UNNS Research Collective · February 2026 · Chambers LI–LV Structural Arc

Structural Completeness, Robustness,
and Hierarchical Non-Isometry

Five chambers close a complete geometric arc — from dimensional constraint through curvature sensitivity to metric stratification — establishing a structural theory of recursive operator admissibility that is internally falsifiable, categorically precise, and computationally predictive.
Status: CERTIFIED  ·  Vertical Arc: CLOSED  ·  Decision B: CONFIRMED
LI → LII → LIII → LIV → LV · Closed Progression

Executive Summary

Five computational chambers — each a falsification instrument — answer a single question: What is the geometry of admissibility under recursive operator dynamics?

The answer arrived in stages. Each chamber had explicit falsifiers that could have terminated the arc. None did. Instead each result forced the next question, and the questions form a chain that closes on a structural theorem.

The chain: Interaction geometry is dimensionally bounded. Curvature shifts admissibility boundaries without breaking partition identity. No hidden mechanism classes exist. Factorization survives adversarial perturbation. Under operator tier change, factorization survives — but metric geometry stratifies.

"Recursive admissibility factorizes into orthogonal constraint channels whose partition structure is invariant under perturbation and tier lift, while their metric geometry stratifies under operator bias according to a predictable curvature functional." — The Distilled Result

Open in Fullscreen!

The Structural Arc: Five Chambers, One Progression

The progression is not accidental. LI establishes geometric constraint. LII establishes curvature sensitivity. LIII–LIV establish partition invariance. LV reveals that invariance is structural but metric geometry is not. Non-isometry is the natural culmination of pre-metric geometric asymmetry elevated to a metric theorem.

LI n≤2 Cap Dimensional Constraint Interaction geometry ≤2D LII P3 Phase Curvature Deformation Boundaries shift, partitions hold LIII No Gate 5 Channel Completeness 4-gate basis is complete LIV R2=R3=0 Factorization Robustness Invariance survives stress LV Non-Iso Hierarchical Stratification Partitions persist, metrics stratify
CHAMBER LI

Dimensional Interaction Constraint

Pairwise admissibility gates produce non-additive interaction geometry (Δ = 0.580, κ = 0.309). Triple compositions do not sustain independent interaction structure — exclusion geometry vanishes at n = 3.

n ≤ 2 CONSTRAINT CONFIRMED
CHAMBER LII / PHASE P3

Curvature-Responsive Bifurcation

Curvature deformation shifts admissibility boundary location. Identity partitions remain stable throughout. Geometry is not rigid — it is structurally anchored. This is pre-metric symmetry.

BOUNDARY MOTION CERTIFIED
CHAMBER LIII

Channel Completeness

Adversarial relaxation across 56,877 mechanisms reveals no hidden mechanism classes. All residual structure collapses into a single unified basin at the G3 bifurcation boundary. No fifth gate exists.

4-GATE BASIS CLOSED
CHAMBER LIV v1.1.0

Factorization Robustness

Four perturbation families, identity-preserving tracking. R2 = R3 = 0 throughout all perturbation types and strengths. Factorization is structural, not a tuning artifact.

R2 = R3 = 0.0000 CERTIFIED
CHAMBER LV v1.1.0

Tier-2 Selective Transfer

Channels commute under lift (R2 = R3 = 0 throughout). Retention geometry does not. V2×V4 exhibits systematic erosion 2–3× steeper than stable families. Normalization null rules out encoding artifacts.

HIERARCHICAL NON-ISOMETRY

Chamber LIV v1.1.0 — Witness-Separability Robustness

The question Chamber LIV asked is simple and important: Is the factorization structure we discovered a fundamental property, or was it an artifact of the specific domain we tested it in? The answer required putting the structure under systematic adversarial stress.

Protocol: Four Perturbation Families

Chamber LIV applied four distinct identity-preserving perturbation families to the validated Tier-1 witness pool (N = 2000 mechanisms per family). "Identity-preserving" means the same mechanism IDs are tracked throughout — changes in witness status reflect genuine structural change, not sampling noise.

P1 Encoding Perturbation Most aggressive. G3 R1 drops below 0.70 — boundary thinning detected P2 Parametric Noise Continuous scalar perturbations across gate input space. R1 ≥ 0.70 maintained P3 Adversarial Shifts Targeted attacks on witness boundary regions. Channels orthogonal P4 Domain Stress Boundary extension and sampling stress across mechanism All metrics pass

The Four Robustness Metrics

R1 — Identity Retention
≥ 0.70
Fraction of baseline witness IDs surviving perturbation. G3/P1 edge case: boundary thinning, not collapse.
R2 — Channel Overlap
0.0000
Cross-channel witness contamination. Exactly zero across all perturbations and strengths.
R3 — Dual-Unlock Interference
0.0000
Mechanisms requiring two simultaneous gate relaxations. None found. Channels remain independent.
R4 — Witness Migration
Controlled
G3/P1 shows migration toward boundary-adjacent territory without crossing into another channel.

Theorem 3.1 — Factorization Robustness (Certified)

Under all four perturbation families within the characterized domain, witness-separability and channel orthogonality are preserved. R2 = R3 = 0 under all perturbation types at all tested strengths. G3's R1 drop under P1 (encoding perturbation) is classified as parametric sensitivity — boundary thinning — not structural failure. The distinction matters: structural stability (R2, R3, R4 jointly) is separate from parametric robustness (R1).

Structural vs. Parametric: A Key Distinction

Chamber LIV introduced a separation that carries forward into all subsequent analysis. A failure of R1 alone, when R2–R4 pass, indicates boundary thinness — the witnesses for that gate sit close to the admissibility threshold, so perturbations can displace them. It does not indicate channel collapse — the gate's conceptual independence is intact. This separation proved critical for interpreting Chamber LV.

Chamber LV v1.1.0 — Tier-2 Selective Transfer Diagnostic

The question was no longer whether factorization survives perturbation within a tier. The question became: what happens when the operator tier itself changes? Does the factorization structure transfer, or does tier-change break it?

The answer was unexpected in its precision: factorization transfers completely — but geometry does not transfer uniformly.

Three Operator Families Under Test

0.70 threshold 1.00 0.88 0.75 0.55 0.40 τ=0.1 τ=0.3 τ=0.5 τ=0.7 τ=0.9 F1 V2×V3 (Isometric) F3 V6×V7 (Mild erosion) F2 V2×V4 (Non-isometric — stratification) G2 Retention Trajectory (ρ₂) across Tier-2 Lift Levels

Three Independent Structural Signals

Signal I — Channel Orthogonality Invariance (R2, R3)

Across all three families, all five τ₂ levels, and both evaluation tracks: R2 = 0.0000 and R3 = 0.0000 throughout. Not approximately zero. Exactly zero. Channel orthogonality is not a Tier-1 artifact. Witness-separability and factorization survive operator tier change completely.

Signal II — Retention Erosion: Monotone and Family-Specific

FamilyGateτ=0.1τ=0.3τ=0.5τ=0.7τ=0.9R5-T SlopeClass
V2×V3G11.0001.0001.0001.0001.000+0.0000Stable
V2×V3G21.0000.9940.9440.9260.889−0.0290Stable
V2×V3G31.0001.0001.0001.0001.000+0.0000Stable
V2×V3G41.0001.0001.0001.0001.000+0.0000Stable
V2×V4G10.9750.9240.8470.8560.763−0.0492Mild erosion
V2×V4G20.8830.7220.5860.3770.340−0.1432Stratification
V2×V4G30.9230.7720.6910.5990.448−0.1122Stratification
V2×V4G40.9670.8740.7420.6420.517−0.1132Stratification
V6×V7G10.9830.8780.8430.8260.748−0.0522Mild erosion
V6×V7G20.9800.9320.8160.8030.769−0.0551Mild erosion
V6×V7G30.9790.9510.8880.8280.825−0.0432Stable
V6×V7G40.9650.9080.8440.6740.645−0.0872Mild erosion

Signal III — The Normalization Null (Decision Criterion B1)

The critical falsification test. If V2×V4's retention erosion were caused by encoding artifacts — curvature-induced scale drift in the gate input distributions — then encoding-invariant normalization should restore retention. Three normalization maps were applied: dk_rel (G2), B_rel (G3), L_rel (G4).

The Prediction That Failed

The encoding artifact hypothesis predicted: normalization would restore V2×V4 retention to passing levels. The prediction failed at every gate, every τ₂ level, to six decimal places. Δ(passRate) = 0.0000000 everywhere. Not approximately zero. Structurally zero — because the normalization maps rescale thresholds proportionally with population statistics, absorbing global shifts but leaving witness-specific boundary erosion untouched. This is a structural null, not a null of insufficient statistical power.

FamilyGateRaw passRateNorm passRateΔInterpretation
V2×V3G1–G41.0001.0000.000Control
V2×V4G20.4000.4000.000Normalization insufficient → B
V2×V4G30.4000.4000.000Normalization insufficient → B
V2×V4G40.6000.6000.000Normalization insufficient → B
V6×V7G1–G41.0001.0000.000Control
V6×V7G40.6000.6000.000Control (mild erosion)

Staged Degradation: A Geometric Signature

The migration metric (R4) reveals a particularly important ordering. Under V2×V4, the degradation proceeds in distinct stages — and the ordering of those stages carries structural information.

PROJECTION STRESS R1 declining · R2=R3=R4=0 PROJECTION FAILURE R1 below threshold · R4 approaching BREAKDOWN R4 > 0.10 · R2=R3=0 τ=0.1 R4=0.006 τ=0.3 R4=0.025 τ=0.5 R4=0.031 τ=0.7 R4=0.099 τ=0.9 R4=0.1235 ⚠ R2 = R3 = 0.0000 throughout all stages — channel identity preserved

The ordering is the discovery: R1 fails first. R4 crosses threshold later. R2 and R3 never move from zero. This means witnesses lose single-gate identity before the partition structure breaks down, and migration follows erosion rather than preceding it. This is the signature of geometric boundary thinning, not channel coupling.

Hierarchical Stratification: Three Geometric Strata

OPERATOR FAMILY GEOMETRIC STRATA ISOMETRIC STRATUM F1 · V2×V3 · |slope| ≤ 0.029 for all gates MILD EROSION STRATUM F3 · V6×V7 · |slope| ≤ 0.087 for all gates NON-ISOMETRIC STRATUM F2 · V2×V4 · |slope| ≥ 0.113 for G2, G3, G4 Curvature-augmented lift · Encoding artifact hypothesis FALSIFIED d(i,i)≈0 d(i,i)<0.3 d(i,i)>0.5

Decision B — Confirmed: Both Criteria Triggered

  • B1 (Normalization Null): V2×V4 fails after normalization. Δ(passRate) = 0.000000 across all gates. Encoding hypothesis falsified.
  • B2 (Slope Stratification): V2×V4 normalized slopes (−0.11 to −0.14) are 2–3× steeper than stable families. Structural stratification confirmed.

Conclusion: Factorization survives operator tier change. Admissibility geometry does not lift uniformly. That is a stronger statement than universal transfer.

Factorization Inevitability: From Discovery to Necessity

The paper "Factorization Inevitability in Recursive DAG Admissibility" answers the deeper question: why must admissibility factorize? The shift from "we found four gates" to "factorization is inevitable" is the decisive compression moment.

Before
Taxonomy
Four gates found through chamber validation
After
Theory
Factorization derived from recursive DAG structure
Mechanisms Tested
56,877
Adversarial LIII completeness testing
Hidden Gates Found
0
Basin unification at G3 boundary

Theorem 1 — Factorization Inevitability

If a gate set is witness-separable, admissibility irreducibly factorizes into k independent constraint channels.

  • Irreducibility: No gate is implied by the conjunction of the remaining gates. Each has a witness violating only it.
  • Channel-specific unlock: Each gate's witnesses become viable only when that specific gate is relaxed — not when any other is relaxed.
  • Orthogonal failure modes: The failure channels are pairwise distinguishable by witnesses. No monolithic constraint can reproduce this structure — channel information is structurally irreducible.

Theorem 2 — The DAG Correspondence

Any genuinely independent new gate discovered in an extended domain must correspond either to refinement of existing DAG-level invariant classes, or to extension of the substrate model. In neither case does it falsify the factorization principle. The theory is refineable, not brittle.

The Four Invariant Channels

GateNameDAG InvariantPredicate
G1Geometric CurvaturePath geometrycurvature_method = turning_angle
G2Baseline SeparabilityNode distinguishabilityΔκ > 0.05
G3Bifurcation CapabilityBranch structurebifurcation ∧ sharpness > 0.10
G4Locality ConsistencyCausal orderinglocality_score < 0.10

Why This Matters

Before: "We found four gates through chamber validation." After: "Admissibility must factorize; current domain reveals four channels corresponding to four DAG invariant classes." The four are not a list. They are a theorem. The 72% contraction of mechanism-class space is now understood as the joint restriction of four independent projection channels — each eliminating a distinct mechanism family.

The Unified Structural Theory

The paper "Structural Completeness, Robustness, and Hierarchical Non-Isometry" consolidates five chambers into a coherent theory with seven formal results.

1. Dimensional Interaction Constraint (LI)

Admissibility interaction geometry exists in n ≤ 2 compositions and collapses at n = 3. Interaction geometry is inherently low-dimensional.

2. Curvature-Responsive Bifurcation (LII)

Curvature deformation alters admissibility boundary position without altering partition identity. Geometry moves. Partitions hold. Pre-metric symmetry.

3. Local Structural Completeness (LIII)

No fifth independent gate exists within the characterized domain. The four-gate basis is closed and complete.

4. Factorization Robustness (LIV)

R2 = R3 = 0 under all four perturbation families at all tested strengths. Partition invariance is structural.

5. Projection Non-Isometry (LV)

Admissibility partitions are preserved under Tier-2 lift (R2 = R3 = 0), but retention geometry ρᵢ(τ₂) is not uniformly metric-preserving across operator families.

6. Critical Erosion Law

Migration onset satisfies τ_mig ≥ τc + (θ/A(F))^{1/p(F)}. Staged degradation has a predictable onset derivable from kernel parameters.

7. Categorical Formulation — The Cleanest Statement

The Tier-2 lift functor L is the identity on channel objects — empirically: R2 = R3 = 0 — but fails to be an isometry in the Lawvere-metric enriched category ChanF — empirically: self-distances d(i,i) = 1 − ρᵢ change as witnesses are lost.

Channel objects commute under lift. Metric geometry does not. That is the categorical core of hierarchical non-isometry. This is not collapse. This is stratification.

The Curvature Functional: Prediction Before Execution

The non-isometry constant LF — governing how steeply retention erodes — can now be predicted before running the chamber, from kernel bias parameters:

Kernel Bias Amplitude
B_F
From operator specification: γ_F · E[|ξ_F|]
Boundary Density
λᵢ
Estimated from baseline witness pool histogram
Non-Isometry Bound
L_F ≤ C(F)
Curvature functional predicts migration onset before chamber execution

What Has Been Gained: A Structural Theory of Operator Stratification

The Vertical Progression

This is not incremental work. Five chambers answer five structurally ordered questions that form a complete arc:

ChamberQuestionResultSignificance
LIDoes interaction geometry extend freely?No. 2D-bounded.Dimensional cap law
LIICan curvature deform without breaking identity?Yes. Geometry moves, partitions remain.Deformation-without-collapse principle
LIIIAre there hidden admissibility channels?No. 4-gate basis complete.Completeness theorem
LIVIs factorization robust under perturbation?Yes. R2=R3=0.Robustness theorem
LVDoes operator lift preserve admissibility geometry?No. Partitions commute, metric stratifies.Non-isometry theorem

The Decisive Conceptual Contribution

Most frameworks collapse two distinct properties: partition structure (whether witnesses remain single-gate) and retention geometry (whether witness density is preserved under lift). These are independent. The data show partition structure is preserved and retention geometry is family-dependent. A framework that tests only partition structure would miss the hierarchical stratification entirely; a framework that conflates geometry failure with partition failure would incorrectly declare factorization broken. The LI–LV arc separates them rigorously.

Methodological Significance

FALSIFICATION-FIRST METHODOLOGY Witness Separability Operational test for independence Orthogonality Metrics R2,R3 Channel mixing detected exactly Identity- Preserving Track Same IDs tracked through all lifts Normalization Falsification Null rules out encoding artifacts Lipschitz Migration Bound Staged onset is mathematically bounded Curvature Functional Predicts L_F before chamber

This is not philosophical structuralism. Each chamber has explicit falsifiers, specified in advance. The normalization null was allowed to kill the hypothesis — it didn't. The metric upgrade did not overwrite earlier results. No step relies on reinterpretation of prior failure. That is methodological integrity, and it is rare.

Implications for Operator Theory and Physics

Operator Theory

Operator families form geometric strata. Not all lifts are equivalent. Hierarchy exists in operator geometry with direct implications for multi-scale systems, recursive constraint composition, and stratified admissibility spaces.

Systems Theory

Constraint composition is not freely extensible across dimensions. Curvature sensitivity produces stratified stability classes — resembling phase transitions and renormalization-type phenomena, but derived from a discrete admissibility substrate without assuming physics.

Computational

A kernel → curvature functional → non-isometry constant → migration onset pipeline now exists. Hierarchical stratification is predictable before chamber execution. Discovery becomes computation.

Key Numbers

Mechanisms Tested
56,877
LIII adversarial completeness
Hidden Channels Found
0
Basin unification at G3 boundary
R2, R3 All Families
0.0000
Exactly zero. Not approximately.
Normalization Δ
0.000000
Six decimal places. Structural null.
F2 Slope (G2)
−0.1432
2–3× steeper than stable families
Mechanism Contraction
72%
Gate set eliminates mechanism families
Basin δ_max (LIII)
0.239
< 0.5 threshold; no fifth channel
G3 Residual Conc.
73–89%
Bifurcation captures critical dimension

The Clean Final Statement

No hype. No inflation. Each chamber had explicit falsifiers. No step relies on reinterpretation of prior failure. The normalization null was allowed to kill the hypothesis — it didn't. That is methodological integrity.

Recursive admissibility factorizes into orthogonal constraint channels whose partition structure is invariant under perturbation and tier lift, while their metric geometry stratifies under operator bias according to a predictable curvature functional.

Chamber LV: CERTIFIED. Vertical Arc: CLOSED. Discovery first. Reformulation later.

References and Instrumentation

All chambers operate as self-contained falsification instruments with preregistered protocols. UNNS Research Collective, February 2026.