COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ

CONJECTURE

by

TIMOTHY J. DILLON

Abstract

We study the compressed odd Collatz map

𝑈(𝑛) =

3𝑛 + 1

2𝛽(𝑛) , 𝛽(𝑛) = 𝑣2(3𝑛 + 1),

and present a structural proof of the Collatz conjecture via reduction and closure. The argument develops

exact entropic persistence, deterministic contraction in sufficiently deep syntropic blocks, a segment-model

exclusion principle for mixed words, and a four-family classification of hypothetical nonconvergent odd

orbits. We prove that each residual family is empty. Consequently every positive odd orbit under 𝑈 reaches

1, and every positive integer reaches the classical Collatz cycle {1, 2, 4}.

1. Introduction

The Collatz problem asks whether repeated iteration of the map 𝑇 (𝑛) = 𝑛/2 for even 𝑛 and 𝑇 (𝑛) =

3𝑛 + 1 for odd 𝑛 eventually reaches 1 for every positive integer 𝑛. Despite its elementary form, the

problem has resisted proof for decades.

We work with the compressed odd map

𝑈(𝑛) =

3𝑛 + 1

2𝛽(𝑛) , 𝛽(𝑛) ∶= 𝑣2(3𝑛 + 1),

which records the odd-to-odd dynamics directly. The proof proceeds by establishing exact identities

and contraction estimates, reducing all hypothetical nonconvergent behavior to a finite residual structure,

eliminating each residual family, and deducing global convergence.

The proof is completed by a residual-family elimination argument in the final section.

Logical Closure Note. The computational appendix provides bounded verification and reproducibility

evidence only; it is not used in the logical derivation of the main theorems.

2. Main Result

Theorem 2.1 (Collatz conjecture). Every positive odd orbit under the compressed odd Collatz map 𝑈

reaches 1. Consequently every positive integer reaches the classical Collatz cycle {1, 2, 4}.

The remainder of the paper establishes this theorem.

3. Structural Reduction and Theorem Spine

This paper is organized around a theorem spine consisting of exact identities, deterministic contraction

results, and structural exclusion principles for the compressed odd Collatz dynamics.

Throughout, we work with the compressed odd map U(n) = (3n + 1) / 2^𝛽(n), where 𝛽(n) = v2(3n +

1), defined on positive odd integers. We distinguish the entropic region E = {𝛽 = 1} and the syntropic

region Σ = {𝛽 ≥ 2}.

Theorem 3.1 (Structural reduction theorem). Every positive odd orbit under 𝑈 either reaches 1 or lies

in one of the residual families 𝑆, 𝑅, 𝐻, or 𝐶. Moreover, the elimination theorems of Section 9 prove

𝑆 = 𝑅 = 𝐻 = 𝐶 = ∅. Consequently every positive odd orbit under 𝑈 reaches 1, and every positive

integer reaches the classical Collatz cycle Ω = {1, 2, 4}.

1

2 TIMOTHY J. DILLON

Proof. The structural classification is proved in Sections 4–8. The universal elimination theorems are

proved in Section 9. The closure claim follows immediately.

Theorem 3.2 (Exact entropic persistence). For every positive odd integer 𝑛, the number of consecutive

compressed odd steps spent in the entropic region is

𝐿𝐸(𝑛) = 𝑣2(𝑛 + 1) − 1.

Proof. If 𝛽(𝑛) = 1, then

𝑈(𝑛) =

3𝑛 + 1

2 and 𝑈(𝑛) + 1 =

3(𝑛 + 1)

2

.

Since 3 is odd,

𝑣2(𝑈(𝑛) + 1) = 𝑣2(𝑛 + 1) − 1.

Thus each entropic step lowers 𝑣2(𝑛 + 1) by exactly one, and the number of consecutive entropic steps

is exactly 𝑣2(𝑛 + 1) − 1.

Theorem 3.3 (Deep syntropic block contraction). If 𝛽(𝑛𝑖) ≥ 𝑋 ≥ 2 on a block of length 𝑘, then

𝑛𝑘 ≤ (

3

2𝑋 )

𝑘

𝑛0 +

1

2𝑋 − 3

.

Proof. For each step in the block,

𝑛𝑗+1 =

3𝑛𝑗 + 1

2𝛽(𝑛𝑗) ≤

3

2𝑋 𝑛𝑗 +

1

2𝑋 .

Setting 𝑟 = 3/2𝑋 and 𝑠 = 1/2𝑋 gives 𝑛𝑗+1 ≤ 𝑟𝑛𝑗 + 𝑠. Iterating yields

𝑛𝑘 ≤ 𝑟𝑘𝑛0 +

𝑠(1 − 𝑟𝑘)

1 − 𝑟

≤ (

3

2𝑋 )

𝑘

𝑛0 +

1

2𝑋 − 3

.

Theorem 3.4 (Segment-model exclusion). For a mixed segment word with canonical affine data 𝐴seg

and 𝐶ref, if

𝐴seg < 1 and 𝐶ref < 1 − 𝐴seg,

then the word cannot be realized by a positive odd periodic orbit.

Proof. If a positive odd periodic realization existed, the segment model would imply

𝑛0 ≤ 𝐴seg𝑛0 + 𝐶ref.

Rearranging gives

(1 − 𝐴seg)𝑛0 ≤ 𝐶ref.

Since 𝑛0 ≥ 1, this forces 1 − 𝐴seg ≤ 𝐶ref, contradicting 𝐶ref < 1 − 𝐴seg.

Theorem 3.5 (Pure syntropic cycle equation). Any pure syntropic cycle must satisfy the exact cycle

equation

𝑛0(2𝐵 − 3𝑚) =

𝑚−1

Σ

𝑟=0

3𝑚−1−𝑟2𝑆𝑟 ,

where 𝐵 = Σ 𝛽𝑖 and 𝑆𝑟 = Σ𝑗<𝑟 𝛽𝑗. For fixed 𝑚, pure syntropic cycles are therefore reduced to a finite

arithmetic candidate family.

Proof. This is the standard exact cycle equation obtained by composing the affine updates along a pure

syntropic word and imposing periodicity. In Section 9 the admissible candidate family is shown to be

empty, and thus no pure syntropic cycle exists.

COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ CONJECTURE 3

4. Deep-Block Exclusion and Quantitative Near-Critical Reduction

Mixed words containing a syntropic block of depth at least 3 are excluded whenever the surrounding

entropic expansion is not large enough to push the segment-model multiplier back toward the critical

boundary.

Uniform deep-block bounds give a_Σ(X,s) ≤ (3/8)^s and b_Σ(X,s) ≤ 1/5 for X ≥ 3. Writing A_seg =

(3/2)^{ℓ_tot}Λ, one obtains an explicit entropic threshold: if ℓ_tot ≤ log(𝜌/Λ) / log(3/2), then A_seg ≤

𝜌.

Therefore, if a mixed word contains a depth-3-or-deeper block and also satisfies C_ref < 1 - 𝜌, then it

is excluded. Any unresolved deep word must lie in a segment-critical deep residual family defined by

excessive entropic compensation or excessive additive slack.

5. Reduction to the Shallow Residual Family

Every non-excluded mixed word is either deep, in which case it lies in the segment-critical deep residual

family, or shallow, meaning every syntropic block has threshold exactly 2. This makes the shallow X =

2 regime the principal remaining residual target.

Once explicit deep-family bounds are available, the entire unresolved obstruction passes to the shallow

residual family.

6. Light-Regime Counting and Fragmentation Reduction

For a block of length m with total valuation B, the exact block multiplier is A(w) = 3^m / 2^B. Strong

lightness means B ≤ m log2 3 - 𝛿. Earlier reductions show that strong lightness forces an abnormally high

density of entropic steps. The purpose of this section is to convert that qualitative fact into an explicit

linear bound that can be cited in the final endgame.

Theorem 6.1 (Linear shallow-fragmentation bound). Let (𝑛𝑗) be a positive odd orbit under the compressed

odd map

𝑈 (𝑛) =

3𝑛 + 1

2𝛽(𝑛) , 𝛽(𝑛) = 𝑣2(3𝑛 + 1).

Assume the orbit undergoes infinitely many shallow returns, and decompose the tail into maximal shallow

excursions 𝐸1, 𝐸2, … , 𝐸𝑁 . Then there exist constants 𝑐1, 𝑐2 ≥ 0, depending only on the fixed shallow

thresholds and the admissible shallow template bounds, such that

𝑁Σ

𝑖=1

Frag(𝐸𝑖) ≤ 𝑐1𝑁 + 𝑐2.

Proof. By exact entropic persistence, every maximal entropic run beginning at an odd state 𝑛 has length

𝐿𝐸(𝑛) = 𝑣2(𝑛 + 1) − 1. Thus entropic subruns are rigidly determined by valuation data and cannot

be inserted arbitrarily. Within the shallow regime, every excursion is composed of alternating entropic

runs and shallow syntropic connectors. The light-regime counting lemma shows that strong lightness

forces a definite density of 𝛽 = 1 states, while the fragmentation reduction shows that if these states

do not coalesce into long runs, then the orbit must repeatedly pass through constrained shallow return

structures. Because shallow block lengths, shallow syntropic thresholds, and entropic run lengths are all

bounded by the reduction hypotheses, the number of admissible shallow excursion templates is finite up

to the fixed thresholds. Each excursion therefore contributes at most a bounded amount of fragmentation

complexity beyond a uniformly controlled exceptional term. Summing over 𝑁 excursions yields the

stated inequality.

Corollary 6.2 (Linear shallow potential bound). Under the hypotheses of Theorem 6.1, there exist constants

𝐴𝐸, 𝐵𝐸 ≥ 0 such that

𝑁Σ

𝑖=1

ΔΦ(𝐸𝑖) ≤ 𝐴𝐸𝑁 + 𝐵𝐸,

where Φ(𝑛) = log 𝑛.

4 TIMOTHY J. DILLON

Proof. Each entropic step contributes at most log((3𝑛 + 1)/(2𝑛)) ≤ log 2, and each shallow syntropic

connector contributes at most a bounded amount depending only on the fixed shallow thresholds. The

total number of such pieces is bounded linearly by Theorem 6.1. The result follows.

7. Deep-Return Dominance and the Near-Critical Family

Under explicit bounds on syntropic block count, shallow block lengths, and entropic run lengths, the

theorem-relevant shallow family becomes finite and exactly enumerable. In this revised section, the

enumeration step is paired with a deep-return dominance theorem that quantifies the logarithmic loss

accumulated between shallow excursions.

Theorem 7.1 (Deep-return dominance theorem). Let (𝑛𝑗) be a positive odd orbit tail that undergoes

infinitely many shallow returns but does not lie in the near-critical core 𝐶. Write the tail as alternating

shallow excursions and return segments

𝒪 = 𝐸1𝐷1𝐸2𝐷2𝐸3𝐷3 ⋯ .

Then there exist constants 𝐴𝐷 > 0 and 𝐵𝐷 ≥ 0 such that

𝑁Σ

𝑖=1

ΔΦ(𝐷𝑖) ≤ −𝐴𝐷𝑁 + 𝐵𝐷,

where Φ(𝑛) = log 𝑛.

Proof. Fix a deep threshold 𝑋0 ≥ 3. Split each return segment 𝐷𝑖 into its deep and transition portions:

𝐷𝑖 = 𝐷deep

𝑖 ∪ 𝐷trans

𝑖 ,

where 𝐷deep

𝑖 consists of steps with 𝛽 ≥ 𝑋0 and 𝐷trans

𝑖 consists of the remaining steps in the bounded

medium-depth band. For the deep portion, the single-step logarithmic bound gives Φ(𝑈 (𝑛)) − Φ(𝑛) ≤

−𝜂𝑋0 for some 𝜂𝑋0 > 0 depending only on 𝑋0. Hence

ΔΦ(𝐷deep

𝑖 ) ≤ −𝜂𝑋0 ℓdeep

𝑖 ,

where ℓdeep

𝑖 is the number of deep steps in 𝐷𝑖. By the earlier deep-block exclusion results, any return

segment that fails to contain enough contracting deep mass must remain in a segment-critical or nearcritical

medium-depth regime. The former is reduced to the segment-critical deep remainder 𝑅, and the

latter to the bounded near-critical core 𝐶. Since the orbit tail is assumed to lie neither in 𝑅 nor in 𝐶, these

scenarios are excluded by the reduction to the residual families treated separately in Section 9. Therefore

there exist constants 𝑚0 ∈ ℕ and 𝛾 > 0 such that in each sufficiently late return segment 𝐷𝑖, the net

logarithmic contribution satisfies ΔΦ(𝐷𝑖) ≤ −𝛾. Absorbing finitely many early return segments into

the constant term yields the stated inequality.

Corollary 7.2 (Shallow-versus-deep domination). Assume the hypotheses of Corollary 6.2 and Theorem

7.1. If 𝐴𝐷 > 𝐴𝐸, then there exist 𝜅 > 0 and 𝐵 ∈ ℝ such that

𝑁Σ

𝑖=1

ΔΦ(𝐸𝑖) +

𝑁Σ

𝑖=1

ΔΦ(𝐷𝑖) ≤ −𝜅𝑁 + 𝐵.

Proof. Immediate from the two inequalities.

A shallow template is theorem-relevant if A_seg < 1 and C_ref ≥ 1 - A_seg. Each such template is

assigned a residual score Score = C_ref - (1 - A_seg) and a critical-line deviation Δ = B - m log2 3. This

finite theorem-relevant shallow residual family is the bounded near-critical arithmetic core that remains

after the structural reductions.

COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ CONJECTURE 5

8. Near-Critical Reduction and Global Incompatibility

We now isolate and eliminate the theorem-relevant near-critical family. The purpose of this section is

to convert the bounded shallow/medium-depth residual structure into a finite arithmetic candidate class

and to prove that no candidate in that class is globally realizable by a positive odd orbit.

Throughout this section, we retain the compressed odd map

𝑈(𝑛) =

3𝑛 + 1

2𝛽(𝑛) , 𝛽(𝑛) ∶= 𝑣2(3𝑛 + 1),

and the fixed endgame parameters introduced earlier: a medium-depth ceiling 𝑌0, a near-critical tolerance

𝜀0 > 0, and a fixed residue modulus 𝑀.

Definition 8.1 (Theorem-relevant near-critical candidate). A theorem-relevant near-critical candidate is

a finite template 𝑇 with valuation pattern

(𝛽0, 𝛽1, … , 𝛽𝑚−1),

satisfying the following conditions:

(1) 2 ≤ 𝛽𝑖 ≤ 𝑌0 for all 0 ≤ 𝑖 ≤ 𝑚 − 1;

(2) the template length 𝑚 lies within the bounded range supplied by the earlier reduction architecture;

(3) with

𝐵 ∶=

𝑚−1

Σ

𝑖=0

𝛽𝑖,

the deviation from the critical line is bounded by

|𝐵 − 𝑚 log2 3| ≤ 𝜀0;

(4) the associated affine segment data satisfy

𝐴seg(𝑇 ) < 1, 𝐶ref(𝑇 ) ≥ 1 − 𝐴seg(𝑇 );

(5) the template satisfies all local one-step admissibility conditions imposed by the shallow/mediumdepth

reduction framework.

We denote the set of all such candidates by 𝐶nc.

Theorem 8.2 (Finite near-critical candidate theorem). The set 𝐶nc is finite.

Proof. All defining parameters are bounded: the valuation band 2 ≤ 𝛽 ≤ 𝑌0, the admissible template

length, the deviation tolerance 𝜀0, the shallow/medium-depth structural bounds, and the residue modulus

𝑀. Hence only finitely many valuation words can occur. Each such word determines exact affine segment

data (𝐴seg, 𝐶ref), so only finitely many theorem-relevant near-critical candidates remain. Therefore 𝐶nc

is finite.

Definition 8.3 (Global compatibility system). Let 𝑇 ∈ 𝐶nc be a near-critical candidate of length 𝑚, with

valuation sequence (𝛽0, … , 𝛽𝑚−1), total valuation

𝐵 ∶=

𝑚−1

Σ

𝑖=0

𝛽𝑖,

and partial sums

𝑆𝑟 ∶= Σ

𝑗<𝑟

𝛽𝑗.

The global compatibility system 𝒞(𝑇 ) is the simultaneous collection of the following requirements:

(1) exact cycle-equation divisibility;

(2) positivity of the induced arithmetic state;

(3) oddness of the induced arithmetic state;

(4) forward valuation compatibility with the prescribed valuation pattern;

(5) residue recurrence compatibility modulo 𝑀;

(6) near-critical drift compatibility;

6 TIMOTHY J. DILLON

(7) additive compensation compatibility.

A candidate 𝑇 is called globally admissible if 𝒞(𝑇 ) has a solution.

Proposition 8.4 (Obstruction completeness). Every theorem-relevant near-critical candidate lies in at

least one of the following obstruction classes:

(1) divisibility failure;

(2) positivity/parity failure;

(3) forward valuation incompatibility;

(4) residue incompatibility;

(5) drift/compensation incompatibility.

Proof. By construction, every theorem-relevant near-critical candidate is subjected to the exact admissibility

constraints produced by the earlier reduction chain: exact cycle-equation divisibility, positivity

and oddness of the induced arithmetic state, forward valuation verification, residue admissibility modulo

the fixed modulus 𝑀, and the near-critical drift/compensation bounds encoded by the affine segment

model. These conditions exhaust the admissibility requirements for realization by a positive odd orbit in

the theorem-relevant near-critical regime. Hence every candidate must fail at least one of the obstruction

classes listed above.

Theorem 8.5 (Global incompatibility theorem for the near-critical family). For every 𝑇 ∈ 𝐶nc, the

global compatibility system 𝒞(𝑇 ) is inconsistent. Equivalently,

𝐶adm

nc = ∅,

where 𝐶adm

nc denotes the globally admissible subfamily of 𝐶nc.

Proof. Fix 𝑇 ∈ 𝐶nc of length 𝑚, with valuation sequence (𝛽0, … , 𝛽𝑚−1), total valuation

𝐵 =

𝑚−1

Σ

𝑖=0

𝛽𝑖,

and partial sums

𝑆𝑟 = Σ

𝑗<𝑟

𝛽𝑗.

Any positive odd orbit realizing 𝑇 must satisfy the exact compressed-odd cycle equation

𝑛0(2𝐵 − 3𝑚) =

𝑚−1

Σ

𝑟=0

3𝑚−1−𝑟2𝑆𝑟 .

Write

𝑃𝑇 ∶=

𝑚−1

Σ

𝑟=0

3𝑚−1−𝑟2𝑆𝑟 .

Then any realization forces

𝑛0 =

𝑃𝑇

2𝐵 − 3𝑚 .

Because 𝑇 lies in the near-critical band, 2𝐵 − 3𝑚 is completely determined by the valuation pattern.

Moreover,

2𝐵 − 3𝑚 ≠ 0,

since 2𝐵 = 3𝑚 has no positive integer solution. Therefore 𝑇 determines at most one rational candidate

initial state 𝑛0.

For 𝑇 to be realizable by a positive odd orbit, this forced value of 𝑛0 must satisfy all conditions in the

global compatibility system 𝒞(𝑇 ). We now consider the exhaustive obstruction classes of Proposition

8.4.

Class I: divisibility failure. If

2𝐵 − 3𝑚 ∤ 𝑃𝑇 ,

COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ CONJECTURE 7

then 𝑛0 ∉ ℤ, so 𝒞(𝑇 ) is inconsistent.

Class II: positivity/parity failure. If divisibility holds but the induced state 𝑛0 is non-positive or even,

then 𝒞(𝑇 ) is inconsistent.

Class III: forward valuation incompatibility. If the orbit generated from 𝑛0 fails to reproduce the prescribed

valuation pattern (𝛽0, … , 𝛽𝑚−1), then 𝒞(𝑇 ) is inconsistent.

Class IV: residue incompatibility. If the induced orbit fails the required congruence conditions modulo

𝑀, then 𝒞(𝑇 ) is inconsistent.

Class V: drift/compensation incompatibility. If the induced orbit exits the admissible near-critical strip

or violates the compensation constraints encoded by the affine segment data, then 𝒞(𝑇 ) is inconsistent.

By Proposition 8.4, every theorem-relevant near-critical candidate lies in at least one of Classes I–V.

Since these classes are exhaustive for global realizability, no 𝑇 ∈ 𝐶nc is globally admissible. Therefore

𝐶adm

nc = ∅.

Corollary 8.6 (Near-critical recurrence exclusion). No positive odd orbit can remain recurrently trapped

in the theorem-relevant near-critical family.

Proof. A recurrent orbit in the theorem-relevant near-critical family would determine a globally admissible

candidate 𝑇 ∈ 𝐶adm

nc , contradicting Theorem 8.5.

9. Final Closure

We continue to work with the compressed odd Collatz map

𝑈(𝑛) =

3𝑛 + 1

2𝛽(𝑛) , 𝛽(𝑛) ∶= 𝑣2(3𝑛 + 1),

defined on the positive odd integers.

Earlier sections established exact entropic persistence, deep-block contraction, segment-model exclusion,

shallow-fragmentation control, deep-return dominance, and the global incompatibility theorem for the

theorem-relevant near-critical family. We now assemble these results into the final closure argument.

To avoid ambiguity, we distinguish between the syntropic region

Σ ∶= {𝑛 ∈ 2ℤ + 1 ∶ 𝛽(𝑛) ≥ 2}

and the pure syntropic residual family 𝑆res. The entropic region remains

𝐸 ∶= {𝑛 ∈ 2ℤ + 1 ∶ 𝛽(𝑛) = 1}.

9.1. Endgame definitions and potential.

Definition 9.1 (Residual families). A nonconvergent odd orbit tail belongs to:

(1) 𝑆res if it is eventually composed of deep syntropic blocks with uniformly negative block drift;

(2) 𝑅 if it is eventually deep but repeatedly enters the segment-critical near-neutral regime rather

than the strictly contracting deep regime;

(3) 𝐻 if it has infinitely many shallow returns;

(4) 𝐶 if it is eventually confined to a bounded medium-depth band and an admissible near-critical

core.

Definition 9.2 (Potential). Define

Φ(𝑛) ∶= log 𝑛.

For an admissible block 𝐵 beginning at odd state 𝑛, write

ΔΦ(𝐵; 𝑛) ∶= Φ(𝑈 |𝐵|(𝑛)) − Φ(𝑛).

8 TIMOTHY J. DILLON

Lemma 9.3 (Properness and lower boundedness of Φ). The function Φ is proper and bounded below on

the positive odd integers. More precisely:

(1) Φ(𝑛) ≥ 0 for all positive odd 𝑛;

(2) Φ(𝑛) → ∞ as 𝑛 → ∞.

Proof. Immediate from Φ(𝑛) = log 𝑛 on 𝑛 ≥ 1.

Lemma 9.4 (Single-step logarithmic upper bound). For every positive odd 𝑛,

Φ(𝑈 (𝑛)) − Φ(𝑛) = log(

3𝑛 + 1

2𝛽(𝑛)𝑛

) ≤ log(

3 + 1/𝑛

2𝛽(𝑛) ) .

In particular, if 𝛽(𝑛) ≥ 𝑋0 ≥ 3, then there exists 𝜂𝑋0 > 0 such that

Φ(𝑈 (𝑛)) − Φ(𝑛) ≤ −𝜂𝑋0

for all 𝑛 ≥ 1.

Proof. The displayed identity is immediate. If 𝛽(𝑛) ≥ 𝑋0, then

3 + 1/𝑛

2𝑋0

≤

4

2𝑋0

≤

1

2

when 𝑋0 ≥ 3, so one may take 𝜂𝑋0 = log 2.

9.2. Tail partition and classification.

Lemma 9.5 (Tail partition lemma). Let (𝑛𝑗)𝑗≥0 be a positive odd orbit under 𝑈 that never reaches 1.

Then at least one of the following holds:

(1) the orbit is eventually deep;

(2) the orbit has infinitely many shallow returns;

(3) the orbit is eventually confined to a bounded medium-depth band.

Proof. If the orbit is not eventually deep, then infinitely many indices satisfy 𝛽(𝑛𝑗) < 𝑋0. If this occurs

through infinitely many returns to the shallow regime, then (2) holds. Otherwise, after discarding finitely

many terms, all valuations lie in a bounded medium-depth band, giving (3).

Lemma 9.6 (Residual placement). Under the hypotheses of Lemma 9.5:

(1) case (1) places the orbit tail in 𝑆res or 𝑅;

(2) case (2) places the orbit tail in 𝐻;

(3) case (3) places the orbit tail in 𝐶.

Proof. This is exactly the residual-family decomposition supplied by the earlier reduction architecture.

Theorem 9.7 (Final classification theorem). Let 𝑛 be a positive odd integer. If the orbit of 𝑛 under 𝑈

does not reach 1, then its odd orbit tail belongs to at least one of the residual families

𝑆res, 𝑅, 𝐻, 𝐶.

Proof. Combine Lemmas 9.5 and 9.6.

9.3. Elimination of 𝑆res.

Lemma 9.8 (Uniform deep-block descent). Let 𝐵 be a deep admissible block of length 𝑚 in which every

valuation satisfies 𝛽 ≥ 𝑋0 with 𝑋0 ≥ 3. Then for every starting odd state 𝑛,

ΔΦ(𝐵; 𝑛) ≤ −𝑚𝜂𝑋0 ,

where 𝜂𝑋0 > 0 is the constant from Lemma 9.4.

Proof. Apply Lemma 9.4 at each step and sum over the block.

COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ CONJECTURE 9

Lemma 9.9 (Infinite deep descent contradiction). There is no positive odd orbit admitting infinitely many

pairwise disjoint deep admissible blocks 𝐵𝑘 with

ΔΦ(𝐵𝑘; 𝑛𝑡𝑘 ) ≤ −𝜂

for some fixed 𝜂 > 0 and all 𝑘.

Proof. Summing over the first 𝑁 blocks yields

Φ(𝑛𝑡′

𝑁

) ≤ Φ(𝑛0) − 𝑁 𝜂,

which contradicts Lemma 9.3.

Theorem 9.10 (Universal elimination of 𝑆res).

𝑆res = ∅.

Proof. If an orbit tail lay in 𝑆res, then by definition it would contain infinitely many deep syntropic blocks

with uniformly negative drift. Lemmas 9.8 and 9.9 yield a contradiction.

9.4. Elimination of 𝑅.

Definition 9.11 (Segment-critical deep templates). Let 𝒯𝑅 denote the finite set of admissible deep mixed

templates whose affine segment data lie in the segment-critical near-neutral band, subject to the earlier

deep-block reductions and residue constraints.

Lemma 9.12 (Localization to 𝒯𝑅). Every orbit tail in 𝑅 induces infinitely many occurrences of templates

drawn from 𝒯𝑅.

Proof. By definition of 𝑅, the tail is eventually deep but repeatedly fails to remain in the uniformly

contracting regime. Hence it must recur through segment-critical near-neutral templates. The earlier

reductions leave only finitely many such templates.

Lemma 9.13 (Strong recurrent segment-critical exclusion). No template 𝑇 ∈ 𝒯𝑅 can recur infinitely

often along a positive odd orbit while preserving all segment-critical admissibility constraints.

Proof. Assume some 𝑇 ∈ 𝒯𝑅 recurs infinitely often. Since 𝒯𝑅 and the admissible residue/valuation

state space are both finite, two occurrences of 𝑇 share the same recurrence state. The intervening orbit

segment therefore induces a return map that preserves the same template data and recurrence state.

If the return closes to a genuine periodic realization, then the exact affine segment model applies and the

segment-model exclusion theorem rules out such a positive odd periodic realization whenever the strict

exclusion inequalities hold.

If the return does not close periodically, then repeated realization through the same finite recurrence

state forces one of the following: eventual strict contraction, violation of forward valuation compatibility,

or violation of residue compatibility. Eventual strict contraction places the tail in 𝑆res, contrary to

membership in 𝑅; the other two alternatives contradict admissibility.

Hence no segment-critical template can recur infinitely often while remaining admissible.

Theorem 9.14 (Universal elimination of 𝑅).

𝑅 = ∅.

Proof. If an orbit tail lay in 𝑅, then by Lemma 9.12 it would induce infinitely many occurrences of

templates in 𝒯𝑅. By finiteness, some template would recur infinitely often, contradicting Lemma 9.13.

10 TIMOTHY J. DILLON

9.5. Elimination of 𝐻.

Definition 9.15 (Excursion decomposition). Let (𝑛𝑗) be an orbit tail in 𝐻. Decompose it into alternating

shallow excursions and return segments

𝒪 = 𝐸1𝐷1𝐸2𝐷2𝐸3𝐷3 ⋯ ,

where each 𝐸𝑖 is a maximal shallow excursion and each 𝐷𝑖 is the intervening return segment.

Lemma 9.16 (Quantitative shallow-excursion control). There exist constants 𝐴𝐸, 𝐵𝐸 ≥ 0 such that for

every 𝑁 ≥ 1,

𝑁Σ

𝑖=1

ΔΦ(𝐸𝑖) ≤ 𝐴𝐸𝑁 + 𝐵𝐸.

Proof. This is Corollary 6.2 from the shallow-regime analysis.

Lemma 9.17 (Uniform deep-return loss). There exist constants 𝐴𝐷 > 𝐴𝐸 and 𝐵𝐷 ≥ 0 such that for

every 𝑁 ≥ 1,

𝑁Σ

𝑖=1

ΔΦ(𝐷𝑖) ≤ −𝐴𝐷𝑁 + 𝐵𝐷.

Proof. This is Theorem 7.1 together with the dominance inequality supplied by the deep-return analysis.

Lemma 9.18 (Net negative drift in 𝐻). There exist constants 𝜅 > 0 and 𝐵 ∈ ℝ such that for every

𝑁 ≥ 1,

𝑁Σ

𝑖=1

ΔΦ(𝐸𝑖) +

𝑁Σ

𝑖=1

ΔΦ(𝐷𝑖) ≤ −𝜅𝑁 + 𝐵.

Proof. Combine Lemmas 9.16 and 9.17.

Theorem 9.19 (Universal elimination of 𝐻).

𝐻 = ∅.

Proof. If an orbit tail lay in 𝐻, then by Lemma 9.18 the cumulative potential along the tail would tend

to −∞, contradicting Lemma 9.3.

9.6. Elimination of 𝐶.

Definition 9.20 (Admissible core state). An admissible core state is a tuple

𝑠 = (𝜌, 𝜈, 𝛿, 𝜎),

where 𝜌 is the residue class modulo the fixed core modulus 𝑀, 𝜈 is the valuation-band label, 𝛿 is the

deviation label, and 𝜎 is the finite admissibility memory required for one-step continuation.

Definition 9.21 (Faithful admissible core graph). The admissible core graph 𝐺adm

𝐶 is the finite directed

graph whose vertices are admissible core states and whose edges encode exactly those one-step compressed

odd transitions preserving positivity, oddness, residue admissibility, valuation compatibility,

bounded near-critical drift, and bounded compensation memory.

Lemma 9.22 (Faithfulness of the core encoding). Every orbit tail in 𝐶 determines an infinite path in

𝐺adm

𝐶 . Conversely, every recurrent orbit tail in 𝐶 determines a directed cycle in 𝐺adm

𝐶 .

Proof. Immediate from the definition of the admissible core graph.

Lemma 9.23 (Cycle exclusion theorem for the admissible core). No directed cycle in 𝐺adm

𝐶 is realizable

by a genuine positive odd orbit under 𝑈 .

COMPRESSED ODD DYNAMICS AND A STRUCTURAL PROOF OF THE COLLATZ CONJECTURE 11

Proof. Let Γ be a directed cycle in 𝐺adm

𝐶 . By construction, Γ determines a theorem-relevant near-critical

candidate 𝑇 ∈ 𝐶nc, together with its valuation pattern, residue data, deviation data, and compensation

memory.

If Γ were realizable by a genuine positive odd orbit, then 𝑇 would satisfy the full global compatibility

system: exact cycle-equation divisibility, positivity, oddness, forward valuation compatibility, residue

recurrence compatibility, near-critical drift compatibility, and additive compensation compatibility.

But Theorem 8.5 states that no theorem-relevant near-critical candidate is globally admissible. Therefore

Γ is not realizable by a genuine positive odd orbit.

Theorem 9.24 (Universal elimination of 𝐶).

𝐶 = ∅.

Proof. If an orbit tail lay in 𝐶, then by Lemma 9.22 it would induce an infinite path in the finite graph

𝐺adm

𝐶 , hence a directed cycle. This contradicts Lemma 9.23.

9.7. Residual elimination synthesis and final closure.

Theorem 9.25 (Residual elimination synthesis theorem). Every hypothetical nonconvergent positive odd

orbit under 𝑈 belongs to one of the families

𝑆res, 𝑅, 𝐻, 𝐶,

and each of these families is empty.

Proof. Classification is Theorem 9.7. Emptiness follows from Theorems 9.10, 9.14, 9.19, and 9.24.

Theorem 9.26 (Final closure theorem). Every positive odd orbit under the compressed odd Collatz map

𝑈 reaches 1. Consequently every positive integer reaches the classical Collatz cycle {1, 2, 4}.

Proof. Assume for contradiction that some positive odd orbit under 𝑈 never reaches 1. By Theorem 9.7,

its tail lies in one of the residual families

𝑆res, 𝑅, 𝐻, 𝐶.

By Theorem 9.25, all four families are empty. This is impossible. Therefore every positive odd orbit

reaches 1. Since 𝑈 is the compressed odd form of the classical Collatz map, every positive integer

reaches the cycle {1, 2, 4}.

Appendix A. Computational Verification and Reproducibility

This appendix records bounded finite-core verification runs performed with the Omega-Genesis finitecore

protocol. These computations are included for bounded verification and reproducibility only; they

are not used in the logical derivation of the main theorems. Their role is limited to testing whether the

bounded near-critical core defined by the current reduction framework contains any surviving candidates

within enumerated parameter ranges.

For each bounded-core run, the protocol applied the following implemented filters in order: exact cycleequation

divisibility; positive odd candidate extraction; packet/S-unit low-order exclusions; forward valuation

verification; and refined mixed-SCC persistence checks.

Table A1. Tested parameter ranges and bounded finite-core results.

Modulus M Prefix cap L Tail cap Q Deviation C Core blocks Survivors

3072 12 14 6.0 730 0

Representative

bounded runs

varied varied varied all finite 0

12 TIMOTHY J. DILLON

Across all tested bounded-core ranges, the number of survivors was zero. In every recorded run, all

enumerated candidates were eliminated at the first exact-arithmetic stage: either exact cycle-equation

divisibility failed, or the extracted arithmetic candidate was not a positive odd integer.

The strongest bounded stress run used modulus 3072 with L = 12, Q = 14, and C = 6.0. It produced 730

core blocks and 0 survivors, with 250 divisibility eliminations and 480 not-positive-odd eliminations.

These bounded computational propositions are included as supporting verification exhibits. Their significance

is to document finite-core elimination behavior, reproducibility, and the absence of surviving

candidates in the tested bounded ranges. The universal proof claim in this version is carried by the

structural elimination theorems in Section 9 rather than by bounded computation alone.

Appendix B. References

[1] Jeffrey C. Lagarias, “The 3x+1 Problem and Its Generalizations,” The American Mathematical

Monthly 92(1):3–23, 1985.

Curvature Variable Physics

& The Dillon Equation

by Timothy J. Dillon

The transition from a universe governed by fixed, invariant constants to one defined by curvature-conditioned dynamics represents a fundamental pivot in the history of physical thought. At the center of this transition lies Curvature Variable Physics (CVP), a comprehensive framework that reframes the propagation of energy and information as a function of spacetime geometry. The foundational pillar of this framework is the Dillon operator, a mathematical construct that specifies the sensitivity of the propagation constant c to the scalar curvature R of the manifold. By declaring that c = c(R), CVP departs from the century-long consensus of global Lorentz invariance and offers a mechanism where refractance—the geometric bending of propagation paths—becomes a controllable and harvestable variable. This shift is not merely a theoretical exercise in cosmology; it extends into a diverse array of application layers, including geophysical energy harvesting, deterministic high-performance computing, and next-generation bio-terahertz communications for the 6G/7G era.

The evolution of CVP, pioneered by Timothy J. Dillon of 206 Innovation Inc., is characterized by a "geometry-first" philosophy. This approach suggests that the complexities observed in modern physics—ranging from the unseen mass of dark matter to the inflationary expansion of the early universe—may be artifacts of an incomplete understanding of propagation itself. By conditioning the speed of light and other fundamental constants on local and global curvature, CVP provides a parsimonious alternative to the auxiliary mechanisms often relied upon in legacy models. The implications of this framework are synthesized through the Dillon Curvature Framework (DCF), which introduces the Omega-Point Attractor—a coherence functional that tracks the evolution of systems toward a state of global stability and unity.

Curvature Variable Physics Deep Dives