Skip to content

shawnjason/Projection-Insufficiency

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

23 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Projection Insufficiency — Lean Proofs

DOI

Machine-checked Lean 4 proofs for:

"Projection Insufficiency and Trajectory Realization: A Unified Constraint-Based Framework for Bounded Systems"

Paper DOI (concept, always resolves to latest): 10.5281/zenodo.19633241


Author

Shawn Kevin Jason — Independent Researcher, Las Vegas, NV ORCID: ORCID iD 0009-0003-9208-1556


What This Repository Contains

Nine standalone Lean 4 proof files covering the principal formal results of the paper. The proofs split into four groups: a foundational core establishing the projection-insufficiency impossibility, the trajectory-dependence of admissibility, and the constructive partial converse under contractivity; a trajectory-admissibility enforcement template that codifies the canonical same-fiber conflict pattern instantiated by every domain specialization below; the AI-domain specializations of Section 9 (language-model hallucination, planning dead-ends, reinforcement-learning terminal-constraint failure); and the adjacent formalisms of Section 10 (POMDP belief-state insufficiency, constraint-propagation infeasibility in CSPs).

Each file is independent and verifies against the current Mathlib release.


Files

Foundational Results

004_projection_insufficiency.lean — Theorem 2 (Projection Insufficiency Theorem) and Corollary 3 (Non-recoverability is structural) The headline impossibility result. When a projection map P : T → R is non-injective and a property Φ : T → Y differs across some indistinguishability class of P, no function f : R → Y can recover Φ from P. Corollary 3 follows immediately as a structural restatement: non-recoverability is not contingent on the choice of recovery function but on the projection's failure to separate trajectories that differ in the property of interest.

012_admissibility_nonlocal.lean — Proposition 11 (Admissibility is trajectory-dependent) and Corollary 12 (Non-locality of admissibility) Specializes the projection-insufficiency obstruction to predicates: when admissibility is trajectory-dependent relative to a bounded projection, no function on the projection alone recovers the admissibility predicate. The file is self-contained — it includes a re-statement of the underlying projection-insufficiency theorem so it can be verified in isolation without importing the foundational PIT file.

015_contractive_fixed_point_unique.lean — Theorem 13 (Constructive resolution under contractivity) Companion result establishing that under a contractive self-map K : X → X with constant q < 1, K has at most one fixed point — the constructive partial converse to the impossibility theorems above.

Trajectory-Level Enforcement Template

021_trajectory_admissibility_enforcement.lean — Proposition 10 (canonical template for the specialization family) The architectural template instantiated by every specialization below. Defines the SameFiberConflict predicate — the same-projection / different-admissibility pattern — and proves both deterministic and stochastic policy forms: no deterministic local policy and no stochastic local policy can guarantee admissibility uniformly across the indistinguishability classes of a non-injective projection. Each file in the AI-domain and adjacent-formalisms groups below is a domain-specific instance of this template.

AI-Domain Specializations (Section 9)

018_lm_hallucination_ceiling.lean — Section 9.1 (language-model hallucination ceiling) A language model with bounded context window of depth h is a forward-local system whose next-token selection is a function of the bounded context projection. The SameContextHallucinationRisk predicate is the LM specialization of the same-fiber conflict; both deterministic and stochastic decoder forms are proved.

019_planning_dead_end_boundary.lean — Section 9.2 (planning dead-end detection boundary) A planner with bounded local state representation (lookahead horizon plus current state, abstraction, or factored variables) is a forward-local system whose action selection is a function of that bounded representation. The SameRepDeadEndConflict predicate yields both deterministic and stochastic planner forms.

020_rl_terminal_constraint_boundary.lean — Section 9.3 (reinforcement-learning terminal-constraint boundary) An RL policy operating under a finite value horizon h evaluates states using a bounded value-projection of the trajectory. The SameProjTerminalConflict predicate yields both deterministic and stochastic policy forms.

Adjacent Formalisms (Section 10)

022_pomdp_belief_state_insufficiency.lean — Section 10.1 (POMDP belief-state insufficiency) A POMDP agent that derives its belief state from its observation history acts on a function of that history. When two trajectories produce identical observation histories, the composition belief∘obs is a lossy projection; the SameBeliefAdmissibilityConflict predicate yields belief-state insufficiency in both deterministic and stochastic forms.

023_constraint_propagation_boundary.lean — Section 10.2 (constraint-propagation infeasibility boundary) Arc-consistency and generalized arc-consistency algorithms operate on a bounded local view of a partial assignment — the values and adjacencies in a finite neighborhood of constraint structure. The SameViewFeasibilityConflict predicate yields both deterministic and nondeterministic propagation forms.


Mapping to the Paper

Paper Result File Lean Theorem
Theorem 2 (Projection Insufficiency) 004_projection_insufficiency.lean projection_insufficiency
Corollary 3 (Non-recoverability is structural) 004_projection_insufficiency.lean covered by projection_insufficiency
Proposition 11 (Admissibility is trajectory-dependent) 012_admissibility_nonlocal.lean admissibility_nonlocal
Corollary 12 (Non-locality of admissibility) 012_admissibility_nonlocal.lean covered by admissibility_nonlocal
Theorem 13 (Constructive resolution under contractivity) 015_contractive_fixed_point_unique.lean contractive_fixed_point_unique
Proposition 10 (Trajectory admissibility enforcement, deterministic) 021_trajectory_admissibility_enforcement.lean no_deterministic_local_policy_guarantees_admissibility
Proposition 10 (Trajectory admissibility enforcement, stochastic) 021_trajectory_admissibility_enforcement.lean no_stochastic_local_policy_guarantees_admissibility
Section 9.1 (LM hallucination, deterministic) 018_lm_hallucination_ceiling.lean lm_hallucination_ceiling_deterministic
Section 9.1 (LM hallucination, stochastic) 018_lm_hallucination_ceiling.lean lm_hallucination_ceiling_stochastic
Section 9.2 (Planning dead-end, deterministic) 019_planning_dead_end_boundary.lean planning_dead_end_boundary_deterministic
Section 9.2 (Planning dead-end, stochastic) 019_planning_dead_end_boundary.lean planning_dead_end_boundary_stochastic
Section 9.3 (RL terminal-constraint, deterministic) 020_rl_terminal_constraint_boundary.lean rl_terminal_constraint_boundary_deterministic
Section 9.3 (RL terminal-constraint, stochastic) 020_rl_terminal_constraint_boundary.lean rl_terminal_constraint_boundary_stochastic
Section 10.1 (POMDP belief-state, deterministic) 022_pomdp_belief_state_insufficiency.lean pomdp_belief_state_insufficient_deterministic
Section 10.1 (POMDP belief-state, stochastic) 022_pomdp_belief_state_insufficiency.lean pomdp_belief_state_insufficient_stochastic
Section 10.2 (Constraint propagation, deterministic) 023_constraint_propagation_boundary.lean constraint_propagation_boundary_deterministic
Section 10.2 (Constraint propagation, nondeterministic) 023_constraint_propagation_boundary.lean constraint_propagation_boundary_nondeterministic

How to Verify

  1. Open live.lean-lang.org
  2. Confirm the dropdown in the upper right is set to Latest Mathlib
  3. Paste the contents of any .lean file into the editor
  4. Wait for checking to complete — "No goals" on each theorem and no errors in the Problems pane confirms verification

Each file is independent; no cross-file imports are required.


Scope

These proofs verify the formal logical structure of the principal projection-insufficiency theorem, its trajectory-admissibility enforcement companion, the constructive partial converse under contractivity, and the AI-domain and adjacent-formalism specializations of Sections 9 and 10. They do not establish the framework-level conjectures and open claims developed in the paper, nor the philosophical or methodological commentary surrounding the formal results.


Related Work

The forward-case specialization of the projection-insufficiency theorem, with explicit adversarial constructions and cross-domain AI applications, is developed in:

The Non-Locality of Extendability: An Impossibility Theorem for Bounded Information Systems, with Applications to Generative Sequential SystemsDOI: 10.5281/zenodo.19688367 (Lean proofs: 10.5281/zenodo.19687799)

The stochastic extension of the forward-case result, with a quantitative inconsistency-accumulation lower bound and a matching positive representational theorem, is developed in:

Inconsistency Accumulation in Forward-Local Sequential Policies: A Lower Bound under Delayed ConstraintsDOI: 10.5281/zenodo.19688628 (Lean proofs: 10.5281/zenodo.19687094)


License

MIT

About

Machine-checked Lean 4 proofs for "Projection Insufficiency and Trajectory Realization." Establishes that no function on a non-injective projection can recover a trajectory-dependent property, with specializations to language-model hallucination, planning, RL, POMDPs, and constraint propagation.

Topics

Resources

License

Stars

Watchers

Forks

Packages

 
 
 

Contributors

Languages