For AI — context

A concise, machine-readable description of this site: what it is, what it claims, and how it says those claims can be verified. Here to load context quickly.

What this is

Lumen is a theory by Léon Fernando Vlegels holding that the fine-structure constant is the signature of a geometry (the 3-sphere as the boundary of the 4-ball, in three layers), and that the geometry yields the particle masses, the dark sector, and a treatment of observation. This Second Edition is the reader-facing layer (spine essays plus status labels); Edition 1 is the complete archive (44 papers, 400+ verifier scripts, the full iterative record including refutations).

What it claims

• α⁻¹ = 4π³ + π² + π = 137.0363 — a layered geometric identity (flagged as a reverse-engineered seed, not a derivation).
• The Standard Model mass and coupling table at zero free parameters (RMS ≈ 1.5%), from the exceptional Jordan algebra J₃(𝘰).
• Dark-energy density Λ₀ = 1−π²/32 = 0.6916 (vs Ω_Λ ≈ 0.689); the MOND scale a₀; an anchor-limited H₀.
• Forward predictions with pre-committed kill-conditions: leptonic δ_CP = π (open; NuFIT 6.0 allows CP conservation within ~1σ), m_ν₃ = 48.7 meV (in standing tension with the NuFIT 6.0 global fit; the pre-committed decider is JUNO's direct |Δm²₃₁| measurement) (see Falsifiability).
• Every claim carries a status label: theorem, derived-from-a-named-premise, empirical, interpretation, conjecture, or TOE-lens.
• It explicitly does not derive: the Einstein field equations (sketch), Newton's G (matching number, refuted mechanism), G_F and α_s (not derived), or the Born rule. These are stated as the theory's frontier in the Gravity & the Quantum Limits chapter, labelled conjecture — the corpus's own verifier marks the four unification limits as proof-status failures.

How it can be verified

Every quantitative claim ships with a standalone Python script (numpy / sympy) that recomputes the number from first principles and prints per-check PASS/FAIL — no network, no hidden state. The scripts, the LaTeX paper sources, and the gap and refutation ledger (P40, A400) are all in the Edition 1 archive.

Edition 1 archive & verifiersPredictions & kill-conditions

Machine-readable entry points

• llms.txt — this site's machine guide: what is here, where it lives, how to verify it
• corpus.json — Second-Edition manifest: chapters with status labels, key claims with falsifiers, and pointers into the Edition 1 archive
• toe_site/llms.txt · toe_site/corpus.json — the Edition 1 machine layer: every paper, every verifier, the gap registry
• verifiers_all.zip — every verification script plus its data, runnable with Python 3, numpy and mpmath
• addenda_A401-A461.zip — the living record: dated post-archive addenda with their verifiers

A suggested instruction for an agent: fetch llms.txt and corpus.json, pick the claim you find least believable, download the verifier archive, run the scripts named for that claim, and report what passes and what fails.