Derivation 29: The Mediant Is Not an Axiom

Derivation 29: The Mediant Is Not an Axiom#

The challenge#

The entire framework rests on the mediant (a+c)/(b+d) being the primitive combining operation. Derivation 10 declared it a primitive. Derivation 28’s proof of Ω_Λ = 13/19 is conditional on this.

If the mediant is merely an axiom — a choice — then the framework is one choice among many, and the predictions are coincidences that follow from a lucky choice.

This derivation shows: the mediant is the UNIQUE operation satisfying two physical properties of coupled oscillators. It is not chosen. It is forced.

Two physical properties#

Property 1: Betweenness#

When two oscillators at frequencies ν₁ and ν₂ couple, the resulting locked frequency ν lies between them:

min(ν₁, ν₂) ≤ ν ≤ max(ν₁, ν₂)

This is not an assumption. It is energy conservation. A coupled system cannot produce a frequency outside the range of its inputs without an external energy source. The coupling redistributes energy between the two oscillators; it does not create energy.

In terms of frequency ratios: if the two oscillators have winding numbers a/b and c/d (with a/b < c/d), then the locked ratio ω satisfies:

a/b ≤ ω ≤ c/d

Property 2: Minimality (stability)#

Among all possible locked frequencies between ν₁ and ν₂, the system locks to the one with the SMALLEST DENOMINATOR.

This is not an assumption. It is the Arnold tongue structure of the circle map. The tongue width at rational p/q scales as:

w(p/q, K) ~ (K/2)^q

The tongue width DECREASES exponentially with the denominator q. A mode with smaller q has a wider tongue — it is stable over a larger range of bare frequency and coupling. The coupled system enters the widest available tongue first, because it is the first one reached as coupling increases from zero.

The most stable lock is the simplest lock. The simplest fraction between two given fractions (the one with the smallest denominator) is the first to appear as coupling increases. This is not a principle of economy or aesthetics — it is the topology of Arnold tongues.

The theorem#

Theorem (Stern-Brocot, 1858/1860). Let a/b and c/d be adjacent fractions (|ad − bc| = 1). The unique fraction in the open interval (a/b, c/d) with the smallest denominator is:

(a + c) / (b + d)

the mediant.

Proof sketch. By the theory of continued fractions, the fraction with the smallest denominator in any interval (α, β) of width |β − α| = 1/(bd) (where b, d are the denominators of the endpoints) has denominator b + d. Its numerator is a + c (forced by the requirement that it lie in the interval and be irreducible). The Farey adjacency condition |ad − bc| = 1 ensures that no fraction with smaller denominator exists in the interval. □

(Full proof: Hardy & Wright, “An Introduction to the Theory of Numbers,” Chapter III; or Brocot’s original construction.)

The derivation#

Step 1. Physical systems of coupled oscillators satisfy betweenness (energy conservation) and minimality (Arnold tongue stability).

Step 2. The Stern-Brocot theorem says: the unique operation satisfying betweenness and minimality on adjacent rationals is the mediant.

Step 3. Therefore the mediant is the unique combining operation for coupled oscillator frequency ratios.

The mediant is not an axiom. It is a theorem about the ONLY operation consistent with energy conservation and Arnold tongue stability applied to frequency ratios.

What this replaces#

Derivation 10 listed four primitives:

  1. Integers Z

  2. Mediant (a+c)/(b+d)

  3. Fixed-point x = f(x)

  4. Parabola x² + μ = 0

With this derivation, primitive (2) is replaced by:

2’. Coupled oscillators satisfy betweenness and minimality.

The mediant is then a DERIVED operation — the unique one consistent with (2’). The framework’s primitives become:

  1. Integers Z (counting)

  2. Coupled oscillators with betweenness and minimality (→ mediant)

  3. Fixed-point x = f(x) (self-reference)

  4. Parabola x² + μ = 0 (bifurcation)

Primitive (2’) is more physical and less algebraic than (2). It refers to energy conservation and stability — properties that can be tested experimentally — rather than to an algebraic operation that must be taken on faith.

The different shapes#

Different combining operations on pairs (a, b) produce different algebraic structures:

Operation

Formula

Structure

Physical meaning

Complex multiplication

(a,b)·(c,d) = (ac−bd, ad+bc)

ℂ

Rotation + scaling

Quaternion multiplication

4-component

ℍ

3D rotation

Component-wise multiplication

(ac, bd)

Coordinate scaling

Independent axes

Mediant (component-wise addition)

(a+c, b+d)

Stern-Brocot tree

Mode-locking

Each operation answers a different physical question:

  • Complex multiplication: “what happens when you compose two rotations?” → phase composition

  • Mediant: “what happens when two oscillators couple?” → frequency locking

The physical context determines the operation. For coupled oscillators (the Kuramoto model, the framework’s substrate), the relevant question is mode-locking, not rotation. The mediant is the answer to the mode-locking question. Complex multiplication is the answer to the rotation question. They are different because the physics is different.

The “shape” of each algebra:

  • Complex numbers: the unit circle (S¹). Multiplication preserves the circle.

  • Mediants: the Stern-Brocot tree. The mediant preserves Farey adjacency.

These are genuinely different topological structures. S¹ is a smooth manifold. The Stern-Brocot tree is a discrete binary tree. The framework uses the tree, not the circle, because the physical process (synchronization) produces a tree of rational lockings, not a smooth rotation.

The chain, axiom-free#

With the mediant derived from betweenness + minimality:

Energy conservation + Arnold tongue stability
→ mediant is the unique combining operation (Stern-Brocot theorem)
→ Stern-Brocot tree is the configuration space (D10-D11)
→ Klein bottle selects {q₂=2, q₃=3} (D19)
→ Farey count |F₆| = 13 (number theory)
→ SO(2) invariance → (|F_n|, n) are the only scalars (D28 Step 0)
→ Mediant-consistent partition: C/(C+S) (D28 Steps 2-4)
→ Ω_Λ = 13/19 (D25)

No axioms beyond “coupled oscillators conserve energy and lock to the most stable ratio.” The rest is mathematics.

Why the Stern-Brocot tree and not a continuum#

The continuum fails minimality#

A continuous frequency space ℝ satisfies betweenness: given any two reals, their average (or any convex combination) lies between them. But ℝ does NOT satisfy minimality: between any two reals there are uncountably many others, and no canonical “simplest” one exists. The arithmetic mean, geometric mean, and harmonic mean are all “between” but none is “simplest” because ℝ has no ordering by complexity.

Minimality requires a DISCRETE ordering by denominator: p/q is simpler than p’/q’ when q < q’. This ordering is not imposed — it is the stability ordering of the Arnold tongues. A mode with smaller q has tongue width w ~ (K/2)^q — exponentially wider, therefore exponentially more stable. The simplest fraction is the most physically robust one.

The Stern-Brocot tree IS this ordering. It enumerates all rationals by increasing denominator, with each mediant being the simplest rational in its interval. No continuum structure has this property.

Why not a different discretization#

The Arnold tongue structure at rational p/q follows from Fourier analysis of the coupling function. The Kuramoto coupling sin(θ_j − θ_i) has a single Fourier harmonic. Its iterates produce tongues at ALL rationals p/q, with width scaling as (K/2)^q — ordered by denominator.

Any periodic, antisymmetric coupling function can be expanded in a Fourier series. The fundamental mode (sin) dominates at weak coupling. The tongue structure at all rationals, ordered by denominator q, is universal — it does not depend on the specific coupling function, only on its periodicity and antisymmetry.

A different discretization (powers of 2, decimals, algebraic numbers) would not reproduce this tongue ordering because the tongues occur at ALL rationals, not at a subset. The Stern-Brocot tree is the unique enumeration that respects the tongue width ordering.

The universality argument#

The result rests on three properties of the coupling:

  1. Periodicity (phases are circular: θ ∈ S¹)

  2. Antisymmetry (coupling is mutual: f(θ) = −f(−θ))

  3. Smoothness (Fourier series converges)

Any coupling satisfying (1)-(3) produces Arnold tongues at all rationals, ordered by denominator. The Stern-Brocot tree is the configuration space of the resulting dynamics. No other structure emerges.

If the coupling were not periodic (e.g., linear springs), there would be no mode-locking and no rational structure. If it were not antisymmetric, the coupling would not be mutual and the system would not synchronize. If it were not smooth, the Fourier analysis would fail.

The three properties (1)-(3) are the defining properties of coupled oscillators on a circle. They are not choices — they are the definition of the physical system. The Stern-Brocot tree is their unique consequence.

Status#

Derived. The mediant is the unique operation satisfying betweenness (energy conservation) and minimality (widest Arnold tongue). The Stern-Brocot theorem (1858/1860) proves this. The framework’s primitive (2) is replaced by a physical property (2’) that is experimentally verifiable.

The Stern-Brocot tree is the unique configuration space arising from coupled oscillators with periodic, antisymmetric, smooth coupling. It is not a choice of discretization — it is the Arnold tongue structure, which is universal for this class of dynamical systems.

The prediction Ω_Λ = 13/19 now rests on:

  • Coupled oscillators conserve energy (testable)

  • Coupled oscillators lock to the most stable ratio (testable, demonstrated in every synchronization experiment since Huygens)

  • The Klein bottle topology (D18-D19, simulation-confirmed)

  • Number theory (the Farey count, Euler totient)

  • SO(2) invariance (the Kuramoto symmetry)

None of these are axioms in the sense of “assumed without justification.” They are physical properties and mathematical theorems.

Proof chains#

This derivation is the starting proposition (P2) in both end-to-end proof chains: