Every physical system has constants.
The speed of light. Boltzmann’s constant. The fine-structure constant. These numbers define how a system behaves — how fast energy moves, how heat distributes, how fields interact. Without them, the equations are empty. With them, the equations describe reality.
Most constants are measured. You build an instrument, run an experiment, and read a number off the result. The number is what it is. You do not derive it — you observe it.
A few constants are derived. They fall out of the mathematics itself, before any experiment is run. π is one. The golden ratio is another. You do not measure φ ≈ 1.618 — you prove it. It is the unique positive root of x² = x + 1, and its value is fixed by algebra, not observation.
The distinction matters. Measured constants might have been different. The speed of light could, in principle, have been another number. But π could not. The golden ratio could not. They are consequences of structure — and structure does not negotiate.
The Three-Body Problem of Propagation
In Edition 1, we introduced σ — the unique positive root of x⁴ = x + 1, approximately 1.22074. We showed that σ is the natural organizing constant of the A₄ root lattice: a four-dimensional geometric structure with 20 nearest neighbors, governed by the symmetry group of the regular pentatope.
In Edition 2, we proved that σ extends Van der Laan’s self-similar subdivision to four dimensions, closing the system at exactly 35 proportional types — the combinatorial number C(7,3) — answering a question that had been open for sixty years.
But both editions treated σ as a parameter — a number the geometry selects. The deeper question is whether the geometry has something stronger. Not just a constant, but a law.
Here is what we mean.
How Wavefronts Propagate on a Lattice
When a wavefront propagates on a lattice — any lattice, in any dimension — it does not move uniformly. At each lattice site, the propagation has choices: it can step along a shared facet, traverse a full simplex interior, or continue along a higher-dimensional boundary. The geometry determines which paths exist. The constants determine how the growth distributes among them.
On the A₄ lattice, there are exactly two characteristic topological delays:
Mode 1: Three-step boundary path. Propagation proceeds through a tetrahedral face — a three-dimensional boundary shared between two simplices. This is the short path, the local connection. In the dominant-delay model, this path introduces a characteristic delay of 3 steps, scaling by a factor of σ⁻³.
Mode 2: Four-step interior path. Propagation proceeds through the full simplex — traversing all four dimensions of the pentatope. This is the long path, the deep connection. In the dominant-delay model, this path introduces a characteristic delay of 4 steps, scaling by a factor of σ⁻⁴.
These correspond to the shortest topological path lengths through the codimension-1 boundaries and the full dimensional interior of the simplex. Every propagation path on the A₄ lattice factors through these two characteristic delay channels.
The Identity
Now compute the growth partition fractions.
σ ≈ 1.22074, so:
σ⁻³ ≈ 0.54970 σ⁻⁴ ≈ 0.45030
Add them:
σ⁻³ + σ⁻⁴ = 1
Not approximately 1. Not 0.99997. Not “within numerical precision.”
Exactly 1.
Why This Is Not a Coincidence
This identity is not an empirical observation. It is an algebraic consequence of the defining equation x⁴ = x + 1.
Here is the proof:
Start with σ⁴ = σ + 1. This is the definition of σ — the equation that creates it.
Divide both sides by σ⁴:
1 = σ/σ⁴ + 1/σ⁴
Simplify:
1 = σ⁻³ + σ⁻⁴
That is the entire proof. Three lines. The identity is not discovered by computation — it is inherited from the polynomial that defines the constant. It is as provable as the Pythagorean theorem, and it is true on every machine, in every implementation, forever.
What This Means for a Lattice
In physical systems, conservation laws ensure that quantities like energy are neither created nor destroyed — only transferred between modes. In a dynamical propagation network, the partition of growth behaves in an analogous way.
The identity σ⁻³ + σ⁻⁴ = 1is the growth-rate partition law of the A₄ lattice wavefront.
It says: when a wavefront propagates across the lattice, its asymptotic growth is distributed between its two characteristic delay channels in a way that is exactly conserved. The two growth fractions — the short boundary path and the long interior path — sum to unity. The asymptotic growth rate partitions between the channels in a ratio fixed by pure algebra.
This is not a design choice. Nobody tuned σ to make this work. Nobody selected the ratios to satisfy a constraint. The partition law is the constraint, and σ is the only number in the real line that satisfies it for dimension 4.
We call this the Semantic Octave — because it functions like an octave in music. In music, an octave is the interval where a frequency doubles and returns to “the same note.” In the A₄ lattice, the wavefront growth splits into two characteristic delay modes and partitions to unity. The geometry has a natural period, a fundamental cycle where the growth factors balance. The octave is that cycle.
The Family Pattern
The identity is not unique to dimension 4. It is universal.
In dimension 2, the golden ratio φ satisfies x² = x + 1, which gives:
φ⁻¹ + φ⁻² = 1
Two characteristic delays on the triangular lattice. One-step (edge) and two-step (face). The growth fractions sum to 1.
In dimension 3, the plastic constant ρ satisfies x³ = x + 1, which gives:
ρ⁻² + ρ⁻³ = 1
Two characteristic delays on the tetrahedral lattice. Two-step and three-step. The growth fractions sum to 1.
In dimension 4, the σ-constant satisfies x⁴ = x + 1, which gives:
σ⁻³ + σ⁻⁴ = 1
Two characteristic delays on the A₄ lattice. Three-step (tetrahedral face) and four-step (full simplex interior). The growth fractions sum to 1. This is the identity we derived above — the one Calera Computing’s architecture is built on.
In dimension 5, c₅ ≈ 1.16730 satisfies x⁵ = x + 1:
c₅⁻⁴ + c₅⁻⁵ = 1
In dimension 6, c₆ ≈ 1.13472 satisfies x⁶ = x + 1:
c₆⁻⁵ + c₆⁻⁶ = 1
In general, for any dimension d ≥ 2, the constant c_d satisfies c_dᵈ = c_d + 1, and the same three-line derivation yields its own partition law:
c_d⁻⁽ᵈ⁻¹⁾ + c_d⁻ᵈ = 1
The identity is inherited from the polynomial, not discovered by computation. It holds in every dimension, for every constant in the hierarchy, by algebra alone — confirmed computationally for d = 2 through 20 using the verification scripts published alongside our second paper. The general derivation is in the appendix below.
Every lattice in the A_d family has its own growth partition law. Every constant in the xᵈ = x + 1 hierarchy carries its own octave. The fractions change — φ splits roughly 62/38, ρ splits roughly 57/43, σ splits roughly 55/45,c₅ splits roughly 54/46, c₆ splits roughly 53/47 — but the sum is always exactly 1.
The deeper the dimension, the more equal the split.
A lattice with a strict growth partition law is a lattice where the relative structure of propagating information is asymptotically stable.
If the growth rate is perfectly partitioned between the delay channels, then a pattern propagating through the lattice preserves its relative mode distribution. The ratio between the boundary and interior components converges to a fixed value. It does not drift over time, and it is not subject to numerical decay. It can be decoded by factoring out the global growth rate, leaving the relative structure identical to when it was initiated. Not “approximately the same.” The same. Because the algebra guarantees the partition ratio.
This is the difference between a parameter and an identity.
A system built on tuned parameters is fragile. Change the parameter — adjust it by 0.1%, retrain a model, update a coefficient — and the system behaves differently. The old results are no longer reproducible. The new results depend on the new parameter. There is no mathematical guarantee that the system will behave the same way twice.
A system built on an algebraic identity is permanent. σ⁻³ + σ⁻⁴ = 1 does not change when you update software. It does not drift when you retrain. It does not vary between hardware platforms. It is an algebraic fact, derivable in three lines, and it will be true in a hundred years for the same reason it is true today: because x⁴ = x + 1 has exactly one positive real root, and that root has this property. And if the architecture ever moves to a different dimension — a 5D lattice, a 6D lattice, any lattice in the family — the partition law moves with it. The octave is not a feature of one geometry. It is a feature of the entire hierarchy.
Architecture we build at Calera Computing is grounded in this identity. Not as an aspiration. As a mathematical constraint grounded in the geometry of the A₄ lattice and the universal hierarchy it belongs to.
What Comes Next
Edition 1 introduced the constant. Edition 2 proved the counting formula. Edition 3 derives the growth partition law.
Edition 4 will go further: what happens when the A₄ lattice tiles aperiodically — when it fills space without repeating? The connection to quasicrystals and Penrose tilings is not a metaphor. It is a geometric consequence of the same root system.
Casey Lee Race
Founder King
📄 Paper: “The σ-Constant: A Universal Algebraic Invariant for Energy Propagation in d-Dimensional Simplicial Lattices” 🔗 DOI: doi.org/10.5281/zenodo.20350425 💻 Verification Scripts: github.com/Calera-Computing-Inc/sigma-constant-verification
📄 Paper 2: “The x^d = x + 1 Hierarchy: Cross-Dimensional Spectral Validation on A_d Root Lattices” 🔗 DOI: doi.org/10.5281/zenodo.20692936 💻 B02 Verification: github.com/Calera-Computing-Inc/cross-dimensional-hierarchy-verification
Appendix: The General Identity
For any dimension d ≥ 2, let c_d be the unique positive real root of xᵈ = x + 1. Then:
c_dᵈ = c_d + 1 — the defining equation.
1 = c_d / c_dᵈ + 1 / c_dᵈ — divide both sides by c_dᵈ.
1 = c_d⁻⁽ᵈ⁻¹⁾ + c_d⁻ᵈ — simplify.
The two growth fractions partition unity in every dimension. The first members of the family are φ (d = 2), ρ (d = 3), and σ (d = 4). The identity was verified computationally for d = 2 through 20.