Theorems
Every major theorem in the paper, ordered from the weakest hypothesis to the strongest conclusion. The first three form the central three-tier escalation; the rest extend the same machinery to discrete, multi-turn, stochastic, and pipelined settings.
The three tiers at a glance
| Tier | Theorem | Extra hypothesis | Conclusion |
|---|---|---|---|
| T1 | Boundary Fixation | connected + Hausdorff | ∃ z with f(z)=τ and D(z)=z |
| T2 | ε-Robust Constraint | f is L-Lipschitz, D is K-Lipschitz | f(D(x)) ≥ τ − LK·d(x,z) |
| T3 | Persistent Unsafe Region | transversality: G > ℓ(K+1) | positive-measure set with f(D(x)) > τ |
All theorems
Core impossibility
- T1 · Boundary Fixation — the pointwise result (paper Thm 4.1, Lean
MoF_08). - T2 · ε-Robust Constraint — the Lipschitz neighborhood bound (paper Thm 5.1, Lean
MoF_11). - T3 · Persistent Unsafe Region — the measure-theoretic result under transversality (paper Thm 6.3, Lean
MoF_11). - Defense Dilemma (K*) — the tradeoff in choosing the defense's Lipschitz constant (paper Thm 7.3, Lean
MoF_19).
Discrete and continuous
- Discrete Impossibility — discrete IVT + non-injectivity dilemma; no topology required (paper Thm 8.2–8.3, Lean
MoF_12). - Continuous Relaxation (Tietze) — the bridge from finitely many observations to the continuous theory (paper Thm 8.1, Lean
MoF_ContinuousRelaxation).
Extensions
- Multi-Turn Impossibility — the impossibility recurs at every turn (paper Thm 9.1, Lean
MoF_13). - Stochastic Impossibility — expected behavior inherits boundary fixation (paper Thm 9.2, Lean
MoF_13). - Pipeline Degradation — Lipschitz constants multiply; depth makes defense harder (paper Thm 9.4, Lean
MoF_15).
Unification
- Meta-theorem — a single representation-independent statement that implies T1, the discrete dilemma, and the stochastic impossibility (Lean
MoF_14).
Quantitative bounds
- Volume bounds — explicit lower bounds on
and the cone-based persistent region (paper Thm 7.1–7.2, Lean MoF_17,MoF_18).