UNNS Substrate Research Program · Foundational Theory · 2026

The Margin-Confinement Law:
Why Coherent Structures Never Truly Fragment

Across supernovae, Voyager transport, nuclear events, seismic systems, and neutrino detector corpora, the same structural pattern repeatedly emerged: systems approached fragmentation asymptotically while preserving hidden relational continuity. The Margin-Confinement framework proposes that admissible structures do not truly cross realizability boundaries. Instead, apparent collapse may reflect projection-induced fragmentation, while latent coherence remains structurally recoverable under locality-preserving transforms.
Non-Crossability FCC Regime Δ-Lifting RISC Latent Continuity 15,401 Evaluations 0 Genuine Crossings
Instruments: STRUC-I v1.0.4 · STRUC-PERC-I v2.4.0–v2.5.0 Domains: 14 physical · 1 biological · 1 astrophysical · 1 detector reconstruction Status: Foundational theory manuscript · 2026

Abstract

Physical systems under extreme forcing — nuclear detonations, stellar explosions, heliopause crossings, particle detector reconstruction pipelines — appear to approach structural disintegration. And yet they do not cross into persistent Hard fragmentation. The Margin-Confinement Law explains why: the admissibility boundary ∂ℳadm is a dynamically non-penetrable invariant manifold for identity-preserving flows.

Derived from the Universal Structural Law, the Percolative Realizability Principle, and Bounded Structural Rigidity, the law proves non-crossability by contradiction via a background-chain structural invariant and a Lyapunov-type confinement functional. Empirical support spans 15,401 evaluations across sixteen domains with zero genuine crossings.

A subsequent Δ-lifting stage on the neutrino detector corpus recovers Full percolation in all nine previously fragmented ladders (100% recovery rate), expanding FCC-like states from 5 to 34 and establishing the first operational evidence for latent structural continuity — continuity not recreated but uncovered. The manuscript frames confinement and recoverability as two sides of a single deeper invariance.

🔒 The Margin-Confinement Law

Physical systems under extreme forcing approach the admissibility boundary ∂ℳadm. Seismic waveforms fragment. Detector outputs lose connectivity. Plasma boundaries reorganize. And yet: no admissible system in the tested corpus ever crossed into persistent Hard fragmentation. Not as a statistical tendency. As a structural fact.

Theorem 1 — The Margin-Confinement Law
Once a physical system is represented inside the admissible region adm of realizability space, its subsequent evolution under identity-preserving dynamics remains confined to adm for all time. The admissibility boundary ∂ℳadm, defined by connectivity margin m(Lt) = 0, is a dynamically non-penetrable invariant manifold: trajectories may approach it asymptotically — generating the Forced Coherent Collapse regime — but cannot cross into persistent Hard fragmentation without discontinuous rupture of structural identity.

The proof proceeds by contradiction. Assuming a crossing occurs in finite time, one derives a structural impossibility: the background chain — a connected 1-subcomplex of the vulnerability graph guaranteed by the USL — cannot dissolve continuously. Its dissolution is required by the crossing hypothesis. Therefore no crossing occurs. A Lyapunov-type confinement functional supplies the dynamical counterpart.

Two Pillars — Confinement and Recoverability

The law operates simultaneously at two levels: a topological level (the background chain invariant prevents discontinuous escape) and a dynamical level (the Lyapunov functional bounds margin from below). Together they yield two findings: forbidden collapse (Theorem 1 prohibits genuine Hard fragmentation under identity-preserving dynamics) and recoverable continuity (admissibility is an intrinsic local-relational property that re-emerges under locality-preserving transforms).

Conceptual diagram: upper panel shows Voyager, supernovae, and nuclear trajectories evolving within an admissible manifold toward the FCC regime. Lower panel compares observational RISC projections with Δ-space recovered paths.
Figure 1 — Trajectory Approaching the Admissibility Boundary: Confined Dynamics vs. Apparent Crossing. Identity-preserving trajectories evolve within admissible state space while approaching the realizability boundary asymptotically. Near-boundary dynamics produce Forced Coherent Collapse (FCC), where structural margin approaches zero without trans-boundary escape. The lower panel illustrates how observational fragmentation in raw charts (RISC) may falsely appear non-admissible, while Δ-space recovery reveals latent continuity.
Geometric visualization: colored trajectories spiral toward a glowing non-crossable boundary. Inset shows margin approaching zero from above while forbidden region remains inaccessible.
Figure 6 — Confinement via Non-Crossable Invariant Manifold (Geodesic Trapping). Identity-preserving trajectories evolve toward realizability boundaries while remaining confined within admissible state space. Near-boundary dynamics produce tangential sliding, spiraling convergence, and asymptotic trapping. The inset illustrates the asymptotic limit m(Lt) → 0⁺ — admissible systems may approach collapse arbitrarily closely without entering forbidden non-realizable states.
Video 1 — Forbidden Crossing and Asymptotic Confinement Near the Admissibility Boundary. Animated visualization of the Margin-Confinement principle. Structural trajectories evolve within admissible state space while approaching the realizability boundary asymptotically. As margin collapses toward m(Lt) → 0⁺, trajectories become compressed, tangent to the boundary, and dynamically trapped without crossing into forbidden non-admissible regions. Confinement as geometric asymptotic process, not reflective barrier.

⚡ The Forced Coherent Collapse Regime

When admissible systems approach ∂ℳadm under extreme forcing, they enter the Forced Coherent Collapse (FCC) regime: tail dominance TD → 1, connectivity margin m(Lt) → 0⁺, giant-component coherence GR ≥ 0.97 persisting. The system compresses toward the boundary asymptotically. FCC is not a failure mode — it is admissibility-protected near-boundary compression, the geometric consequence of non-crossability.

FCC threshold
TD≥0.985
GR = 1.000 maintained
Highest TD (deepL)
0.9985
neutrino background train
NK corpus
TD 0.997
GR ≥ 0.971 · zero HARD
Voyager 1
97.4%
Full across heliopause
Δ-FCC states
34
up from 5 raw (6.8×)

FCC Coherence Is Locally Relational — A New Finding

The Δ-lifting stage reveals that FCC-like states persist under amplitude stripping, drift removal, and coordinate flattening. The bkg2*C14 series (raw TD ≈ 0.25–0.49) reaches TDΔ ≈ 0.99 in Δ-space. FCC coherence is carried by the local gap structure of the realizability ladder, not by global amplitude values. FCC is a local-relational boundary phenomenon — and a substantially stronger theoretical finding than previously recognized.

Multi-panel FCC diagram. Upper: systems evolving toward FCC corridor. Lower: phase-space map of GR vs TD separating admissible, soft fragmented, hard fragmented, and FCC regions. Insets: nuclear and supernova trajectories.
Figure 3 — Forced Coherent Collapse (FCC) Phase Space. FCC states emerge when admissible systems undergo extreme forcing while preserving giant-component coherence. The phase-space diagram operationally distinguishes FCC from soft and hard fragmentation regimes and illustrates the trapping behavior observed in nuclear, seismic, and astrophysical corpora.
NK nuclear case study: four plots showing giant ratio approaching high values, connectivity margin positive, margin scaling decreasing toward zero with yield.
Figure 4 — NK Nuclear Case Study: Forcing vs. Yield. As calculated yield increases, the structural margin decreases toward zero while giant connectivity remains high — FCC-like asymptotic compression rather than hard fragmentation. Increasing yield drives FCC and confirms asymptotic non-crossability.

🔍 RISC — Observational vs. Ontological Collapse

The law's most practically important corollary is Representation-Induced Structural Collapse (RISC): a source system genuinely in ℳadm producing a HARD verdict through a representation that breaks local connectivity. The corpus motivates a distinction that sharpens the law considerably:

Observational Collapse

Representation-specific HARD fragmentation in a particular chart. Arises from RISC mechanisms: binning artifacts, orientation reversals, n-poverty, attenuation. Correctable by locality-preserving transforms. The source never left ℳadm. Exists in one chart; absent from locality-preserving transforms of the same source.

Ontological Collapse

Persistent loss of admissibility across all locality-preserving transforms — total destruction of giant-component structure no representation can recover.

Not observed in any tested corpus. Forbidden for identity-preserving flows by Theorem 1.

RISC's Principled Foundation Under the Law

RISC is not merely a classification of representational failures. Under Theorem 1, it is a precise account of apparent observational escape beyond ∂ℳadm while the underlying trajectory remains confined. The representation failed; the source did not. This gives RISC a rigorous theoretical foundation rather than a descriptive one.

⚛️ Neutrino Detector Corpus — A New Kind of Evidence

The neutrino detector observational corpus is the manuscript's theoretically richest single domain. 67 ladders across five distinct representation groups of the same underlying detector process — the most direct testbed for RISC in the UNNS program. The corpus completes the full manufactured-collapse / embedding-recovery / transform-recovery cycle within a single physical source.

STRUC-PERC-I pipeline and domain validation: Voyager 2 trajectory and neutrino tracker clusters shown in admissibility phase space.
Figure 2 — STRUC-PERC-I Pipeline & Domain Validation (Voyager 2 & Neutrino Trackers). The STRUC-PERC-I framework transforms heterogeneous physical datasets into unified structural ladder representations. Voyager transport trajectories and compact neutrino detector structures remain confined within admissible structural regions despite strong forcing.

Stage 1 — Raw Corpus: RISC as Operational Fact

Full percolation
50
74.6% of corpus
Giant component
8
11.9%
RISC pairs
6
TMVA/deepL matched
USL violations
0
across all 67

The Strongest Single RISC Case

TMVA h_SigSB_significance_2 produces HARD (GR = 0.985, 1 isolated node) through classifier binning geometry. The corresponding deepL h_SigSB_significance_2 returns FULL (GR = 1.000, 0 isolated). Same observable. Same physics. Different chart. Different class. This HARD → FULL crossing is the clearest single RISC demonstration in the UNNS program.

Stage 2 — Δ-Lifting: Latent Continuity Recovery

Δ-Full percolation
61
91.0% · up from 74.6%
Recovery rate
9/9
100% fragmented → Full
Δ-FCC states
34
up from 5 raw (6.8×)
Giant → Full
7/7
all promoted

100% RISC Recovery Under a Local, Non-Parametric, Deterministic Transform

Every one of the nine fragmented raw ladders — three HARD and six TAIL — recovers to GR = 1.000 in Δ-space. This includes all three HARD outcomes with identified artifact mechanisms: deepL_Graph;2 (n-poverty), hBkgeff (orientation reversal), TMVA_h_SigSB_2 (classifier binning). The gap structure of the source process retains intact admissibility geometry that classifier projections obscure.

The Most Profound Implication: Continuity Was Never Lost

Recovery language implies the system broke and later healed. But the 100% recovery rate under a local transform suggests something more profound: in every RISC case, admissible continuity was never lost — it was only hidden.

Δ-lifting does not restore continuity. It uncovers continuity that was present all along.

Three-panel illustration: fragmented observational chart (left), faint hidden admissible backbone (middle), Δ-space recovery restoring giant connected component (right).
Figure 5 — Hidden Continuity Beneath Fragmentation (Δ-Recovery of Latent Continuity). Raw fragmented charts (left) conceal an underlying admissible continuity (middle), which becomes visible after locality-preserving Δ-transformation (right). The restored giant component demonstrates that continuity is not recreated but uncovered — RISC is a projection artifact, not genuine ontological collapse.
Video 2 — Hidden Continuity Beneath Apparent Fragmentation (Δ-Recovery of Latent Structure). Animated demonstration of latent admissible continuity beneath fragmented observational charts. Locality-preserving Δ-transforms progressively uncover the hidden admissible backbone. The recovered giant component reveals that continuity was not recreated but uncovered — RISC is a projection artifact, not genuine ontological collapse.

The Complete Anti-Collapse Cycle

Three Stages — Same Source, Three Different Recoveries

(1) TMVA manufactures HARD — classifier binning geometry creates structural gaps.
(2) Deep-learning embedding removes it — deepL recovers Full for the same observable (Stage 1 RISC).
(3) Δ-lifting recovers Full from every fragment — regardless of artifact type (Stage 2, 9/9).

The source system never left ℳadm. The HARD verdict was a property of the observational chart, not the physics.

🌐 Representation-Covariant Admissibility

The Δ-lifting results motivate a theoretical extension beyond the law's proven content. Theorem 1 governs dynamical confinement under identity-preserving flows. The Stage 2 results suggest a complementary principle governing representational transforms.

Conjecture 1 — Representation-Covariant Admissibility (Corpus-Supported)
Let L be a realizability ladder generated by a physical process satisfying the IPF conditions, and let φ be a locality-preserving representational transform — a map preserving the local ordering and gap structure of L, such as ΔL = (|xi+1−xi|). Then:

L ∈ ℳadm ⟹ φ(L) ∈ ℳadm

The Δ-lifting stage of the neutrino corpus provides the first operational support: all nine fragmented ladders recover Full; no natural-representation admissible ladder fails except under identified n-poverty artifacts.

Transform Class Taxonomy

Transform ClassStructural EffectCorpus Example
Locality-preservingPreserve or recover admissibility; resolve RISC artifactsΔ-lifting → 9/9 Full recovery
Relational / embeddingMay recover latent connectivity; approach admissibilitydeepL embeddings → 21/23 Full
Projection / binningCan induce RISC; destroy local connectivityTMVA histograms → TAIL/HARD
Orientation-reversingInverts ladder chart; creates void at boundaryhBkgeff → HARD (RISC Type-II)
Identity-breakingMay genuinely destroy admissibility; violates IPFNon-invertible coarse-graining

📊 Empirical Support — 15,401 Evaluations, 0 Crossings

The law is supported across sixteen domains. In every case, zero genuine dynamical crossings of ∂ℳadm are observed. All HARD outcomes have identified representation artifact mechanisms consistent with RISC. At 99% confidence, the genuine crossing rate is bounded above by p < 3.0 × 10⁻⁴, consistent with the theoretical prediction of p = 0.

STRUC-I · 14 Domains

Physical Ladders

5,233 evaluations across atomic, molecular, nuclear, geoid, CMB, crystallographic, atmospheric, solar, cosmic web, biological systems.

0 violations · all domains
Phase Mapping

α–μ Constant Grid

9,826 phase-mapping evaluations at deformed physical constant values. Zero verdict changes. Zero commutators. Zero hard USL violations.

0 changes · 0 commutators
Voyager 1

Heliopause Crossing

3,500 |B| heliosheath windows. 97.4% Full percolation across a 7-year boundary approach and physical crossing.

97.4% Full · 0 HARD
NK Nuclear

Explosion Corpus

29 station-events. TD reaching 0.997. GR ≥ 0.971. Margin decreases toward zero. FCC confirmed as the near-boundary state.

TD 0.997 · 0 HARD
Neutrino Detector

Stage 1 + Stage 2

67 ladders × 2 stages. Stage 1: 6 RISC pairs, 5 FCC-like. Stage 2: 9/9 recovery, 34 FCC-like, first latent continuity corpus.

100% recovery · 0 USL violations
Corpus Coverage — 15,401 Evaluations · 0 Genuine Crossings Phase mapping 9,826 STRUC-I physical 5,233 Voyager 1 |B| 3,500 STRUC-PERC 81 NK nuclear 29 Neutrino (S1+S2) 67+67 0 genuine HARD crossings · all HARD outcomes = RISC artifacts

💡 The Minimal Structural Principle and Programmatic Synthesis

Across RISC, Δ-lifting, FCC persistence, Voyager boundary transport, and latent continuity extraction, the corpus results admit a concise synthesis:

Minimal Structural Principle (corpus-scoped)
Admissible structures tend to preserve or recover relational continuity under identity-preserving evolution and locality-preserving representation transforms.

This principle is not a derived theorem. It is an empirical generalization recording the consistent pattern: confinement under dynamics (Theorem 1), representational recovery under Δ-lifting (Conjecture 1), boundary-adjacent resilience under extreme forcing (FCC), and long-range transport coherence (Voyager) all instantiate the same underlying structural tendency. The principle unifies these phenomena without asserting universality beyond the tested evidence.

The Conceptual Transition

The UNNS program's subject matter shifts from what class does a system occupy? to how does a system move within the invariant admissible manifold? This is the transition from classification theory to realizability dynamics. Admissibility is no longer a static property to check — it is a dynamical invariance to understand geometrically.

Alignment with Beyond Fragmentation

The Margin-Confinement Law and its companion manuscript Beyond Fragmentation are the empirical and theoretical halves of the same boundary framework:

DimensionBeyond FragmentationMargin-Confinement Law
Central questionWhat do extreme systems do near realizability boundaries?Why do admissible systems fail to cross those boundaries?
FCCDiscovered as a descriptive regimeExplained as near-boundary asymptotic state under confinement
VoyagerEmpirical near-boundary observationFirst realization of a non-crossability trajectory
RISCClassification of representational failuresApparent escape while trajectory remains confined
m → 0⁺Observed without explanationDerivable consequence: m cannot reach zero under IPF

The Unified Boundary Theory

Together the two manuscripts suggest that physical collapse, in the sense relevant to realizability geometry, is asymptotic rather than trans-boundary: realizable structures may become arbitrarily stressed, tail-dominated, and near-critical while remaining structurally admissible.

Beyond Fragmentation assembled the evidence. The Margin-Confinement Law supplies the missing theoretical principle.

Resources & References

UNNS Substrate Research Program · The Margin-Confinement Law · 2026 · Instruments: STRUC-I v1.0.4, STRUC-PERC-I v2.4.0–v2.5.0 · 15,401 evaluations · 0 genuine HARD crossings · all HARD outcomes = RISC artifacts · Theorem territory · All data available for independent verification · unns.tech