The Algebra of Admissibility: Formal Structure of Deterministic Deduction
Working Draft
Abstract
We develop a lattice-theoretic framework for classifying configurations as forced, admissible, or inadmissible without enumerating paths. A relaxation operator on the constraint lattice classifies constraints into foundational, ontological, and epistemic types. Epistemic and foundational status are stable under tightening; ontological status can degrade. The forced/admissible/inadmissible partition is invariant under epistemic relaxation at a given grain. In self-determined systems, the constraint-generation operator is antitone; its generic product is a 2-cycle, and a fixed point is a non-trivial physical condition. Four instances are exhibited: finite-energy admissibility forcing non-uniformity, the γ-threshold partitioning concentration dynamics, slow-variation as epistemic constraint, and unique factorization as ontological constraint.
1. Introduction
The problem: given a system whose constraints determine which configurations can exist, classify what is forced (must exist), admissible (can exist), and inadmissible (cannot exist) — without enumerating the configuration space.
The framework originates in Gradient Field Theory, where a field Φ both determines and satisfies its own action functional: δ_Φ S[Φ; Φ] = 0. This paper abstracts that structure. Section 2 gives definitions. Section 3 proves the partition theorem and classification stability. Section 4 treats self-determined systems. Section 5 exhibits instances.
2. Definitions
Definition 2.1 (Configuration space). A triple (X, Σ, μ) where X is a nonempty set, Σ a σ-algebra, and μ a σ-finite measure.
Definition 2.2 (Constraint lattice). A complete lattice (𝒞, ≤) with an admissibility map 𝒜: 𝒞 → Σ satisfying: 𝒜(C₁ ∧ C₂) = 𝒜(C₁) ∩ 𝒜(C₂), 𝒜(C₁ ∨ C₂) = 𝒜(C₁) ∪ 𝒜(C₂), and 𝒜(⊤) = X. Stronger constraints are lower: C₁ ≤ C₂ ⟹ 𝒜(C₁) ⊆ 𝒜(C₂). We work with finitely generated constraint systems throughout; completeness is invoked only for the self-determination results of Section 4.
Definition 2.3 (Partition). Fix a constraint C ∈ 𝒞 and an equivalence relation ~ on X with measurable classes (a determinacy equivalence). A configuration x is:
Forced under (C, ~) if x ∈ 𝒜(C) and 𝒜(C) ⊆ [x]_~.
Admissible under (C, ~) if x ∈ 𝒜(C) and 𝒜(C) ⊄ [x]_~.
Inadmissible under C if x ∉ 𝒜(C).
Forcedness is a property of ~-classes, not of individual configurations.
Definition 2.4 (Relaxation). Given a system S = C₁ ∧ ⋯ ∧ Cₙ, the relaxation with respect to Cₖ is R_k(S) = ∧_{j≠k} Cⱼ. Since S ≤ R_k(S), the admissible set expands: 𝒜(S) ⊆ 𝒜(R_k(S)). The relaxation expansion is Δ_k = 𝒜(R_k(S)) \ 𝒜(S).
Definition 2.5 (Constraint types). A constraint Cₖ in a system S, relative to a determinacy equivalence ~, is classified by its relaxation expansion:
Foundational: removing Cₖ makes other constraints in S uninterpretable. The relaxation R_k(S) is not a well-formed constraint system. No Δ_k exists to examine. (This is a meta-level property: some constraints are preconditions for the formulation of others. The constraint lattice as an algebraic object cannot represent this; it is extra structure on the system. Finite action in GFT is foundational because the variational principle presupposes it.)
Ontological: R_k(S) is well-formed, and Δ_k contains configurations that violate a structural invariant of X — a property derivable from the definition of (X, Σ, μ) without reference to S. (Finite energy in a non-variational system is ontological: removing it admits infinite-energy configurations, which are structurally pathological.)
Epistemic-for-~: R_k(S) is well-formed, and every configuration in Δ_k is ~-equivalent to some configuration in 𝒜(S). The expansion reveals finer structure but no ~-distinguishable novelty. (Slow-variation in GFT is epistemic-for-~_geo: removing it reveals variable coupling constants, but all new configurations agree with existing ones on coarse-grained geometric observables.)
The classification is relative to both the system S and the equivalence relation ~. The same constraint can be foundational in one system (if another constraint depends on it), ontological in a second (if nothing depends on it but its relaxation admits pathology), and epistemic in a third (if the system already excludes the pathology through other constraints).
3. The Partition Theorem and Classification Stability
Theorem 3.1 (Partition). For any C ∈ 𝒞 and determinacy equivalence ~, the partition of X into forced, admissible, and inadmissible is exhaustive and exclusive.
Proof. Every x ∈ X either belongs to 𝒜(C) or does not. If x ∈ 𝒜(C), then either 𝒜(C) ⊆ [x]_~ or not. □
Theorem 3.2 (Epistemic invariance of the forced quotient). If Cₖ is epistemic-for-~ in S, then:
$$\text{[x]}\sim \text{ is forced under } S \iff \text{[x]}\sim \text{ is forced under } R_k(S)$$
Proof. (⟹) 𝒜(S) ⊆ [x]~. Since Cₖ is epistemic-for-~, every y ∈ Δ_k is ~-equivalent to some z ∈ 𝒜(S), hence y ∈ [x]~. So 𝒜(R_k(S)) = 𝒜(S) ∪ Δ_k ⊆ [x]~. (⟸) 𝒜(S) ⊆ 𝒜(R_k(S)) ⊆ [x]~. □
Theorem 3.3 (Classification stability under tightening). Let S’ ≤ S (S’ tighter).
(a) Foundational is preserved. If Cₖ is foundational in S (some Cⱼ in S depends on Cₖ), then Cₖ is foundational in S’. The dependent constraint is still present; dependencies are properties of constraints, not systems.
(b) Epistemic is preserved under independent tightening. If Cₖ is epistemic-for-~ in S, and S’ = S ∧ C’ where C’ does not depend on Cₖ, then Cₖ is epistemic-for-~ in S’. Proof: Δ’_k ⊆ 𝒜(R_k(S)) = 𝒜(S) ∪ Δ_k. Every element of 𝒜(S) is well-formed. Every element of Δ_k is ~-equivalent to something in 𝒜(S). So every element of Δ’_k is ~-equivalent to something in 𝒜(S) ⊆ 𝒜(S’).
(c) Ontological can degrade. If Cₖ is ontological in S, a tighter S’ may independently exclude the pathological configurations in Δ_k, making Δ’_k structurally benign. Cₖ is then epistemic in S’.
(d) Any type can become foundational. If S’ adds a constraint that depends on Cₖ, then Cₖ is foundational in S’ regardless of its type in S.
(e) Inadmissibility is preserved. x ∉ 𝒜(S) ⟹ x ∉ 𝒜(S’). Immediate from 𝒜(S’) ⊆ 𝒜(S).
(f) Forced ~-classes are preserved under nonempty tightening. If 𝒜(S) ⊆ [x]~ and 𝒜(S’) ≠ ∅, then 𝒜(S’) ⊆ [x]~. Proof: 𝒜(S’) ⊆ 𝒜(S) ⊆ [x]_~.
4. Self-Determined Systems
Definition 4.1. A configuration-dependent constraint system is a map 𝒮: X → 𝒞. A configuration x is locally self-determined if x ∈ 𝒜(𝒮(x)): it satisfies the constraints it generates.
Definition 4.2. The constraint-generation operator is
$$\Gamma: \mathcal{C} \to \mathcal{C}, \quad C \mapsto \bigwedge_{x \in \mathcal{A}(C)} \mathcal{S}(x)$$
A globally self-determined constraint is a fixed point of Γ: a constraint C* such that the configurations admissible under C* collectively regenerate C*.
Proposition 4.3 (Γ is antitone). C₁ ≤ C₂ ⟹ Γ(C₁) ≥ Γ(C₂).
Proof. 𝒜(C₁) ⊆ 𝒜(C₂), so Γ(C₁) is a meet over a subset, hence at least as high. □
Corollary 4.4. Γ² is monotone. By Knaster-Tarski, Γ² has fixed points on any complete lattice.
Theorem 4.5 (2-cycle structure). A Γ²-fixed point p satisfies either Γ(p) = p (self-determined) or Γ(p) ≠ p, in which case {p, Γ(p)} is a 2-cycle: constraint p generates configurations that determine Γ(p), and Γ(p) generates configurations that determine p. No abstract condition on the lattice alone excludes 2-cycles. Fixed-point existence requires structure beyond lattice completeness — either constructive exhibition, a Lyapunov functional that is strictly decreasing on non-fixed orbits (with compactness or well-foundedness to guarantee convergence), or a proof that no proper 2-cycles exist.
Local self-determination (a single Φ satisfying its own equations) does not give a Γ-fixed point. Global self-determination requires that all configurations admissible under C* collectively regenerate C*. In GFT, this is resolved when all extremizers of S[·; Φ] agree on the structure field λ. Whether this holds in full generality is open.
A 2-cycle represents an oscillation: the laws determine configurations that determine different laws. Self-determination — reality determining its own constraints — is the assertion that this doesn’t happen. When it holds, the partition theorem (Section 3) applies at C*, and the forced/admissible/inadmissible classification is the structure of self-determined reality.
5. Instances
5.1 Finite-Energy Admissibility → Forced Non-Uniformity
Configuration space: sections of a fiber bundle over ℝ³. Constraint: E[Φ] = ∫ e[Φ, ∇Φ] dμ < ∞. Equivalence relation ~: Φ₁ ~ Φ₂ iff both are spatially uniform or both are non-uniform.
Any non-zero uniform configuration Φ₀ has E[Φ₀] = e(Φ₀) · Vol(ℝ³) = ∞, hence Φ₀ ∉ 𝒜(C). The admissible set 𝒜(C) contains only non-uniform configurations. Therefore [non-uniform]_~ is forced. Non-uniformity is not a contingent feature of the solution — it is a consequence of the constraint.
Classification of finite energy: foundational in the GFT system (the variational principle depends on it); ontological in a non-variational system (it excludes pathological configurations but nothing depends on it for interpretability).
5.2 The γ-Threshold → Forced/Admissible Partition
The AGC equation dAᵢ/dt = Φ_in · Aᵢ^γ / Σⱼ Aⱼ^γ − βAᵢ has qualitatively different behavior depending on γ. Constraint: the field equations with state-dependent coupling. Equivalence relation: Φ₁ ~ Φ₂ iff they exhibit the same qualitative concentration behavior (concentrating vs. distributing).
For γ > 1: concentrating behavior is forced. The superlinear feedback makes concentration self-reinforcing; the constraints require it. For γ < 1: distributing behavior is forced. For γ = 1: the boundary. Both behaviors are admissible; the specific outcome is contingent on initial conditions.
The threshold γ = 1 is a constraint intersection that splits the admissible set into two forced classes. Above and below, the system is determined. At the boundary, it is not. This is the partition theorem applied to a physical system.
5.3 Slow-Variation → Epistemic Constraint
Constraint: |∇λ|/|λ| ≪ 1/L (structure-field gradients negligible at scale L). System: full GFT field equations with finite-energy admissibility. Equivalence relation ~_geo: configurations agreeing on macroscopic geometric observables (curvature, geodesics, causal structure).
Relaxation expansion: configurations with non-negligible ∇λ — variable coupling constants, approximate symmetries, sector coupling. Every such configuration is ~_geo-equivalent to a slow-variation configuration: the macroscopic geometry is the same; only the fine-grained parameter structure differs.
Slow-variation is epistemic-for-~_geo. Removing it yields the full GFT field equations; adding it yields general relativity with fixed constants. The forced ~_geo-class (Einstein’s equation as a coarse-grained statement) is invariant under this relaxation (Theorem 3.2). GR is not approximate in the sense of being wrong — it is the exact description of the forced quotient at the coarse grain.
Under a finer ~_struct that distinguishes structure-field values, slow-variation is not epistemic: the relaxation expansion contains ~_struct-inequivalent configurations. The classification depends on the grain.
5.4 Unique Factorization → Ontological Constraint
Configuration space: multiplicative structures on ℕ (maps ℕ × ℕ → ℕ satisfying associativity and commutativity). Constraint: unique prime factorization. Structural invariant: the Peano axioms and the definition of multiplication force unique factorization as a theorem, not an axiom.
Unique factorization is ontological: it is derivable from the structure of X itself. Relaxing it admits non-UFD structures (e.g., ℤ[√−5] where 6 = 2·3 = (1+√−5)(1−√−5)), which are well-defined algebraic objects but structurally pathological relative to ℕ.
The primes are forced under unique factorization: the admissible multiplicative structures all agree on the prime decomposition. The specific ordering or labeling of primes is admissible (convention-dependent), but their existence and distribution are forced by the constraint.
6. Remarks
On the Galois connection. The admissibility map 𝒜 and its adjoint (the tightening map sending a set B to the strongest constraint admitting B) form a monotone Galois connection. This is standard lattice theory and is not developed here because no theorem in this paper requires it. It becomes relevant when the paper extends to the approximation hierarchy, where coarse-graining maps between configuration spaces induce pullback maps between constraint lattices.
On reachability. A quantitative measure of determinacy — what fraction of configurations are excluded by a constraint — requires a well-defined ratio. For σ-finite spaces, this should be measured relative to a reference constraint (not the unconstrained space), using the pushforward measure q_*μ on the quotient X/~. The formal development is deferred until the paper has theorems about it.
On computational complexity. Deterministic deduction bypasses enumeration when lattice operations (meet, join, relaxation) and quotient-containment checks are tractable in the given representation. Forced and inadmissible classifications are then verification problems; admissible classification remains a search problem. The framework identifies when the problem structure permits the bypass, not a universal complexity reduction.
On the approximation hierarchy. In GFT, the epistemic constraints are organized into a hierarchy: slow-variation, isolated-subsystem, fast-relaxation. Each level is obtained by relaxing an epistemic constraint. The forced quotient at each level is invariant under further relaxation (Theorem 3.2 applied iteratively). At the finest level, no epistemic constraints remain — all constraints are foundational or ontological. The formal treatment (coarse-graining maps as functors, pullback maps as lattice homomorphisms, the inverse limit) is deferred.