Regime Classification

The Three Regimes of Collapse

Every system in GCD occupies exactly one of three regimes — Stable, Watch, or Collapse — determined by four frozen gates applied to the kernel invariants. The regime is derived, never asserted. Stability is rare: 87.5% of the Fisher manifold lies outside it.

Ruptura est fons constantiae — dissolution is the source of constancy.

§1 — Why Three Regimes?

The regime system translates continuous kernel invariants into discrete structural phases. Verdicts are derived from gates, never asserted.

The Partition of Fisher Space

The kernel K : [0,1]n × Δn → (F, ω, S, C, κ, IC) produces continuous invariants. Four frozen gates discretize this into three regimes, partitioning the entire 4-dimensional Fisher manifold. The gates are conjunctive for Stable — all four must be satisfied simultaneously — which is why stability is structurally rare.

12.5%
24.4%
63.1%
Stable Watch Collapse

Stable

Low drift, high fidelity, low entropy, low curvature. All four gates satisfied. The quiet zone — only 12.5% of the manifold qualifies.

Watch

Moderate drift or a Stable gate not met. The intermediate zone where most “interesting” dynamics live — curvature cost often dominates.

Collapse

High drift (ω ≥ 0.30). The dominant regime — 63.1% of the manifold. Not failure but the generative boundary that makes return meaningful.

Key structural insight: Collapse is not an error state. Ruptura est fons constantiae — dissolution is the source of constancy. The axiom (“only what returns is real”) requires a boundary to return from. Collapse provides that boundary. 63.1% of the manifold is generative territory.

§2 — The Four Gates

Four frozen thresholds — seam-derived, not prescribed — classify every kernel output into a regime. Stable is conjunctive: all four gates must be satisfied simultaneously.

Gate Symbol Stable Threshold What It Measures
Drift Gate ω < 0.038 How much signal is lost to collapse (1 − F). Primary regime variable.
Fidelity Gate F > 0.90 What fraction of coherence survives collapse. Arithmetic mean of channels.
Entropy Gate S < 0.15 Bernoulli field entropy — uncertainty of the collapse field.
Curvature Gate C < 0.14 Channel heterogeneity — coupling to uncontrolled degrees of freedom.

Classification Logic

COLLAPSE ω ≥ 0.30 (primary boundary)
STABLE ω < 0.038  AND  F > 0.90  AND  S < 0.15  AND  C < 0.14 (conjunctive)
WATCH everything else (default)

Stable is conjunctive because stability requires all invariants to be simultaneously clean. A single gate violation drops the system to Watch, even if the other three are perfect.

Why ω = 0.038?

3.8% drift is the empirical boundary where all four Stable gates can plausibly be met. Below this threshold F > 0.962, and S and C are typically sufficiently low. Above it, the chances of satisfying all four gates simultaneously drops sharply.

Why ω = 0.30?

At 30% drift, the system has lost nearly a third of its coherence. The drift cost closure Γ(ω) begins its steep cubic ascent. This is where return from collapse requires structural intervention, not incremental correction.

Why F > 0.90?

90% fidelity is the floor for stability. Below it, even with low drift (which is constrained by F + ω = 1), the system cannot maintain coherence across channels reliably.

Why S < 0.15 and C < 0.14?

These thresholds ensure that the collapse field is locally ordered (low entropy) and channels are approximately homogeneous (low curvature). The maximum Bernoulli field entropy is ln 2 ≈ 0.693, so 0.15 is about 22% of maximum — very constrained.

§3 — Regime Profiles

Each regime has a distinct signature across all six kernel invariants and a characteristic return dynamic.

Stable

12.5% of Fisher space

Kernel Signature

  • F > 0.90 — very high coherence survival
  • ω < 0.038 — negligible drift
  • S <0.15 — low entropy, collapse field is ordered
  • C < 0.14 — channels are nearly homogeneous
  • IC/F > 0.95 — minimal heterogeneity gap

Return Dynamics

  • Γ(ω) < 10−5 — drift cost negligible
  • τ_R* dominated by memory term Δκ
  • Budget runs surplus — system generates credit
  • Basin requires O(ε²) perturbation to escape

Typical: healthy atoms, well-calibrated instruments, equilibrium systems.

Watch

24.4% of Fisher space

Kernel Signature

  • F ≈ 0.70–0.90 — moderate coherence
  • ω ≈ 0.038–0.30 — measurable drift
  • S ≈ 0.15–0.40 — moderate entropy
  • C ≈ 0.14–0.50 — visible channel spread
  • IC/F ≈ 0.70–0.95 — moderate heterogeneity gap

Return Dynamics

  • Γ(ω) ≈ 0.1–1.0 — drift cost measurable
  • τ_R* dominated by curvature cost α·C
  • Budget near break-even — costly but feasible return
  • Γ and αC compete for dominance

Typical: composites, biological systems, financial portfolios, mid-scale physics.

Collapse

63.1% of Fisher space

Kernel Signature

  • F < 0.70 — majority of coherence lost
  • ω ≥ 0.30 — high or extreme drift
  • S > 0.40 — high uncertainty in collapse field
  • C > 0.50 — severe channel heterogeneity
  • IC/F < 0.70 — large heterogeneity gap, dead channels

Return Dynamics

  • Γ(ω) > 100 — drift cost overwhelms curvature
  • τ_R* dominated by drift cost Γ(ω)
  • Budget in deep deficit — structural intervention needed
  • Beyond ω_trap ≈ 0.682: trapping — no single-step escape

Typical: confinement (quarks → hadrons), phase boundaries, system failure, deep restructuring.

§4 — Critical Overlay

Critical is not a regime — it is a severity flag that can accompany any regime. It signals that composite integrity has collapsed.

CRITICAL: IC < 0.30

IC = exp(κ) < 0.30

What IC < 0.30 means

  • Integrity is dangerously low — the weighted geometric mean has collapsed
  • One or more channels are near-dead (ci → ε)
  • Geometric slaughter: even if F is healthy, IC crashes
  • The seam budget becomes very expensive

Overlay behavior

  • Can appear in any regime — even Watch with moderate ω
  • In the phase diagram, Critical is rendered as a violet tint
  • Takes precedence in the classification cascade
  • Signals structural fragility regardless of drift level
Example: A system with ω = 0.20 (Watch) but one dead channel can have IC = 0.08 — the system appears moderate by drift alone, but is structurally fragile. The Critical overlay catches what the drift gate misses. This is the power of multiplicative coherence: one dead channel kills the geometric mean.

§5 — Geometry of the Regime Manifold

The kernel lives on the Bernoulli manifold — the space of binary channel probabilities c ∈ [0,1]n. Regime boundaries are geometric structures in Fisher coordinates.

Fisher Coordinates

θ = arccos(√F)
  • The Bernoulli manifold is flat in Fisher coordinates: gF(θ) = 1
  • All structure comes from the embedding, not intrinsic curvature
  • F + ω = 1 becomes sin²θ + cos²θ = 1 — an exact trigonometric identity
  • Regime boundaries are arcs at fixed θ values

The Equator

c = ½, θ = π/4
  • S = ln 2 (maximum Bernoulli field entropy)
  • S + κ = 0 (perfect cancellation)
  • F = ω = ½ (equal coherence and drift)
  • Four independent quantities converge simultaneously
  • The equator sits in the Watch regime

Why Stability Is Rare

Stable is the intersection of four 1-dimensional constraints in 4D space. Each constraint carves away a slab of the manifold; their intersection is a tiny corner. The measure of this corner is ∼12.5% of Fisher space.

ω < 0.038
~96.2% passes
F > 0.90
~10% passes
S < 0.15
~21.6% passes
C < 0.14
~28% passes

Conjunction: ~0.962 × 0.10 × 0.216 × 0.28 ≈ 0.006 — the exact figure is 12.5% because the constraints are correlated (F + ω = 1 links the first two gates).

Special Points on the Manifold

c* = 0.7822

Logistic self-dual fixed point. Maximizes S + κ per channel. Sits in the Watch regime near the Stable boundary.

ctrap = 0.3177

Trapping threshold. Below this, Γ(ωtrap) = α and single-step escape is impossible. Deep Collapse territory.

c = ½ (equator)

Quintuple fixed point: S = ln 2, S + κ = 0, Φeq = 0. Exact center of Watch regime.

§6 — Regime Transitions

Regimes follow a monotonic ordering. Unlike classical phase transitions, regime changes in GCD are non-hysteretic — the regime is a function of invariants, not history.

Monotonic Progression

STABLE
ω < 0.038
WATCH
0.038 ≤ ω < 0.30
COLLAPSE
ω ≥ 0.30
POLE
ω → 1

Stable → Watch

Any single gate violation triggers. Most common: ω crosses 0.038 or C crosses 0.14. The system leaves the quiet zone — drift or heterogeneity has become measurable.

Watch → Collapse

ω crosses 0.30. The drift cost Γ(ω) begins its steep cubic ascent. The system has lost 30%+ of its coherence. Return now requires structural intervention.

Near the Pole

As ω → 1, Γ → ∞, information loss becomes total. Rmin diverges: Rmin·(1−ω) → 1/tolseam = 200. No curvature reduction can compensate.

Non-Hysteretic Transitions

GCD regime changes are non-hysteretic. The regime is a function of the current invariants, with no memory of the previous regime:

  • Forward: Stable → Watch → Collapse follows directly from gate evaluation
  • Reverse: Collapse → Watch → Stable requires ω strictly below each threshold
  • No bistability: same invariants always produce the same regime
  • Regime monotonicity (Lemma 22): tightening a threshold monotonically increases the Collapse fraction

§7 — Heterogeneity Gap per Regime

The heterogeneity gap Δ = F − IC measures how much channel heterogeneity costs integrity. It is the central diagnostic across regimes.

Δ = F − IC Arithmetic mean minus geometric mean of channels

Stable: Δ ≈ 0–0.05

All channels strong, approximately homogeneous. IC ≈ F. Negligible gap — the geometric and arithmetic means nearly coincide.

Watch: Δ ≈ 0.05–0.30

Some channels weaken, creating spread. IC drops faster than F. The gap reveals which channels need attention — the “interesting zone” for diagnostics.

Collapse: Δ ≈ 0.30–0.90

Severe heterogeneity. Dead channels (ci → ε) kill IC while F remains positive. This is geometric slaughter — one dead channel destroys multiplicative coherence.

Two-Channel Formula

Δ = (c1 + c2)/2 − √(c1 · c2) = (√c1 − √c2)² / 2

The gap is always non-negative (the integrity bound IC ≤ F). For n channels it satisfies the asymptotic bound Δ ≈ Var(c) / (2F) — the Fisher Information contribution from heterogeneity.

Geometric Slaughter (Lemma 47)

8-channel trace: 7 channels perfect (c = 1), 1 channel dead (c = ε). F stays healthy — the arithmetic mean barely notices one dead channel. But IC crashes to ε1/8 ≈ 0.10 — the geometric mean is destroyed by a single weak link.

§8 — Cross-Domain Evidence

The regime system is not domain-specific. The same gates, thresholds, and transitions appear across all 20 closure domains. Here are canonical examples.

Particle Physics: The Confinement Cliff

Quarks
Watch
IC/F ≈ 0.94
Hadrons
Collapse + Critical
IC/F ≈ 0.01–0.04

The color channel becomes dead at the confinement boundary (ccolor → ε). IC drops 98% while F barely changes. This is geometric slaughter at a phase boundary — the regime transition from Watch to Collapse+Critical is detected structurally by the kernel.

Atomic Physics: The Restoration of Coherence

Hadrons
Collapse
IC/F ≈ 0.04
Atoms (e.g., Nickel)
Stable
IC/F ≈ 0.96

Atoms restore coherence with new degrees of freedom (electron shells). The dead color channel is replaced by high-coherence electronic channels. IC/F recovers from 0.04 to 0.96 — a complete regime restoration from Collapse to Stable.

Cross-Domain Regime Patterns

Quantum Mechanics

Confined phase (Collapse) ↔ Quantum spin liquid (Watch) — detected via IC cliff at confinement boundary.

Materials Science

Ferromagnet (Stable) ↔ Paramagnetic (Watch) ↔ Spin glass (Collapse) — channel heterogeneity tracks magnetic disorder.

Neuroscience

Alert cortex (Watch) ↔ Lesion state (Collapse) ↔ Recovery (Watch) — cortical channels track neural coherence.

Finance

Stable portfolio (Stable) ↔ Stress (Watch) ↔ Crisis (Collapse) — market channels detect systemic coherence loss.

§9 — Live Regime Explorer

Enter channel values and see the kernel invariants, regime classification, and all four gate evaluations in real time.

Equal weights assumed. Enter 2–64 values.

Gate Evaluation

§10 — Interactive Phase Diagram

Two-channel phase space showing regime boundaries. Move the cursor over the c1 × c2 plane to explore kernel invariants at every point.

Regime Phase Diagram

Two-channel phase space showing regime boundaries. Move the cursor to explore kernel invariants at every point.

Stable 12.5% Watch 24.4% Collapse 63.1% Critical

Regime Partition — Fisher Space

Stable 12.5% Watch 24.4% Collapse 63.1%

Stability is rare — 87.5% of the manifold lies outside it. Return from collapse to stability is what the axiom measures.

§11 — Verification

Structural Identities

  • F + ω = 1 — exact to 0.0e+00 across 10K traces
  • IC ≤ F — integrity bound holds for all admissible inputs
  • IC = exp(κ) — log-integrity relation verified to machine precision
  • 44 structural identities re-derived computationally

Test Coverage

  • 14,362 tests across 172 test files
  • Regime classification tested across all 20 closure domains
  • 43 regime-classification theorems verified
  • Gate thresholds frozen per run—consistent across the seam

Key Lemmas

  • L19: Stable ⇒ Γ(ω) < 10−5
  • L22: Regime monotonicity — tightening preserves ordering
  • L39: Super-exponential IC convergence as channels approach 1
  • L47: Geometric slaughter at phase boundaries (8/20 domains)

Frozen Parameters

ε 10−8
p 3
α 1.0
tolseam 0.005
ωstable 0.038
ωcollapse 0.30