Module lambda_data 

Lambda Data (ΛD) — Epistemic State Vector codec.

Formal basis (paper_lambda_data.md):

ΛD: V → (V × O × C × T) ψ = ⟨T, V, E⟩ where E = ⟨c, τ, ρ, δ⟩

This module implements the lossless binary codec for ΛD state vectors. Unlike JSON projection (π_JSON(ψ) = V, which discards T and E), the ΛD binary format preserves the full epistemic tensor across serialization boundaries.

Invariants enforced at encode boundary:

1. Ontological Rigidity: T ∈ O ∧ T ≠ ⊥
2. Singular Interpretation: V ∈ dom(T) (deferred to runtime)
3. Semantic Conservation: type preservation across transformations
4. Epistemic Bounding: c ∈ [0,1] ∧ δ ∈ Δ

Theorem 5.1 (Epistemic Degradation): For any composition f operating on ΛD inputs, c_out ≤ min(c_in₁, c_in₂, …, c_inₙ) Enforced at compose time.

Structs

LambdaData

ψ = ⟨T, V, E⟩ where E = ⟨c, τ, ρ, δ⟩

Enums

Derivation

δ ∈ Δ = {raw, derived, inferred, aggregated, transformed}

LdError

Functions

apply_provenance_ceiling

§Fase 53.d.2 — tainted-overriding (founder refinement A). A PROVENANCE member’s declared default_confidence — an extension member’s ceiling (§53), or the built-in epistemic:<level> axis — is a CEILING on the announced certainty, never a floor. When a value with certainty input_c is annotated with such a member, the announced certainty degrades to min(ceiling, input_c).

compose

Compose two ΛD state vectors under Theorem 5.1 (Epistemic Degradation).

decode

Decode a ΛD state vector from binary. Validates invariants after decode.

encode

Encode a ΛD state vector to binary. Validates invariants at boundary.

from_ir

Create a LambdaData from IR fields (bridge from compiler to runtime).

run_ld

Run axon ld subcommand. Returns exit code.

to_json

π_JSON(ψ) — lossy projection that discards epistemic tensor.