Protected-State-Correction-Theory

Proof Status Map

Claim ID Short Name Exact Statement Status Evidence Type Confidence Key Files Next Action                        
OCP-001 Formal OCP tuple An OCP system can be formalized as a protected/disturbance/correction tuple with exact and asymptotic branches. PROVED repo-local definition and consistency checks High docs/formalism/formal-theory.md Use as the primary formal language.                        
OCP-002 Exact orthogonal projection theorem If H = S ⊕ D with S orthogonal to D, orthogonal projection onto S exactly recovers the protected component of any state s+d. PROVED linear-algebra proof and tests High docs/theorem-candidates/central-theorem.md; tests/math/test_core_projectors.py Promote as the clean finite-dimensional backbone.                        
OCP-003 Indistinguishability no-go If S ∩ D is nontrivial, no single-valued recovery map can recover s from x=s+d for all s in S and d in D. PROVED elementary contradiction proof High docs/impossibility-results/no-go-results.md Promote as the strongest exact no-go statement.                        
OCP-004 Continuous damping theorem The flow xdot = -k P_D x leaves S fixed and exponentially suppresses D. PROVED closed-form solution and tests High docs/theorem-candidates/backup-theorems.md; tests/math/test_core_projectors.py Use as the bridge from exact correction to asymptotic correction.                        
OCP-005 QEC exact anchor Knill-Laflamme style exact correction fits OCP when syndrome sectors are orthogonal and recoverable. CONDITIONAL standard QEC theorem plus repo-local rewrite High docs/qec/qec-in-ocp.md; tests/math/test_qec_knill_laflamme.py Keep exact but conditional on KL-type assumptions.                        
OCP-006 Helmholtz/Leray exact continuous anchor Periodic Helmholtz projection gives an exact OCP recovery operator for divergence-free fields under additive gradient contamination. PROVED FFT operator construction and tests High docs/mhd/divergence-cleaning-in-ocp.md; tests/math/test_mhd_projection.py Promote heavily as the best exact continuous example.                        
OCP-007 GLM as asymptotic OCP GLM divergence cleaning fits OCP as an asymptotic correction architecture rather than an exact projector. CONDITIONAL implementation analysis and empirical reduction tests Medium docs/mhd/glm-and-asymptotic-correction.md; tests/examples/test_glm_decay.py Present as a good asymptotic example, not exact correction.                        
OCP-008 Control-theoretic instantiation Observer/controller designs instantiate OCP when the protected/disturbance split is invariant and feedback acts only on disturbance coordinates. CONDITIONAL structural control analysis Medium docs/control/control-extension.md Keep conditional and design-oriented.                        
OCP-009 Universal scalar correction capacity There exists one scalar correction-capacity number that nontrivially classifies QEC, MHD projection, GLM, and control architectures. OPEN failed unification attempt Low NOVELTY_AND_LIMITS.md; docs/impossibility-results/no-go-results.md Do not promote without a better category-specific definition.                        
OCP-010 Topological Adam / ML extension Current optimizer work is a genuine OCP instantiation rather than a loose bridge. ANALOGY ONLY local sibling-repo evidence is too weak Low docs/disproven-or-weak/weak-extensions.md Keep secondary and unpromoted.                        
OCP-011 Engineering design value OCP yields practical design rules for building correction operators and recognizing when a correction architecture is underpowered. CONDITIONAL operator examples and no-go criteria Medium docs/applications/practical-use-cases.md Use as the main practical pitch, but keep the domain assumptions explicit.                        
OCP-012 Exact-to-asymptotic bridge OCP gives a meaningful bridge between discrete exact correction and continuous asymptotic correction. CONDITIONAL formalism + exact and asymptotic branches Medium docs/formalism/exact-vs-asymptotic.md Promote after upgrading the bridge from framework language to a sharper theorem statement.                        
OCP-013 Invariant-split generator theorem If K annihilates S, preserves D, and its restriction to D is exponentially stable, then the flow xdot=-Kx preserves S and asymptotically suppresses D. PROVED linear ODE proof and tests High docs/theorem-candidates/generator-theorems.md; tests/math/test_continuous_generators.py Promote as the strongest next-step theorem beyond the projector case.                        
OCP-014 Self-adjoint PSD corollary If K is self-adjoint positive semidefinite with ker(K)=S and spectral gap lambda_*>0 on S^perp, then   e^{-tK}x-P_S x   <= e^{-lambda_* t}   P_D x   . PROVED spectral proof and tests High docs/theorem-candidates/generator-theorems.md; tests/math/test_continuous_generators.py Use as the cleanest grounded continuous-time strengthening.        
OCP-015 Mixing no-go for linear flows If P_S K P_D is nonzero, then disturbances generically leak into the protected coordinates and the flow is not an OCP correction flow preserving S pointwise. PROVED derivative-at-zero argument and tests High docs/impossibility-results/no-go-results.md; tests/math/test_continuous_generators.py Promote as the strongest new linear-flow failure criterion in the repo.                        
OCP-016 Exact correction rank lower bound For an exact linear recovery on V = S ⊕ D with correction operator C = I - R, one must have rank(C) >= dim(D) and rank(R) >= dim(S). PROVED linear-algebra proof and tests High docs/theorem-candidates/capacity-theorems.md; tests/math/test_capacity.py Promote as the cleanest minimum-structure theorem in the repo.                        
OCP-017 Sector distinguishability lower bound Pairwise orthogonal nonzero exact-correction sectors require at least as many distinguishable correction labels as sectors. CONDITIONAL sector-based argument plus QEC example Medium docs/theorem-candidates/capacity-theorems.md; tests/math/test_capacity.py Keep this branch-specific and avoid overstating it beyond the exact sector model.                        
OCP-018 Category-specific capacity view OCP supports branch-specific correction-capacity notions even though a universal scalar capacity remains unsupported. CONDITIONAL definitions, lower bounds, and examples across branches Medium docs/theorem-candidates/capacity-theorems.md; docs/open-questions/viable-next-directions.md Develop this further before treating it as a mature theory.                        
OCP-019 Exact sector recovery theorem Pairwise orthogonal coordinate-compatible disturbance sectors admit an exact sector-conditioned recovery operator mapping each sector back to the protected space. PROVED operator construction, proof sketch, and tests High docs/theorem-candidates/sector-recovery-theorems.md; tests/math/test_sector_recovery.py Promote as the finished exact sector branch beyond the simple projector model.                        
OCP-020 Finite-time exact recovery no-go for smooth linear flows If D is nontrivial, a finite-time linear flow e^{-tK} cannot satisfy Φ_t(s+d)=s for all s in S, d in D at any finite t>0. PROVED invertibility argument plus workbench and test examples High docs/impossibility-results/advanced-no-go-results.md; tests/math/test_continuous_generators.py Use this as the sharp separator between exact recovery and asymptotic linear correction.                        
OCP-021 Sector-overlap detection no-go If two candidate disturbance sectors overlap nontrivially, exact unique sector detection is impossible on the shared family. PROVED subspace contradiction proof and overlapping-sector test case High docs/impossibility-results/advanced-no-go-results.md; tests/math/test_sector_recovery.py Promote as the sector-branch analog of the overlap no-go.                        
OCP-022 Transverse gauge projection fit On projection-compatible domains, Coulomb-gauge or transverse-field projection is an exact protected-state correction fit by the same projector logic as the Helmholtz/Leray branch. PROVED ON SUPPORTED FAMILY direct corollary of the projection branch plus physics reinterpretation Medium docs/physics/maxwell-coulomb-gauge.md; docs/mhd/divergence-cleaning-in-ocp.md Keep as a physics extension, not as a novel operator theorem.                        
OCP-023 Periodic projector transplant to bounded domains The periodic exact projector can be reused unchanged as an exact bounded-domain correction operator preserving the bounded protected class. DISPROVED explicit boundary counterexample and tests High docs/physics/bounded-domain-projection-limits.md; tests/examples/test_bounded_domain_projection_limit.py Use as the clearest physics-side rejection of naive projector transplantation.                        
OCP-024 Constraint damping in additional physics systems Constraint damping architectures in systems such as numerical relativity fit the asymptotic OCP branch when a protected/constraint-violation split and a stable correction generator are made explicit. CONDITIONAL cross-system operator comparison Medium docs/physics/numerical-relativity-constraint-damping.md; docs/mhd/glm-and-asymptotic-correction.md Keep as a viable physics extension and future application lane.                        
OCP-025 Continuous quantum error correction bridge Continuous quantum error correction and feedback architectures fit the asymptotic or measurement-conditioned OCP branches when syndrome information remains sufficiently distinguishable and feedback preserves the code sector. CONDITIONAL QEC/control literature comparison plus branch mapping Medium docs/physics/continuous-quantum-error-correction.md; docs/qec/qec-in-ocp.md Keep as a citable future bridge, but do not promote it as a proved new theorem.                        
OCP-026 Generic constrained Hamiltonian fit Generic constrained Hamiltonian systems automatically determine a nontrivial OCP correction architecture. ANALOGY ONLY current repo lacks a canonical projector/recovery construction in that generality Low docs/physics/kept-vs-rejected-physics-bridges.md Do not promote without an explicit reduction or correction operator.                        
OCP-027 Periodic incompressible projection fit Periodic incompressible velocity projection is an exact protected-state correction fit by the same Helmholtz/Hodge projector logic as the exact continuous branch. PROVED ON SUPPORTED FAMILY operator corollary plus executable periodic CFD tests High docs/cfd/incompressible-projection.md; docs/theorem-candidates/cfd-projection-results.md; tests/math/test_cfd_projection.py Keep as the narrow strongest CFD extension.                        
OCP-028 Divergence-only bounded recovery no-go On a nontrivial bounded incompressible protected class, any recovery map that factors only through the divergence scalar field cannot be an exact recovery operator. PROVED elementary distinguishability argument plus bounded-state witness High docs/cfd/bounded-vs-periodic-projection.md; docs/theorem-candidates/cfd-projection-results.md; tests/math/test_cfd_projection.py Promote as the sharp CFD-facing limitation theorem.                        
OCP-029 Bounded-domain projection classification Bounded-domain incompressible projection belongs to the exact branch only when the correction step realizes the domain-compatible Hodge projector for the actual protected class and boundary data. CONDITIONAL classification statement tied to the bounded counterexample and Hodge-projector criterion Medium docs/cfd/helmholtz-hodge-velocity-projection.md; docs/cfd/cfd-system-matrix.md; docs/physics/bounded-domain-projection-limits.md Keep this conditional until a full bounded-domain operator theorem is proved.                        
OCP-030 Observation fiber exactness A protected variable is exactly recoverable from a constrained record map if and only if it is constant on the record fibers; equivalently κ_{M,p}(0)=0. PROVED formal derivation plus branch tests High docs/theory/advanced-directions/constrained-observation-formalism.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py Keep as the backbone of the constrained-observation branch, but do not market it as a novelty claim by itself.                        
OCP-031 Restricted linear protected-variable recovery For x = Fz, M(x)=Ox, and p(x)=Lx on a finite-dimensional family, exact linear protected-variable recovery exists if and only if ker(O F) is contained in ker(L F). PROVED linear-algebra derivation and tests High docs/theory/advanced-directions/constrained-observation-formalism.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py Use as the main exact bridge to observability and restricted recovery.                        
OCP-032 Fixed-basis phase-loss no-go A fixed computational-basis record cannot exactly recover a phase-sensitive protected variable on a qubit family that contains the same amplitudes with varying phase, even though weaker protected variables such as the z coordinate remain recoverable. PROVED toy-model derivation and computational sweep Medium docs/impossibility-results/constrained-observation-no-go.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Keep as the cleanest quantum-side no-go inside the branch.                        
OCP-033 Periodic record classification On the tested periodic incompressible family, full vorticity is an exact protected-variable record, truncated vorticity is approximate only, and divergence-only observation is a no-go. CONDITIONAL conventional computational benchmark with generated artifacts Medium docs/theory/advanced-directions/constrained-observation-results-report.md; data/generated/recoverability/recoverability_summary.json; tests/math/test_recoverability.py Keep as a strong computational result, but do not overstate it beyond the tested family.                        
OCP-034 Finite-history versus asymptotic recovery split In the two-state scalar-output control toy model, one-step exact recovery fails, two-step protected-variable recovery is exact whenever ε(a-b) ≠ 0, and observer-based asymptotic recovery still converges from the ongoing record. PROVED model-specific derivation, explicit formula, and computational sweep Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Keep as the branch’s clearest exact-versus-asymptotic observation example, but only at the toy-model level.                        
OCP-035 Collapse-modulus noise lower bound For any estimator receiving records with adversarial observation error at most η, the worst-case protected-variable error is bounded below by κ_{M,p}(η)/2. PROVED metric derivation plus analytic benchmark and tests High docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py Promote as the strongest operational theorem currently supporting the κ branch.                        
OCP-036 Restricted observation rank lower bound On a finite-dimensional admissible linear family x=Fz with record OFz and protected variable LFz, exact protected-variable recovery implies rank(OF) >= rank(LF). PROVED rank-nullity derivation and tests Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py Keep as a branch-specific minimum-record lower bound without reviving the old universal-capacity idea.                        
OCP-037 Qubit phase-window collision law For the fixed-basis qubit record on the phase-window family φ ∈ [-w,w], the full Bloch-vector fiber ambiguity satisfies κ(0)=2 sin(min(w, π/2)); exact full recovery holds only at w=0, while weaker protected variables such as z remain exactly recoverable. PROVED closed-form derivation and numerical sweep Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Keep as the cleanest family-level phase-transition law in the branch.                        
OCP-038 Periodic cutoff threshold on the two-mode family For the tested two-mode periodic incompressible family, truncated-vorticity recovery is exact if and only if the Fourier cutoff retains both active modes, which first occurs at cutoff 2. PROVED family-level rank argument plus recovery-error sweep Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Promote as a narrow but real coarsening threshold result and keep it explicitly family-specific.                        
OCP-039 Periodic functional-support threshold On the tested four-mode periodic incompressible modal family, exact recovery of a protected linear functional from truncated vorticity is possible if and only if the cutoff retains every Fourier mode used by that functional; the minimal exact cutoff therefore equals the largest protected mode cutoff. PROVED kernel-based derivation, row-space residual checks, discretization checks, and recovery sweeps Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Promote as the branch’s strongest current periodic minimal-record threshold law, but keep it family-specific.                        
OCP-040 Nested linear minimal observation complexity For a nested finite-dimensional linear observation family x=Fz with records O_r F z and protected variable L F z, exact recovery turns on at the first level r for which row(L F) is contained in row(O_r F). PROVED restricted-linear derivation plus independent periodic/control threshold checks Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py Keep as the cleanest current generalization of the branch’s family-level threshold laws.                        
OCP-043 Nested restricted-linear collision-gap threshold law On the bounded coefficient family A_B = {F z :   z   _∞ ≤ B}, the structured collision gap Γ_r(B) is monotone along nested record families, vanishes exactly at the first exact-recovery level, and yields a zero-noise lower bound Γ_r(B)/2 below threshold. PROVED restricted-linear derivation, independent nullspace-on-a-box checks, stress sweeps, and generated-artifact consistency tests High docs/theory/advanced-directions/constrained-observation-formalism.md; docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; tests/math/test_recoverability.py; tests/examples/test_generated_artifact_consistency.py Promote as the branch’s strongest current theorem-grade threshold result, but keep it explicitly within the restricted linear setting.                
OCP-041 Same-record weaker-versus-stronger split A fixed coarse record can exactly recover a weaker protected variable while exact recovery of a stronger protected variable remains impossible whenever the weaker protected row lies in the observation row space and the stronger one does not. PROVED row-space argument plus periodic/control counterexample families Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theory/advanced-directions/constrained-observation-clean-results.md; tests/math/test_recoverability.py Promote as a useful negative result and organizing principle without overselling it as a broad new theory.                        
OCP-042 Diagonal functional interpolation threshold In the tested scalar-output diagonal family with distinct active eigenvalues, exact recovery of a protected linear functional g·x_0 is possible at horizon H if and only if g_j / c_j agrees on the active sensor spectrum with a polynomial of degree at most H-1, with hidden protected directions remaining impossible; the coordinate threshold is a special case. PROVED Vandermonde/interpolation derivation plus independent linear-recovery, nullspace, and sweep tests Medium docs/theory/advanced-directions/constrained-observation-derivations.md; docs/theorem-candidates/constrained-observation-theorems.md; docs/theory/advanced-directions/constrained-observation-results-report.md; tests/math/test_recoverability.py Promote as the stronger control-side threshold law and demote the older coordinate-only wording to a corollary.                        
OCP-044 Boundary-compatible finite-mode Hodge projection On a bounded-domain family built from boundary-compatible divergence-free stream modes and Dirichlet gradient modes, the orthogonal projector onto the protected span is an exact bounded-domain correction operator. PROVED ON SUPPORTED FAMILY integration-by-parts derivation plus independent discrete projector-construction checks and tests Medium docs/theorem-candidates/bounded-domain-hodge-theorems.md; docs/theorem-candidates/cfd-projection-results.md; tests/math/test_cfd_projection.py; tests/examples/test_generated_artifact_consistency.py Promote as a restricted but real partial solution to the bounded-domain exact-correction problem.                        
OCP-045 Restricted-linear minimal augmentation theorem For a restricted linear family x = F z with record O F z and protected variable L F z, the minimum number of unrestricted added linear measurements needed for exact recovery is δ(O, L; F) = rank([O F; L F]) - rank(O F). PROVED row-space derivation, random stress tests, and design-artifact consistency checks High docs/theorem-candidates/capacity-theorems.md; src/ocp/design.py; tests/math/test_design.py; tests/examples/test_design_examples_consistency.py Promote as the first finished category-specific capacity invariant beyond lower bounds alone.                        
OCP-046 Restricted-linear exact-regime upper envelope If exact recovery holds on a restricted linear family and K satisfies K O F = L F, then κ_{M,p}(δ) ≤   K   _2 δ for Euclidean record and protected metrics. PROVED direct linear derivation plus exact-case computational checks and generated recoverability artifacts Medium docs/theory/advanced-directions/pvrt-theory-program.md; docs/theorem-candidates/pvrt-theorem-spine.md; docs/theory/advanced-directions/constrained-observation-derivations.md; tests/math/test_recoverability.py; tests/examples/test_generated_artifact_consistency.py Promote as the strongest current κ upper bound, but keep it explicitly inside the exact restricted-linear branch.                
OCP-047 Same-rank observation insufficiency In the restricted-linear branch, same-rank record families can have opposite exact-recovery verdicts; observation amount alone does not determine exact protected-variable recoverability. PROVED explicit construction plus dimension-stress tests Medium docs/theory/advanced-directions/pvrt-theory-program.md; docs/theorem-candidates/pvrt-theorem-spine.md; docs/theory/advanced-directions/pvrt-falsification-program.md; tests/math/test_recoverability.py Promote as the clean falsification of rank-only or amount-only recoverability language.                        
OCP-048 Detectable-only through target coarsening If q = φ ∘ p, exact recoverability of p implies exact recoverability of q, while the converse can fail even on finite or restricted-linear families; exact recovery of such a nonconstant coarsening while p fails is the branch detectable-only regime. PROVED factorization argument plus finite and restricted-linear witnesses Medium docs/fiber-based-recoverability-and-impossibility/fiber-formalism-core.md; docs/fiber-based-recoverability-and-impossibility/fibers-and-detectability.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py Keep as the fiber-based branch’s clean target-hierarchy theorem and use it carefully when translating detect-vs-correct language across fields.                        
OCP-049 No rank-only exact classifier theorem For every n > r ≥ 1 and every r ≤ k < n, there exist restricted finite-dimensional linear recoverability problems with the same ambient dimension n, protected rank r, and observation rank k but opposite exactness verdicts, so no classifier depending only on (n, rank(LF), rank(OF)) can decide exact recoverability on all such families. PROVED explicit witness construction, exhaustive small-dimension coordinate enumeration, and generated artifacts High docs/fiber-based-recoverability-and-impossibility/restricted-linear-fiber-theory.md; docs/fiber-based-recoverability-and-impossibility/fibers-and-impossibility.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py; data/generated/unified-recoverability/unified_recoverability_summary.json Promote as the fiber-based branch’s strongest surviving negative theorem and as the cleanest statement of where amount-only unification fails.                        
OCP-050 No fixed-library budget-only exact classifier theorem For every n > r ≥ 1 and every r ≤ k < n, there exists a common coordinate candidate library with unit measurement costs and two selections of the same size k and same total cost k that have opposite exact-recovery verdicts, so fixed-library budget/count alone cannot decide exact recoverability on all restricted finite-dimensional linear families. PROVED explicit fixed-library construction, exhaustive subset enumeration, and generated artifacts High docs/fiber-based-recoverability-and-impossibility/restricted-linear-fiber-theory.md; docs/fiber-based-recoverability-and-impossibility/fibers-and-impossibility.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py; data/generated/unified-recoverability/candidate_library_budget_witnesses.csv Promote as the stronger fixed-library version of the anti-universal theorem and use it against sensor-count/budget-only overclaims.                        
OCP-051 Noisy weaker-versus-stronger separation theorem On a restricted linear family x = F z, if a weaker target W F z is exactly recoverable by K O F = W F while a stronger target S F z has positive collision gap Γ on the same bounded family, then bounded record noise   e   ≤ η yields a weak-target error upper bound   K   _2 η while every decoder for the stronger target still has worst-case error at least Γ/2; in particular the weak upper bound stays strictly below the stronger impossibility floor for η < Γ/(2   K   _2). PROVED restricted-linear derivation plus brute-force noise-grid cross-checks and generated artifacts High docs/fiber-based-recoverability-and-impossibility/fibers-and-weaker-vs-stronger-targets.md; docs/fiber-based-recoverability-and-impossibility/restricted-linear-fiber-theory.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py; data/generated/unified-recoverability/noisy_restricted_linear_hierarchy.csv Promote as the branch’s strongest current noisy target-hierarchy theorem and use it to separate stable weak recovery from strong-target impossibility.
OCP-052 Restricted-linear family-enlargement false-positive theorem Let F_s and F_l be restricted-linear admissible family bases with span(F_s) contained in span(F_l). If exact recovery holds on the smaller family but ker(O F_l) is not contained in ker(L F_l), then exact recovery fails on the enlarged family; moreover the enlarged-family collision gap Γ_l yields the lower bound Γ_l/2 for every decoder on the enlarged family, including any decoder exact on the smaller family. PROVED fiber-factorization argument plus collision-gap lower bound, explicit witness construction, generated artifacts, and branch tests High docs/fiber-based-recoverability-and-impossibility/theorem-candidates.md; docs/fiber-based-recoverability-and-impossibility/false-positive-risk-report.md; docs/fiber-based-recoverability-and-impossibility/restricted-linear-fiber-theory.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py; data/generated/unified-recoverability/family_enlargement_false_positive.csv Promote as the branch’s cleanest current admissible-family-enlargement false-positive certificate, with family scope kept explicit.                        
OCP-053 Canonical model-mismatch instability theorem On the canonical restricted-linear family F_beta = span{e1, e2 + beta e3} with record M(x) = (x1, x2) and target p(x) = x3, each fixed beta yields exact target identifiability, but the decoder exact on F_beta0 incurs worst-case exact-data target error beta - beta0 / sqrt(1 + beta^2) on the true family F_beta over the unit coefficient box. Thus exact identifiability of the true family does not imply robustness of a mismatched inverse map. PROVED closed-form derivation plus brute-force equality checks and generated artifacts High docs/fiber-based-recoverability-and-impossibility/theorem-candidates.md; docs/fiber-based-recoverability-and-impossibility/model-mismatch-stress-report.md; docs/fiber-based-recoverability-and-impossibility/final-inverse-observability-pass-report.md; src/ocp/fiber_limits.py; tests/math/test_unified_limits.py; data/generated/unified-recoverability/canonical_model_mismatch.csv Promote as the branch’s cleanest current exact model-mismatch theorem, with canonical family scope kept explicit.