature of the Results

This paper does not prove the Collatz conjecture. All results are conditional on Assumptions A and B, which are heuristically motivated but not rigorously established. The contribution is to clarify the structure underlying the heuristic arguments, identify precise conditions under which Collatz convergence follows, and propose a research program for proving these conditions.

Falsifiability: The framework is falsifiable. If future work shows that Assumptions A or B fail for a class of orbits, the supermartingale argument would not apply to those orbits. This is a feature, not a limitation — it transforms an apparently intractable problem into a structured research program.

Section 7

Conclusion

In this framework, the cycle 4 → 2 → 1 appears as the unique arithmetic attractor of a near-critical symbolic dynamical system with negative logarithmic drift.

The Collatz conjecture, viewed through this lens, is not a mysterious pathology but rather a natural consequence of:

  • 2-adic encoding giving parity-based symbolic structure
  • Valuation blocks determining fluctuation statistics
  • Entropy heuristics suggesting geometric block lengths
  • Negative log drift driving long-term descent
  • Global rigidity restricting admissible symbolic patterns

The framework transforms an apparently intractable problem into a structured research program connecting symbolic dynamics, probability theory, and analytic number theory. Whether the mixing assumptions (A and B) can be proven rigorously remains an open question — but the conditional results presented here demonstrate that such a proof would suffice to resolve the Collatz conjecture.

Acknowledgments

Acknowledgments

The author thanks the mathematical community for decades of work on the Collatz conjecture, particularly C. Terras, J. Lagarias, T. Tao, and the computational efforts of researchers verifying convergence up to 2⁶⁸. The Architecture of Infinity framework [7] provided conceptual guidance on identifying incomplete geometric structures in classical problems.

References

References

  1. R. Terras, "A stopping time problem on the positive integers," Acta Arithmetica, vol. 30, no. 3, pp. 241–252, 1976.
  2. J.-P. Allouche, "Sur la conjecture de Syracuse-Kakutani-Collatz," Séminaire de Théorie des Nombres de Bordeaux, pp. 1–15, 1979.
  3. G. J. Wirsching, The Dynamical System Generated by the 3n+1 Function, Springer Lecture Notes in Mathematics 1681, 1998.
  4. I. Krasikov and J. C. Lagarias, "Bounds for the 3x+1 problem using difference inequalities," Acta Arithmetica, vol. 109, no. 3, pp. 237–258, 2003.
  5. T. Tao, "Almost all Collatz orbits attain almost bounded values," arXiv:1909.03562, 2019.
  6. T. Oliveira e Silva, "Empirical verification of the 3x+1 and related conjectures," in The Ultimate Challenge: The 3x+1 Problem, pp. 189–207, American Mathematical Society, 2010.
  7. J. Farrior, Architecture of Infinity: A Structural-Spectral Framework for Eleven Unsolved Problems in Mathematics and Physics, Christos™ Energy, 2026.
  8. J. C. Lagarias, "The 3x+1 problem and its generalizations," American Mathematical Monthly, vol. 92, no. 1, pp. 3–23, 1985.
  9. D. J. Bernstein and J. C. Lagarias, "The 3x+1 conjugacy map," Canadian Journal of Mathematics, vol. 48, no. 6, pp. 1154–1169, 1996.
  10. M. Chamberland, "A continuous extension of the 3x+1 problem to the real line," Dynamics of Continuous, Discrete and Impulsive Systems, vol. 2, pp. 495–509, 1996.
  11. J. C. Lagarias, "The set of rational cycles for the 3x+1 problem," Acta Arithmetica, vol. 56, no. 1, pp. 33–53, 1990.
  12. K. R. Matthews and A. M. Watts, "A generalization of Hasse's generalization of the Syracuse algorithm," Acta Arithmetica, vol. 43, no. 2, pp. 167–175, 1984.
Appendix A

Computational Data

Table A.1 — Empirical Distribution of k(n)

Empirical distribution of k(n) for 10⁶ random odd integers in the range [10⁹, 2×10⁹].

r Frequency P(k=r) Theoretical 2^(−r) Relative Error
10.50010.5000+0.02%
20.24980.2500−0.08%
30.12510.1250+0.08%
40.06240.0625−0.16%
50.03130.0312+0.32%
60.01560.01560.00%
70.00780.00780.00%
≥80.00790.00790.00%

Observation: The empirical distribution closely matches the theoretical geometric distribution 2^(−r), supporting Heuristic Observation 1.1.

Appendix B

Figure Captions

Figure 1 — Logarithmic Trajectories

Logarithmic trajectories of Collatz orbits starting from n₀ ∈ {27, 127, 837799, 1000000}. The x-axis represents iteration number m ∈ [0, 200], and the y-axis represents log₂(Tᵐ(n₀)). Each trajectory exhibits noisy downward drift with occasional upward spikes corresponding to k(n) = 1 steps. The long-term negative drift μ ≈ −0.415 bits/iteration is clearly visible despite local fluctuations. The trajectory starting from 837799 exhibits a particularly long excursion before descent, consistent with near-critical behavior (a = 3 close to a = 4).

Figure 2 — Phase Diagram

Phase diagram of the accelerated map family T_{a,b}(n) = (an+b)/2^v₂(an+b). The horizontal axis shows the parameter a ∈ [1, 6], and the vertical axis shows the expected logarithmic drift E[Δ log n] ≈ log(a/4) ∈ [−1.4, 0.6]. The critical threshold a = 4 (drift = 0) is marked with a vertical dashed line, separating the contractive regime (a < 4, negative drift) from the expansive regime (a > 4, positive drift). Collatz (a = 3, b = 1) is marked at (3, log(3/4) ≈ −0.288). The proximity to criticality explains both the large trajectory fluctuations and the difficulty of proving convergence.

Open Access — Full Paper
This paper is published as an independent research preprint under open access. Free to read and share with attribution. Not derived from, affiliated with, or sponsored by any external institution. All theoretical frameworks, analytical methods, probabilistic formulations, and structural interpretations presented herein originated through independent reasoning conducted by the author.
Authorship & Copyright

A Conditional Supermartingale Framework for the Collatz Dynamics is an original work developed by Joshua Farrior under the CHRISTOS™ master framework. © 2026 Joshua Farrior · Christos™ Energy, Technology & Harmonic Design Consulting, LLC · Indianapolis, Indiana · Business ID: 202511071941923 · Christos™ Trademark Pending USPTO · All Rights Reserved.