In the real numbers ℝ, every nonzero element has a multiplicative inverse. In the complex numbers ℂ, the same. In quaternions ℍ, still true. But in octonions 𝕆, division persists — yet its structure is fragile. Beyond octonions, in any algebra built by the same recursive doubling process, there is no division at all.
This is the content of Frobenius' theorem (1877): the only finite-dimensional associative division algebras over ℝ are ℝ, ℂ, and ℍ. The Cayley-Dickson ladder stops. There is no associative division algebra beyond quaternions.
But 𝕆 itself is not associative. It is a non-associative division algebra, and it is also the end of that line. What lies beyond?
Baker's theorem (1966) provides a transcendental connection: a logarithmic linear form in algebraic numbers cannot be too small unless it vanishes. More precisely, if α₁, …, αₙ are nonzero algebraic numbers and β₁, …, βₙ are algebraic numbers with β₁ log α₁ + ⋯ + βₙ log αₙ ≠ 0, then this sum is bounded below by an explicitly computable function of the heights and degrees of the αᵢ and βⱼ.
The key insight for us: the logarithms of primes are algebraically independent. That is, if p₁, p₂, …, pₖ are distinct primes, then log p₁, log p₂, …, log pₖ are linearly independent over ℚ.
This linear independence is the keystone. It means that the formal ℚ-vector space spanned by {log p} has uncountable dimension — far richer than any finite-dimensional algebra.
In a Lie algebra, the root system consists of vectors α in a Euclidean space such that the reflection through the hyperplane perpendicular to α maps the root system to itself. The simplest example: the roots of SU(2) are ±2e, where e is a unit vector.
For the exceptional algebra E₈, the root system Φ(E₈) has 240 roots living in an 8-dimensional space. The roots fall into orbits under the Weyl group; the lengths of roots are (typically) either √2 or 2.
Now consider: what if we label each root by the logarithm of a prime? Not literally — not by replacing root vectors with numbers — but by decorating the root diagram. Each prime p gets assigned to a root α_p, and the "length" of that root in the index space is understood as log p.
This is not a Lie algebra structure on {log p}. Rather, it is an embedding of the arithmetic information (the primes, their logarithms, their linear independence) into the combinatorial skeleton of the root diagram of E₈.
In 𝕆, every nonzero element is invertible. The multiplication is nonassociative but has the property that the *real* parts satisfy a composition law: if u and v are octonions, then Re(u · v) depends linearly on Re(u) and Re(v).
But the moment we move to an algebra structure that incorporates the full independence of {log p}, we are no longer in a finite-dimensional setting. The Cayley-Dickson ladder is finite; the log-prime lattice is infinite-dimensional.
In infinite-dimensional settings, division fails. There are elements with no left or right inverse, not for deep reasons but because the dimension exceeds what finite-dimensional invertibility can handle. The bridge from 𝕆 to E₈ crosses from the finite-dimensional non-associative division algebras into a regime where division is no longer a meaningful operation.
The k-nacci numbers (Fibonacci, Tribonacci, etc.) satisfy recurrence relations of increasing depth. Their characteristic polynomials have roots that grow toward 2 as k → ∞. The limit is the transcendental constant τ = 2 itself.
Each k-nacci sequence encodes a rhythm — a weighted path through the dimensions 1, 2, 3, …, k. The Cayley-Dickson ladder also encodes a rhythm: each doubling increases dimension and breaks one algebraic property.
The k-nacci spine connects these two rhythms. The Logs Segment is the bridge at the point where k-nacci beats merge with the spectral structure of E₈, and where primes (via their logarithms) become the numerology of that merge.
Several threads remain unfastened in the current understanding:
1. Explicit Embedding: Can we construct an explicit map from the k-nacci lattice into Φ(E₈) such that prime logarithms serve as decorations on the root diagram in a way that reflects the structure of the Cayley-Dickson ladder?
2. Division Breakdown: At what point along the bridge does invertibility fail? Is there a natural "proto-division" algebra structure on {log p} (possibly via formal groups or Lie algebras) that makes the transition smooth?
3. Baker Constants: Do the lower bounds from Baker's theorem (the effective bounds on linear forms in logarithms) appear naturally in the geometry of E₈, or in the stability constants ε₀ and κ of the dm³ framework?
Baker, A. (1966). "Linear forms in the logarithms of algebraic numbers." Mathematika, 13(2), 204–216.
Lindemann, F. (1882). "Über die Ludolph'sche Zahl." Sitzungsberichte der Königlich Preussischen Akademie der Wissenschaften, 679–682.
Related deposit: k-nacci Spine (Zenodo)