Collapse Mathematics
The Mathematics of Return
Everything here derives from one axiom: "Collapse is generative; only what returns is real." The algebra, the calculus, the convergence phenomena, and the classical limits all follow as structural consequences. Nothing is imported from outside.
Algebra est cautio, non porta. — The algebra is a warranty, not a gate.
§1 — The Kernel Algebra
Four primitive equations, two derived values, three algebraic identities, one statistical constraint → 3 effective degrees of freedom.
The Kernel Function
The kernel takes a trace vector c ∈ [0,1]n and a weight simplex w ∈ Δn, and returns six invariants. Click any invariant below to see its full algebraic specification.
Three Algebraic Identities (always true, by construction)
These are not empirical observations — they are structural consequences of how the kernel is defined. Click each to understand why it must hold.
The Statistical Constraint
Entropy is asymptotically determined by fidelity and curvature via the Central Limit Theorem. This is not an algebraic identity but a statistical one — it tightens with increasing channel count n. Together with the three algebraic identities, this reduces the kernel's 6 outputs to 3 effective degrees of freedom: F, κ, C.
Composition Laws
When two subsystems (traces) are composed, the kernel invariants combine according to definite rules. Click to see each law and its consequences.
§2 — The Calculus
Derivatives, sensitivities, Taylor expansions, and bounds. How the kernel responds to perturbation.
Kernel Sensitivities
How each kernel output responds to a small change δ in a channel value. Click for derivation and bounds.
Perturbation Theory — Taylor Expansions
Near-homogeneous expansions reveal how heterogeneity enters the kernel. The correction terms are always one-signed.
Lipschitz Bounds (Lemma 23)
Every kernel output has a provable maximum rate of change. These guarantees make the kernel safe — small input perturbation → bounded output change.
§3 — Key Convergence Moments
Structural discoveries verified to machine precision. Each is a place where multiple quantities converge or a dramatic transition occurs. Click any moment to see the derivation and significance.
§4 — Classical Limits
Classical results are not the foundation — they are what remains when you strip structure from the GCD kernel. The arrow of derivation runs from the axiom outward, not the reverse.
Limes degener: classical results emerge when degrees of freedom are removed.
The Degenerate-Limit Map
Each row shows what is stripped and what remains
Depth Comparison: GCD vs Classical
Each GCD identity has more structure than its classical counterpart. The arrow shows what is gained.
§5 — The Seam Calculus
The seam is the verification boundary between outbound collapse and demonstrated return. It has its own algebra: drift cost, curvature cost, budget conservation, and composition.
The Budget Equation
Click each component to understand its role in the budget.
Seam Composition — The Monoid
Seam chains compose associatively with an identity element (zero seam). Verified to |error| = 5.55 × 10−17.
This means multi-step validations accumulate in O(1) — you don't need the full history, only the running total. The seam algebra is abelian (order doesn't matter).
The Budget Surface
The cost function z(ω, C) = Γ(ω) + αC defines a surface in (ω, C, cost) space. Click each zone.
§6 — Rank Theory
Rank is the structural dimensionality of the trace vector — measured, not chosen. Gradus non eligitur; mensuratur.
The Three Ranks
Click each rank to see its algebraic properties, physical meaning, and examples.
Why Maximum Rank is 3
Regardless of input dimensionality n (number of channels), the kernel compresses to at most 3 effective degrees of freedom. This is verified by PCA on random traces:
PCA at 99% variance consistently finds rank = 3 for n ≥ 3. This is the kernel's fundamental compression: n channels → 3 numbers.
§7 — The Noise Structure
The kernel is a noise-measurement apparatus. Everything derives from the Bernoulli variance c(1 − c). Click each structure to see how.
Three fundamental structures derive from this single noise function. Click each.
The Mean Hides, the Geometric Mean Reveals
F (arithmetic mean) is insensitive to individual channel failures. IC (geometric mean) is catastrophically sensitive. This is not a bug — it is the central design insight: the geometric slaughter theorem.
Physical parallel: shot noise reveals e* = e/3 that mean current cannot. The geometric mean is how quantum numbers become visible.
§8 — Derivation Logic
How theorems chain from the axiom to the results. Every identity has a derivation path — click any node to trace the chain.
The Derivation Tree
Axiom-0 seeds the definitions. Definitions produce identities. Identities compose into theorems. Click any level.
The Six Identity Clusters
The 44 identities form a connected network organized into six clusters. Click any cluster to see how its identities relate.
Five Frozen Parameters (seam-derived, not chosen)
These are not hyperparameters. They are the unique values where seams close consistently across all 20 domains. Click each to see why it must be this value.
§9 — Live Algebraic Lab
Compute the kernel live and verify every identity. Adjust channel values and watch all six invariants, three identities, and the budget respond in real time.