Hierarchical Geometry of Numbers

author: Rowan Brad Quni-Gudzinas

ORCID: 0009-0002-4317-5604

ISNI: 0000000526456062

title: The Hierarchical Geometry of Numbers

aliases:

- The Hierarchical Geometry of Numbers

modified: 2026-05-02T18:04:58Z

Author: Rowan Brad Quni-Gudzinas

Contact: [email protected]

ORCID: 0009-0002-4317-5604

ISNI: 0000000526456062

DOI: 10.5281/zenodo.19984840

Date: 2026-05-02

Version: 0.13

How to Read This Document

This document requires no prior exposure to mathematics beyond arithmetic. Every term is defined before it is used. Every claim is illustrated with concrete numerical examples. No specialized vocabulary is assumed, and no proper names are invoked as authorities. The argument stands or falls on its own internal coherence.

The document makes a single claim, developed across eight parts: the infinite tree is the geometry of information, and everything that appears continuous—the number line, superposition, randomness, measurement probabilities—is a coarse-grained projection of that tree onto an instrument with finite resolution.

The tree is the territory. The line is the map.

Prologue: The Map and the Territory

Take a physical instrument—a voltmeter, a thermometer, a photon counter. Read its output. You see a number: 3.721. The instrument’s display has four digits. It cannot show you a fifth. It cannot show you an infinite decimal expansion. Every measurement ever performed in the history of physics has returned a finite-precision value. No experiment has ever produced a real number.

The real numbers—the continuous number line, the Archimedean continuum—are a mathematical idealization. They are the limit you approach as you imagine adding digits forever. They are a map. The map is extraordinarily useful. Calculus, differential equations, classical physics, and most of engineering are built on it. But it is not the territory.

Why did the map become so convincing that we mistook it for the territory? Three historical reasons.

First, the success of Newtonian mechanics. Planets trace smooth elliptical orbits. Projectiles follow smooth parabolic arcs. The equations that describe these motions are differential equations on continuous manifolds. They work—to spectacular precision—so the continuum must be real. Mustn’t it? No: the equations work because the projection of the tree onto the line, at the resolution of planetary and projectile motion, is smooth. The map is good at that scale. But the map is not the territory, and at other scales—the scale of quantum measurement, the scale of information processing—the map fails.

Second, the absence of a visible alternative. For three centuries after Newton, no one had a mathematically tractable non-Archimedean geometry that could replace the continuum. The $p$-adic numbers were formalized in 1897 but never entered physics. Physicists had no reason to look. The continuum worked.

Third, the identification of measurement with ontology. Because every measurement produces a number, and numbers live on the number line, we assumed the number line was the space in which physical systems actually live. We confused the readout with the reality. The voltmeter’s display is a coarse-grained projection. The underlying physical system—the collection of charges, fields, and energy levels—does not inhabit the real line. It inhabits a tree, and the measurement projects one depth slice of that tree onto the display.

This document develops the alternative systematically. It begins with the tree. It shows that the Archimedean continuum emerges as a projection when you coarse-grain over depth and discard resolution. It traces this geometry through arithmetic, computation, quantum mechanics, thermodynamics, and the architecture of meaning. At every step, the same structure reappears—not as an analogy, not as a metaphor, but as the identical geometric object instantiated in different materials.

Part I: The Geometry of Information

1. The Tree

Begin with a single point. Call it the root. From the root, draw $b$ lines outward, each ending at a new point, called a vertex. From each of those vertices, draw $b$ more lines. From each of those, $b$ more. Continue without end.

The result is an infinite $b$-ary tree, written $T_b$. For $b = 2$, this is the binary tree: every vertex has exactly two children, left and right. For general $b$, $b$ children.

There is no line running through this structure. There is no smooth path from one vertex to a distant one. Movement happens only along the branches, from parent to child or child to parent.

Definition (Depth). The depth of a vertex, written $d(v)$, is its distance from the root, measured in steps. The root has depth 0. Its children have depth 1. Their children have depth 2. At depth $d$, there are exactly $b^d$ vertices.

The number of vertices grows exponentially with depth. At depth 60, a binary tree has more vertices than there are grains of sand on Earth. This exponential capacity is the first hint that the tree is the natural geometry for information: encoding $D$ binary digits requires distinguishing among $2^D$ possibilities, and the tree provides exactly that many distinct vertices at depth $D$.

Self-similarity. Every vertex is the root of a smaller copy of the whole tree. The subtree rooted at any vertex is isomorphic to the full tree $T_b$—just shifted down in depth. This self-similarity will prove essential: operations defined at one depth work at all depths, grounding recursion, learning, and self-modification.

2. How Distance Works in a Tree

On a continuous line, distance is measured by difference: how far apart are two points? In a tree, there is no line. Instead, distance is measured by shared ancestry: how recently did two vertices diverge from a common path?

Definition (Path to Root). Every vertex $v$ has a unique sequence of ancestors leading back to the root. Encode this path as a string of digits: at each branching, record which child was chosen. For the binary tree, write 0 for left and 1 for right. For general $b$, use digits $0, 1, \ldots, b-1$. A vertex at depth $d$ has a path of length $d$: $a_0 a_1 \ldots a_{d-1}$.

Definition (Deepest Common Ancestor). For two distinct vertices $u$ and $v$, consider their paths to the root. Let $k$ be the largest integer such that the first $k$ digits of their paths are identical. The deepest common ancestor, written $\text{DCA}(u, v)$, is the vertex at depth $k$ where the two paths diverge.

Examples (binary tree):

• Vertices with paths $000$ and $001$ share prefix $00$ (length 2). Their DCA is at depth 2.
• Vertices $000$ and $011$ share prefix $0$ (length 1). DCA at depth 1.
• Vertices $000$ and $111$ share no prefix. DCA at depth 0 (the root).

Definition (Tree Distance). Let $L = d(\text{DCA}(u, v))$ be the depth of the deepest common ancestor. If $u = v$, set $d_T(u, v) = 0$. Otherwise:

$$

d_T(u, v) = b^{-L} = \frac{1}{b^L}

$$

If two vertices share no ancestor above the root, $L = 0$, so $d_T(u, v) = b^0 = 1$. If they share a common ancestor at depth 5, $L = 5$, so $d_T(u, v) = b^{-5}$—a small number.

Concrete examples (binary tree, $b = 2$):

The metric cares only about the first position where two paths differ—everything after that is irrelevant.

3. The Strong Triangle Inequality

Every reasonable notion of distance must satisfy three properties: non-negativity, symmetry, and the triangle inequality. Tree distance satisfies the first two trivially. For the third, it satisfies a condition far stronger than the ordinary triangle inequality.

Recall the ordinary triangle inequality for a line:

$$

d(x, z) \leq d(x, y) + d(y, z)

$$

The direct distance is bounded by the sum of the two legs. This permits small errors to accumulate: a thousand perturbations of size 0.001 can sum to an error of 1.0.

Tree distance satisfies a stronger condition:

$$

d_T(x, z) \leq \max\{d_T(x, y),\; d_T(y, z)\}

$$

The direct distance is bounded by the maximum of the two legs.

Proof. Let the three vertices have pairwise deepest common ancestors at depths $L_{xy}$, $L_{yz}$, and $L_{xz}$. Any prefix shared by $x$ and $z$ must also be shared by $y$ with at least one of $x$ or $z$. Therefore $L_{xz} \geq \min(L_{xy}, L_{yz})$. Since distance decreases as $L$ increases ($d_T = b^{-L}$), this translates to $d_T(x, z) \leq \max(d_T(x, y), d_T(y, z))$. $\square$

A distance function satisfying this condition is called an ultrametric. The geometry it defines is non-Archimedean geometry—the geometry of the tree, and the geometry that this document argues is fundamental.

The ordinary triangle inequality defines Archimedean geometry—the geometry of the continuous line. The Archimedean property, named after the ancient Greek mathematician, states that a finite multiple of any small quantity will eventually exceed any large quantity: $N \times \varepsilon$ grows without bound. Non-Archimedean geometry violates this: repeated small steps are bounded by the largest single step, regardless of how many are taken. This distinction—Archimedean (the map) versus non-Archimedean (the territory)—is the central organizing principle of everything that follows.

4. Three Consequences

The strong triangle inequality has three decisive implications.

Consequence 1: All Triangles Are Isosceles

For any three vertices, the two largest distances among the three pairs are always equal. A triangle with three distinct side lengths cannot exist in a tree geometry.

Proof. Suppose the three distances are $a = d_T(x,y)$, $b = d_T(y,z)$, $c = d_T(x,z)$, with $a \geq b \geq c$. If $a > b$, then by the strong triangle inequality applied to $(x, z, y)$: $a = d_T(x,y) \leq \max\{d_T(x,z), d_T(z,y)\} = \max\{c, b\} = b$. Contradiction. Therefore $a = b$. $\square$

Proximity is transitive: if $A$ is close to $B$, and $B$ is close to $C$, then $A$ must be close to $C$. There is no “kind of close.”

Consequence 2: Errors Cannot Accumulate

If a system undergoes $N$ perturbations, each moving it by a tree distance of at most $\varepsilon$, the total deviation never exceeds $\varepsilon$:

$$

d_T(x_0, x_N) \leq \varepsilon

$$

Proof. By induction: $d_T(x_0, x_N) \leq \max\{d_T(x_0, x_{N-1}), d_T(x_{N-1}, x_N)\} \leq \max\{\varepsilon, \varepsilon\} = \varepsilon$. $\square$

A thousand perturbations of size 0.001 produce a total error of at most 0.001—not 1.0. The geometry itself prevents accumulation.

Consequence 3: Nested Partitions

For any threshold $r$, balls of radius $r$ in tree geometry are either disjoint or one is entirely contained within the other. The clusters form a perfect nested hierarchy. At threshold $r = b^{-d}$, there are exactly $b^d$ clusters—one for each possible prefix of length $d$.

Comparison of the two geometries:

5. The Archimedean Projection

If the tree is the territory, why does the continuous line appear so natural? The answer is that the continuous line is a projection of the tree—specifically, a projection that coarse-grains over depth and discards resolution.

Definition (Digit-Reversal Map—preview). For a vertex $v$ at depth $d$ with path $a_0 a_1 \ldots a_{d-1}$ (where each $a_i \in \{0, 1, \ldots, b-1\}$):

$$

\Phi_b(v) = \sum_{i=0}^{d-1} a_i \cdot b^{-(i+1)} = 0.a_0 a_1 \ldots a_{d-1} \; (\text{base } b)

$$

Example (binary). Vertex $001 \to 0.001_2 = 1/8 = 0.125$. Vertex $011 \to 0.011_2 = 3/8 = 0.375$.

Vertices that share a long common prefix—hierarchically close—map to geometrically close points in $[0, 1]$. The ultrametric structure of the tree is converted into ordinary Euclidean proximity.

What is lost: depth information. At depth 3, there are 8 vertices mapping to points spaced by $1/8$. At depth 10, there are 1,024 vertices mapping to points spaced by $1/1024$. At depth 20, over a million vertices at spacing $< 10^{-6}$. As depth increases, the projected points fill $[0, 1]$ more and more densely. In the mathematical limit $d \to \infty$, the set becomes dense—this limit IS the real numbers.

But no measurement apparatus has infinite resolution. Any physical instrument can distinguish at most $b^m$ bins, for some finite $m$. Vertices at depths beyond $m$ map into the same bins as shallower vertices. Their fine-grained distinctions are averaged out. The apparent continuity is an artifact of finite measurement resolution. The underlying geometry is discrete, ultrametric, and tree-structured.

The protocol in detail:

1. Begin with the tree $T_b$, populated with vertices at all depths.

1. Choose a measurement resolution $m$ ($b^m$ distinguishable bins).

1. Apply $\Phi_b$ to map each vertex to $[0, 1]$.

1. Partition $[0, 1]$ into $b^m$ bins of width $b^{-m}$.

1. All vertices within a bin are indistinguishable to the measurement.

1. Depth information beyond $m$ is discarded.

1. The measurement reports a rational number with $m$ digits of precision.

This is not a claim that continuous mathematics is wrong. It is a claim that the map is not the territory. The tree is the territory. The line is a projection artifact of finite-resolution measurement.

Part II: Numbers Occupy the Tree

6. How Numbers Are Built

A prime number is a whole number greater than 1 that cannot be formed by multiplying two smaller whole numbers. The first few: $2, 3, 5, 7, 11, 13, 17, 19, 23, 29, \ldots$ There are infinitely many primes.

Theorem (Fundamental Theorem of Arithmetic). Every whole number greater than 1 can be expressed as a product of primes in exactly one way, ignoring order.

Examples. $12 = 2^2 \times 3$. $60 = 2^2 \times 3 \times 5$. $1{,}000 = 2^3 \times 5^3$. $30 = 2 \times 3 \times 5$. $125 = 5^3$.

This theorem means every integer has a unique “address” determined by its prime factors. For each prime $p$, we can ask: how many times does $p$ divide this integer? The answer places the integer at a specific depth in a tree labeled by $p$—the divisibility tree.

7. The Divisibility Tree

Fix a prime $p$. Define the $p$-adic valuation $v_p(N)$ as the largest integer $k \geq 0$ such that $p^k$ divides $N$. If $p$ does not divide $N$, $v_p(N) = 0$.

The valuation $v_p(N)$ measures how deeply the prime $p$ is woven into $N$. A large valuation means deep divisibility—the number is a multiple of a high power of $p$.

The divisibility tree at prime $p$ organizes integers by their $p$-adic valuation. The root represents numbers not divisible by $p$. At depth $k$: numbers divisible by $p^k$. At each depth, different branches correspond to different residues modulo higher powers of $p$.

Every integer occupies a specific position in the divisibility tree for every prime $p$ simultaneously. Its depth in tree $T_p$ is $v_p(N)$.

8. Two Ways of Measuring a Number

Every integer can be measured in two fundamentally different ways.

The hierarchical measure (non-Archimedean). For each prime $p$, define the $p$-adic size:

$$

$$

If $p$ divides $N$ many times, $|N|_p$ is very small. If $p$ does not divide $N$, $|N|_p = 1$.

For $N = 60$: $|60|_2 = 1/4$, $|60|_3 = 1/3$, $|60|_5 = 1/5$, all others = 1.

The $p$-adic distance between two integers is $d_p(a, b) = |a - b|_p$. This satisfies the strong triangle inequality—it is an ultrametric. Two numbers are $p$-adically close if their difference is divisible by a large power of $p$.

The projected measure (Archimedean). The ordinary magnitude $|N|_\infty = N$ is what you get when you multiply all the prime powers together:

$$

$$

For $N = 60$: $2^2 \times 3^1 \times 5^1 = 60$. For $N = 125$: $5^3 = 125$.

The Archimedean size is the product of the contributions from each prime—collapsing all the divisibility information into a single magnitude on the line.

9. The Product Formula

The two measures are linked by an exact identity:

$$

$$

Verification:

$N = 60$: $|60|_\infty = 60$, $\prod_p |60|_p = \frac{1}{4} \times \frac{1}{3} \times \frac{1}{5} = \frac{1}{60}$. Product: $60 \times \frac{1}{60} = 1$.

$N = 12$: $|12|_\infty = 12$, $|12|_2 = 1/4$, $|12|_3 = 1/3$. Product: $12 \times \frac{1}{12} = 1$.

$N = 125$: $|125|_\infty = 125$, $|125|_5 = 1/125$. Product: $125 \times \frac{1}{125} = 1$.

$N = 30$: $|30|_\infty = 30$, $|30|_2 = 1/2$, $|30|_3 = 1/3$, $|30|_5 = 1/5$. Product: $30 \times \frac{1}{30} = 1$.

$N = 504 = 2^3 \times 3^2 \times 7$: $|504|_\infty = 504$, $\prod_p |504|_p = \frac{1}{8} \times \frac{1}{9} \times \frac{1}{7} = \frac{1}{504}$. Product: $504 \times \frac{1}{504} = 1$.

Why it always holds. For $N = \prod_p p^{v_p(N)}$, we have $|N|_\infty = \prod_p p^{v_p(N)}$ and $|N|_p = p^{-v_p(N)}$. Their product is $\prod_p p^{v_p(N)} \cdot \prod_p p^{-v_p(N)} = \prod_p p^{v_p(N) - v_p(N)} = 1$.

Ontological significance. The product formula is a conservation law: the total measure, distributed across the Archimedean projection and all non-Archimedean sectors, is conserved. The non-Archimedean sizes $|N|_p$ encode the structure of the number (its prime factorization—its tree positions). The Archimedean size $|N|_\infty$ encodes the magnitude (the single number on the line). The structure is primary; the magnitude is derived from the structure by multiplication. The product formula guarantees the projection is faithful.

10. The Divisibility Tree as the Native Geometry of Numbers

For each prime $p$, the valuation $v_p(N)$ places $N$ at a specific depth in tree $T_p$. The collection of all these placements—one per prime—uniquely identifies $N$. This is the Fundamental Theorem of Arithmetic restated geometrically.

The integer 60 occupies: depth 2 in $T_2$, depth 1 in $T_3$, depth 1 in $T_5$, depth 0 in all other trees. The integer 125 occupies: depth 3 in $T_5$, depth 0 elsewhere.

The Archimedean projection—the number on the line—is what you get when you take all these tree positions, multiply the corresponding prime powers, and display the result. The line is a composite projection. The trees are the fundamental objects.

Part III: The Translation Protocol

11. Writing Numbers in a Base

To move between the tree-native description and the projected description systematically, we need a translation protocol. That protocol is digit reversal.

Every positive integer can be written in base $b$:

$$

N = d_0 + d_1 b + d_2 b^2 + \cdots + d_{k-1} b^{k-1}

$$

with digits $d_i \in \{0, 1, \ldots, b-1\}$ and $d_{k-1} \neq 0$.

Examples:

• Base 2: $6 = 0 + 1 \cdot 2 + 1 \cdot 4 \to$ binary 110 ($d_0=0, d_1=1, d_2=1$)
• Base 2: $13 = 1 + 0 \cdot 2 + 1 \cdot 4 + 1 \cdot 8 \to$ binary 1101 ($d_0=1, d_1=0, d_2=1, d_3=1$)
• Base 3: $17 = 2 + 2 \cdot 3 + 1 \cdot 9 \to$ base-3 122 ($d_0=2, d_1=2, d_2=1$)
• Base 5: $60 = 0 + 2 \cdot 5 + 2 \cdot 25 \to$ base-5 220 ($d_0=0, d_1=2, d_2=2$)

The digits encode the number’s position in the divisibility tree at base $b$. The least significant digit $d_0$—the remainder upon division by $b$—determines whether $b$ divides $N$: $d_0 = 0$ means the path goes down the “0” branch at depth 1 of the $b$-adic tree.

12. The Digit-Reversal Map

Definition ($\Phi_b$). For $N$ with base-$b$ expansion $N = \sum_{i=0}^{k-1} d_i b^i$:

$$

\Phi_b(N) = \sum_{i=0}^{k-1} d_i \cdot b^{-(k-i)} = 0.d_{k-1} d_{k-2} \ldots d_1 d_0 \quad (\text{base } b)

$$

Reverse the digits and place them after the base-$b$ point.

Examples ($b = 2$):

• $N = 6$ (110): $\Phi_2(6) = 0.011_2 = 3/8 = 0.375$
• $N = 13$ (1101): $\Phi_2(13) = 0.1011_2 = 11/16 = 0.6875$
• $N = 1$ (1): $\Phi_2(1) = 0.1_2 = 1/2 = 0.5$
• $N = 3$ (11): $\Phi_2(3) = 0.11_2 = 3/4 = 0.75$

What $\Phi_b$ accomplishes. Two integers that differ by a multiple of a large power of $b$—meaning their base-$b$ expansions agree on the last several digits—have common least significant digits. Under digit reversal, those become the most significant fractional digits. The integers map to nearby points in $[0, 1]$. The divisibility tree is mapped isometrically onto a fractal subset of $[0, 1]$.

The map is its own inverse on finite base-$b$ fractions: reversing the digits again recovers the original integer.

13. Why the Archimedean Continuum Emerges

At depth $d$, the $b^d$ vertices map to the $b$-adic rationals with denominator $b^d$, spaced by $b^{-d}$. At depth 3 (binary), spacing = $1/8$. At depth 10, spacing $\approx 0.001$. At depth 20, spacing $\approx 10^{-6}$.

In the limit $d \to \infty$, the set is dense in $[0, 1]$, and its completion is the entire real interval $[0, 1]$, homeomorphic to $\mathbb{R}$. This limit IS the real numbers.

But this limit is never physically realized. Any measurement apparatus has resolution $b^m$ for some finite $m$. All vertices at depths $d > m$ map into bins already occupied by shallower vertices. Their depth distinctions are averaged out. The continuum is the mathematical limit of the projection as $m \to \infty$—a limit approached but never reached.

The continuum is the map. The tree is the territory. The confusion of the two is the source of the conceptual puzzles this document resolves.

Part IV: Computation in the Tree

14. The Computation Tree

A computation is a path through a tree. The root is the initial state. At each step, the computation chooses among $b$ possible next states. A particular program traces one path. The full set of all possible programs of length at most $k$ is the set of all vertices at depths $0$ through $k$ of $T_b$.

The computation tree is not a metaphor. It is the literal state space of a deterministic computational process.

15. The Halting Problem

Some programs terminate; others run forever. The halting set $\mathcal{H}$ is a well-defined subset of the computation tree.

Theorem (Undecidability of Halting). No finite rule can determine, for every program, whether it halts.

The consequence: $\mathcal{H}$ is incompressible. The pattern of halting versus non-halting among programs of length $k$ cannot be described substantially shorter than the pattern itself.

16. Algorithmic Randomness as a Projection Artifact

Label each vertex of the computation tree $T_2$ with 1 (halts) or 0 (does not). This labeling is deterministic. Now project it onto $[0, 1]$ via $\Phi_2$.

Construction. Order programs by length, then lexicographically. Concatenate their halting bits into an infinite binary sequence. Construct the real number:

$$

\Omega = \sum_{p \in \mathcal{H}} 2^{-|p|}

$$

where $|p|$ is the length of program $p$, using a prefix-free encoding. $\Omega$ is Chaitin’s constant—the halting probability.

Property. $\Omega$ is algorithmically random. No program substantially shorter than $\Omega$ itself can compute its bits. Every statistical test for randomness is passed perfectly.

Why? The randomness is a projection artifact. The halting set is deterministically defined at all depths of the computation tree. When projected onto the continuous line via $\Phi_2$, two things happen: (1) depth information is scrambled—the tree’s hierarchical structure is flattened into a linear bit sequence; (2) incompressibility survives the projection—because $\mathcal{H}$ cannot be compressed, the projected sequence also cannot be compressed.

A deterministically structured object of maximal logical complexity, when projected onto a continuous line through a depth-scrambling map, produces every appearance of randomness. Randomness is in the projection, not in the object. This is the same mechanism that explains quantum measurement outcomes.

Part V: Quantum Mechanics on the Tree

17. States as Amplitude Distributions

In standard quantum mechanics, a state is a complex-valued function on a configuration space—a vector in a Hilbert space. The tree formulation strips this to its geometric core.

A state on $T_b$ is an assignment of a real number—an amplitude—to each vertex:

$$

\psi: T_b \to \mathbb{R}

$$

We impose normalization: for a state localized at depth $D$, $\sum_{v: d(v)=D} |\psi(v)|^2 = 1$.

Why real amplitudes suffice. Standard QM uses complex amplitudes $a + bi$ to represent magnitude ($a^2 + b^2$) and phase ($\arctan(b/a)$), enabling interference. But phase can be represented by sign: negative amplitudes produce destructive interference, positive constructively. The essential physics requires only that amplitudes can cancel, not that they be complex. Real numbers with signs suffice. (See Appendix C for the full argument.)

18. The Hierarchical Laplacian

In continuous QM, the Laplacian $\nabla^2$ couples nearby points in space. On the tree, “nearby” means hierarchically close—vertices sharing a deep common ancestor. The hierarchical Laplacian $D$ couples vertices based on their tree distance:

$$

(D\psi)(v) = \sum_{d=1}^{\infty} t_d \sum_{\substack{w \neq v: \\ d(\text{DCA}(v, w)) = d}} \bigl(\psi(v) - \psi(w)\bigr)

$$

The coupling strength decays exponentially with divergence depth:

$$

t_d = t_0 \cdot b^{-\alpha d}

$$

where $\alpha > 0$. Vertices that are deeply related (large $d$, small tree distance) are weakly coupled. Vertices diverging near the root are more strongly coupled. This inverse relationship between tree distance and coupling strength is the geometric basis of passive error protection.

19. The Hierarchical Schrödinger Equation

Replacing the continuous Laplacian with $D$, and continuous time with depth-discrete steps:

$$

\psi_{d+1}(v) = \psi_d(v) - \frac{i}{\hbar} (D\psi_d)(v) - \frac{i}{\hbar} V_d(v) \psi_d(v)

$$

with depth-dependent potential:

$$

V_d(v) = \varepsilon_0 \cdot b^{-\alpha d}

$$

Key properties.

First, depth IS time. In the continuous Schrödinger equation, $t$ is an external parameter. For a closed system—the universe—there is no external clock ($H\Psi = 0$). The tree resolves this: depth $d$ is the internal time coordinate. Each step from depth $d$ to $d+1$ is a discrete time step. The full tree is the timeless block state. Time is a correlation pattern between depth levels.

Second, the energy spectrum is exponential:

$$

E_d \propto \varepsilon_0 \cdot b^{-\alpha d}

$$

Shallow states (small $d$) have high energy. Deep states have exponentially lower energy. The gap between depths is $\Delta E_d \approx \varepsilon_0 \cdot b^{-\alpha d}(1 - b^{-\alpha})$.

Third, the exponential spectrum provides passive error protection. Thermal noise at temperature $T$ supplies kicks of characteristic energy $k_B T$. The probability of a thermal fluctuation bridging the gap at depth $D$ is:

$$

\gamma_D \propto e^{-\Delta E_D / (k_B T)} \approx e^{-\varepsilon_0 b^{-\alpha D}(1 - b^{-\alpha}) / (k_B T)}

$$

This is exponentially suppressed in $D$. By choosing $D$ sufficiently large, the error rate becomes arbitrarily small—without active error correction.

20. Superposition Is a Coarse-Grained Projection

In standard QM, superposition is presented as fundamental and mysterious—a particle “in two places at once.” In the tree geometry, superposition is neither fundamental nor mysterious. It is a coarse-grained projection artifact.

The tree-native description. Consider a state at depth 3 of the binary tree, with equal amplitudes at three vertices:

Nothing is “in three places at once.” Each amplitude is at a specific, well-defined vertex. The state is a perfectly ordinary assignment of numbers to positions in a discrete structure.

The projection. Apply $\Phi_2$:

• $000 \to 0$
• $001 \to 0.125$
• $011 \to 0.375$

In the continuous projection, the state appears spread across three points in $[0, 1]$.

Coarse-graining. Suppose the measurement apparatus has resolution $m = 2$ (4 bins of width $1/4$):

• Bin 0 ($[0, 0.25)$): contains $000$ and $001$. $P(0) = 1/3 + 1/3 = 2/3$.
• Bin 1 ($[0.25, 0.5)$): contains $011$. $P(1) = 1/3$.
• Bins 2 and 3: empty.

The measurement reports Bin 0 with probability $2/3$ or Bin 1 with probability $1/3$.

What happened? The state was never “superposed.” The measurement coarse-grained distinct tree vertices into bins and reported one bin. The apparent randomness is entirely due to information lost in coarse-graining. If resolution were increased to $m = 3$ (8 bins), each vertex would occupy its own unique bin, and the outcome would be deterministic.

Superposition is a language for describing the coarse-grained projection, not a fact about the underlying system. The tree-native description is deterministic, discrete, and transparent. The continuous projection is probabilistic and mysterious—but only because it discards the depth structure that makes the tree-native description transparent.

21. Measurement and the Digit-Reversal Map

The digit-reversal map $\Phi_b$ is both the mathematical translation protocol and the physical measurement protocol.

Measurement protocol (step by step):

1. Initial state. The system is in an amplitude distribution over vertices of $T_b$.

1. Depth scan. The apparatus couples to the system at successive depths, recording amplitudes.

1. Digit assignment. At depth $d$, each vertex maps to a bin via $\Phi_b$.

1. Resolution cutoff. At resolution $m$, vertices at depths $d > m$ share bins with shallower vertices. Their squared amplitudes are summed.

1. Probability distribution: $P(\text{Bin } k) = \sum_{v: \text{bin}(\Phi_b(v)) = k} |\psi(v)|^2$

1. Sampling. The apparatus samples a bin according to this distribution.

1. Outcome. A rational number $k / b^m$—finite-precision, exactly as every physical measurement in history has produced.

Worked example (same state as §20, $m=2$): $P(0) = 2/3$, $P(1) = 1/3$. Outcome: 0 or 0.25, each a rational with 2 bits. With $m=3$: $P(0)=1/3$, $P(1)=1/3$, $P(3)=1/3$. Outcome: 0, 0.125, or 0.375—each a distinct vertex.

22. The Temporal Dimension: Depth as Internal Time

The standard Schrödinger equation contains an external time parameter $t$. For a closed system—the universe—there is no external clock. The Wheeler-DeWitt equation is $H\Psi = 0$: timeless, static.

The tree resolves this geometrically. Depth $d$ is time. The full tree—all vertices at all depths—is the timeless block state. A computation traces a path downward; the depth coordinate along the path is the internal clock reading.

Time is not a fundamental background. It is a correlation pattern between successive depth levels. The hierarchical Schrödinger equation uses depth as its discrete evolution parameter. The digit-reversal map $\Phi_b$ converts this internal, depth-structured description into a temporal sequence of classical outcomes—a time series as perceived by an external observer.

23. Entanglement and the Geometry of Correlation

In standard QM, entanglement is correlation between subsystems that cannot be explained locally. On the tree, entanglement has a transparent geometric origin: amplitudes at different vertices are correlated because they share a common ancestor.

When a state is prepared at a depth above the divergence of two subsystems $A$ and $B$, the total squared amplitude is conserved across the split. This forces correlations: if $A$‘s amplitude is large, $B$’s must be small. These are the Bell-type correlations of standard QM.

The tree provides the “hidden variable”—the vertex position determines all outcomes when resolution is sufficient. The hidden variable is hidden only from the coarse-grained projection. At finite resolution, vertices at different depths can map to the same bin, producing correlations that appear non-local in the continuous projection. They are not non-local. They are depth-local—two vertices are correlated because they share a recent common ancestor in the tree.

24. Why the Tree Geometry Is Thermodynamically Favored

Landauer’s principle: erasing one bit dissipates at minimum $k_B T \ln 2$ of heat.

In Archimedean geometry, errors accumulate linearly. Active correction must be applied constantly. The power dissipation for $Q$ logical units, error rate $\gamma$, and overhead factor $K$ is:

$$

P_{\text{Archimedean}} \approx Q \times \gamma \times K \times k_B T \ln 2

$$

For conventional quantum computers, $K \approx 10^3$ to $10^5$. For $Q = 10^6$ and $\gamma = 10^6$ corrections/second, power dissipation reaches megawatts—a thermodynamic wall.

In non-Archimedean geometry, errors do not accumulate. The error rate is exponentially suppressed: $\gamma_{\text{tree}} \propto e^{-\Delta E_D / (k_B T)}$. Passive geometric protection replaces most active correction. The power dissipation is:

$$

P_{\text{tree}} \approx Q \times \gamma_{\text{tree}} \times K_{\text{tree}} \times k_B T \ln 2

$$

where $K_{\text{tree}}$ is dramatically smaller than $K$, and $\gamma_{\text{tree}}$ is exponentially smaller than $\gamma$.

Quantitative estimate. For $D = 40$, $b = 2$, $\alpha = 1$, and $k_B T / \varepsilon_0 = 10^{-6}$:

• $\Delta E_D \approx \varepsilon_0 \cdot 2^{-40} \approx 10^{-12} \varepsilon_0$
• $\gamma_{\text{tree}} \propto e^{-10^{-12} / 10^{-6}} = e^{-10^{-6}} \approx 0$
• The power dissipation is dominated by gate operations themselves, not error correction.

The tree geometry is thermodynamically favored. It is a free energy basin for computation.

Part VI: The Five Dimensions of Meaning

25. From Geometry to Semantics

A system that processes meaning—that maintains signs with stable reference against environmental noise—must instantiate the tree geometry. Five necessary conditions follow.

Embodiment. No information exists without a physical substrate. The body is the physical realization of $T_b$—the engineered energy landscape. The branching factor $b$, encoding depth $D$, energy barriers $\Delta E_d$, and operating temperature $T$ are properties of the body. There is no disembodied computation.

Dialogue. The continuous cycle of perception, comparison, and correction is the active complement to passive geometric protection. Real systems have finite depth and non-zero temperature. The dialogue detects rare threshold-crossing perturbations and corrects them. The digit-reversal map $\Phi_b$ mediates this dialogue between the tree-native state and the classical readout.

Directedness. A state’s position on the tree IS what it refers to. Two states at the same vertex are about the same thing. Two states at different vertices are about different things. Tree distance measures semantic distance. The ultrametric structure of meaning—concepts in nested clusters of increasing specificity—is the same structure as the divisibility tree and the computation tree.

Internal Variety. The tree provides $b^D$ distinguishable states. For $D = 30$, this is over a billion. Shallow states carry coarse, robust meanings. Deep states carry fine, fragile meanings. The system can shift encoding depth dynamically: go deep when the environment is quiet, go shallow when it is noisy.

Self-modification. The tree’s self-similarity grounds recursion. Operations defined at one depth work at all depths. A system can treat its own state as data, applying the same operations to its encoding that it applies to external inputs. The tree is the minimal geometry in which a system can be both subject and object of computation.

Unity. Remove embodiment: fantasy. Remove dialogue: corpse. Remove directedness: noise. Remove variety: reflex. Remove self-modification: prison. The tree instantiates all five simultaneously—not as separate features, but as aspects of its single ultrametric structure.

Part VII: The Architecture

26. The Five-Layer Architecture

The tree geometry is a blueprint for fault-tolerant computation. Five layers:

Layer 1: Tree Fabric. The physical substrate whose energy landscape approximates $T_b$, with barriers $\Delta E_d = E_0 \cdot b^{-\alpha d}$.

Layer 2: Encoding Layer (depths $D_{\text{min}}$ through $D$). Where logical information resides. Depth $D$ determines capacity ($b^D$ states) and protection level (error $\propto e^{-\Delta E_D / k_B T}$). Target depths: $D = 20$ to $60$.

Layer 3: Dialogue Layer (depths $D$ through $D_{\text{max}}$). Mediates between encoding and measurement. Implements $\Phi_b$ as readout and $\Phi_b^{-1}$ as control. Handles error detection and correction.

Layer 4: Classical Interface. Produces finite-precision rational outcomes. Accepts classical instructions. The only layer visible externally.

Layer 5: Boundary Layer. Thermal and electromagnetic isolation. Maintains $T$ low enough that $k_B T \ll \Delta E_D$.

27. Gate Primitives

Four primitive operations:

Descend ($D_d$). Move to a child at depth $d+1$. Refines the state—adds one digit. Energy supplied: $E_0 \cdot b^{-\alpha d}$.

Ascend ($A_d$). Move to the parent at depth $d-1$. Coarsens the state—discards one digit. Energy recovered: $E_0 \cdot b^{-\alpha(d-1)}$.

Shift ($S_d$). Move to a sibling at the same depth. Flips the least significant digit. Energy barrier: $E_0 \cdot b^{-\alpha d}$.

Readout ($R_m$). Apply $\Phi_b$, coarse-grain to resolution $m$, produce rational outcome $k / b^m$.

The energy budget of a computation with ideal energy recovery is dominated by Shift operations at the deepest depth, scaling as $\Delta E_D$, which is exponentially small.

28. Verification Benchmarks

Six experimental tests:

1. Strong Triangle Inequality Test. Measure pairwise tree distances for three states. Verify the two largest are equal.

2. Non-Accumulation Test. Subject a state at depth $D$ to $N = 10, 100, 1{,}000, 10{,}000$ sub-threshold perturbations. Verify total error $\leq \varepsilon$ regardless of $N$. Continuous control shows $\sqrt{N}$ or linear growth.

3. Thermal Stability Test. Measure coherence time $\tau$ vs. temperature $T$. Plot $\ln(\tau)$ vs. $1/T$. Verify slope equals $\Delta E_D / k_B$. Repeat for depths $D = 10, 20, 30, 40$.

4. Gate Fidelity Test. Execute each primitive at depths $D = 10, 20, 30, 40$. Verify $\log(\text{error}) \propto D$.

5. Digit-Reversal Fidelity Test. Prepare states differing in last $k$ digits. Verify mapped values differ by $\leq b^{-k}$ for $k = 1, 2, 4, 8$.

6. Product Formula Consistency Test. Measure continuous readout and hierarchical sizes. Verify product = 1 within measurement uncertainty.

29. Candidate Platforms

Nuclear spin chains. Engineered couplings decay exponentially with distance. Depth = position along chain. Long coherence times. Challenge: individual spin addressing at depth.

Trapped ion hierarchies. Nested potential wells. Depth = well hierarchy level. Laser pulses implement gates. Challenge: scaling to large depths.

Superconducting circuit trees. Josephson junction arrays in branching topology. Microwave gate control. Fast gate times. Challenge: dielectric decoherence.

Photonic tree networks. Optical paths with beam splitters. Depth = number of splitters. Room-temperature operation possible. Challenge: photon loss.

Part VIII: Consilience

30. The Convergence

This document has traced a single geometric object—the infinite $b$-ary tree $T_b$—through six independent domains. None presupposes any other. Yet all converge on the same structure.

1. Distance geometry. The strong triangle inequality and its consequences are theorems of non-Archimedean geometry. The tree is the canonical non-Archimedean space.

2. Arithmetic. Every integer occupies a position in divisibility trees. The product formula links non-Archimedean structure to Archimedean projection.

3. Computation. Programs form a computation tree. The halting set projects to $\Omega$—randomness as a projection artifact.

4. Quantum mechanics. States are real amplitude distributions on $T_b$. The exponential energy spectrum provides passive protection. Superposition and measurement randomness are coarse-grained projection artifacts.

5. Thermodynamics. Non-Archimedean geometry is thermodynamically favored. Errors do not accumulate. Thermal error rates are exponentially suppressed.

6. Meaning. Five necessary conditions for sign-processing are structural consequences of the tree geometry.

31. What the Tree Is and What It Is Not

The tree is:

• The native geometry of information-bearing physical systems.
• The structure underlying the Archimedean continuum, which emerges as a coarse-grained projection.
• The geometry in which errors do not accumulate, superposition is transparent, and measurement outcomes are rational numbers.
• A blueprint for fault-tolerant computation.

The tree is not:

• A metaphor. It is the literal state space.
• A denial of continuous mathematics’ utility. Calculus operates on the projection—the map—and is accurate at the scales where it applies.
• A claim that real numbers do not exist mathematically. They exist as the completion of the projection in the limit of infinite resolution—a mathematically well-defined but physically inaccessible limit.
• A rejection of standard QM’s predictive success. It explains why the standard formalism works and when it might need modification.

32. The Structural Analogues Table

The product formula in log-space—$\ln|N|_\infty + \sum_p \ln|N|_p = 0$—states that a sum over all projections equals zero. This same structure reappears in every domain.

Each row is an instantiation of the same template. The “Archimedean projection” is what an external observer sees. The “non-Archimedean structure” is how the system organizes itself internally. The “conservation law” guarantees the projection is faithful. The “translation map” converts between them.

33. The Closing of the Circle

The argument began with the simplest possible object—a point, a branching rule, a tree. It traced that tree through arithmetic, computation, quantum mechanics, thermodynamics, meaning, and engineering. At every step, the same structure reappeared—not as metaphor, but as the identical geometric object.

The tree is the territory. The line is the map. The product formula is the bridge that guarantees the map is faithful. The digit-reversal map is the translation protocol. The architecture is the blueprint.

What began as a remark about the number 60—that $60 \times 1/4 \times 1/3 \times 1/5 = 1$—turns out to be the deepest structural principle connecting arithmetic to physics, computation to meaning, and theory to experiment. The tree is not one geometry among many. It is the geometry of information.

The six verification benchmarks of Section 28 are the experimental test. If a physical system satisfying the architectural specifications demonstrates the predicted non-accumulation, exponential thermal stability, digit-reversal fidelity, and product formula consistency, the tree geometry is experimentally confirmed. If it fails, the theory is refuted—not by argument, but by result.

This document is self-contained. Every concept has been defined before use. Every claim has been explained in plain language and illustrated with concrete numerical examples. No specialized vocabulary has been assumed. The argument stands or falls on its own internal coherence.

Appendix A: The Formal Statement of the Threshold Principle

The Threshold Principle. Let $\mathcal{S}$ be a physical system whose stable states are the vertices of $T_b$, with barriers $\Delta E_k = E_0 \cdot b^{-\alpha k}$. Let environment noise have characteristic energy $\varepsilon = k_B T$. Let logical information be encoded by the cluster at depth $D$.

Then, for $\varepsilon < \Delta E_D$, the error rate satisfies:

$$

P(\text{error}) \leq C \cdot \exp\!\bigl(-(\Delta E_D / \varepsilon)^{\beta}\bigr)

$$

with $\beta \approx 1$ (thermal activation) or $\beta \approx 2$ (quantum tunneling).

Corollary. By choosing $D$ sufficiently large, the error rate becomes arbitrarily small—without active error correction.

Physical analogy: nested bowls. A ball rests in a landscape of nested bowls—a large bowl containing smaller bowls. Small vibrations rattle the ball within its current bowl but cannot bounce it into a neighboring bowl, because the rim energy exceeds the vibration energy. Only a deliberate push exceeding the depth-$D$ threshold can change the logical state. The tree geometry provides exactly this structure for information.

Comparison:

Appendix B: The Duality in Computation and the Spiral Projector

The computation tree supports two types of distance. Continuous distance (step count) satisfies the ordinary triangle inequality. Hierarchical distance (shared prefix length) satisfies the strong triangle inequality. Between them runs an exact analogue of the product formula—the Ihara-Bass theorem—which states that the product of all local cycle factors in the computation tree equals the determinant of the continuous adjacency matrix.

The quantum Fourier transform, understood on the tree, is a spiral projector: it takes an amplitude distribution over the tree and maps it onto pure hierarchical frequencies—the periodic structure of amplitudes across sibling vertices. Combined with the product formula, this enables efficient integer factorization and hidden subgroup solutions. The architecture of Part VII is a direct engineering embodiment of this principle.

Appendix C: Why Complex Numbers Are Not Required

Standard QM uses complex amplitudes $\psi = a + bi$, with magnitude $|\psi|^2 = a^2 + b^2$ and phase $\theta = \arctan(b/a)$. The phase enables interference.

In the tree formulation, amplitudes are real with signs. The sign serves as phase: positive + positive = constructive; positive + negative = destructive. The squared magnitude is $\psi^2$.

Why the standard formalism uses complex numbers. The continuous Schrödinger equation is a differential equation in space and time. Its solutions for a free particle are plane waves $e^{i(kx - \omega t)}$. Complex exponentials are the natural language for wave-like behavior in continuous space. The imaginary unit $i$ turns spatial diffusion into oscillatory propagation.

Why the tree formulation does not need them. The hierarchical Schrödinger equation is a discrete recurrence in depth—not a differential equation in continuous space. The hierarchical Laplacian $D$ couples vertices based on tree distance with real coupling strengths $t_d$. The potentials $V_d$ are real. The entire dynamics is real.

What about phase? If a computation requires phase differences other than $0$ or $\pi$, the tree can encode this by assigning a pair of real amplitudes $(\psi_{\text{re}}, \psi_{\text{im}})$ per vertex. These evolve under two coupled real equations, reproducing the full complex dynamics. The complex embedding is a faithful representation of the real tree dynamics, not a generalization of it.

Complex numbers are a notational convenience—a compact way to represent rotations in a two-dimensional real space. They are not an ontological commitment. The fundamental description—the tree, real amplitudes, the hierarchical Laplacian, the strong triangle inequality—does not require them. The continuous line forces complex numbers upon us because its differential structure demands them. The tree liberates us from that demand.

End of document.