Documentation

Linglib.Theories.Semantics.BSML.FreeChoice

BSML Free Choice Theorems #

@cite{aloni-2022}

Pragmatic enrichment combined with split disjunction derives free choice inferences. This module proves the core FC results from @cite{aloni-2022} for arbitrary NE-free formulas.

Key Results #

ResultStatement
Modal Disjunction (Fact 3)[α ∨ β]⁺ ⊨_S ◇α ∧ ◇β (NE-free α,β; R state-based)
Narrow Scope FC (Fact 4)[◇(α ∨ β)]⁺ ⊨ ◇α ∧ ◇β (NE-free α,β)
Wide Scope FC (Fact 5)[◇α ∨ ◇β]⁺ ⊨ ◇α ∧ ◇β (NE-free α,β; R indisputable)
Dual Prohibition (Fact 11)[¬◇(α ∨ β)]⁺ ⊨ ¬◇α ∧ ¬◇β (NE-free α,β)
Double Negation FC (Fact 12)[¬¬◇(α ∨ β)]⁺ ⊨ ◇α ∧ ◇β (NE-free α,β)
Negative FC failure (Fact 14)◇¬(α∧β) ⊭{BSML+} ◇¬α; ¬□(α∧β) ⊭{BSML+} ¬□α

Proof Architecture #

Free choice is DERIVED from three independent principles:

  1. Split disjunction: team is partitioned for ∨
  2. NE enrichment: [·]⁺ adds non-emptiness constraints recursively
  3. NE non-flatness: empty teams fail NE, so both parts must be non-empty
theorem Semantics.BSML.atom_downwardClosed {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (p : Atom) (t t' : Finset W) :
t' tsupport M (BSMLFormula.atom p) tsupport M (BSMLFormula.atom p) t'

Atoms are downward-closed. Direct corollary of the substrate's Core.Logic.Team.downwardClosed_support_of_isNEFree when one wants to name the atom case explicitly.

theorem Semantics.BSML.ne_not_downwardClosed {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} [Nonempty W] (M : BSMLModel W Atom) :
¬∀ (t t' : Finset W), t' tsupport M BSMLFormula.ne tsupport M BSMLFormula.ne t'

NE is NOT downward-closed: an inhabited team supports NE but ∅ doesn't. This is the obstruction that prevents the substrate's flat_support_of_isNEFree from extending to NE-bearing formulas.

theorem Semantics.BSML.narrowScopeFC {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (α β : BSMLFormula Atom) (t : Finset W) ( : α.isNEFree = true) ( : β.isNEFree = true) (h : support M (enrich (α.disj β).poss) t) :
support M α.poss t support M β.poss t

Narrow-scope Free Choice: [◇(α ∨ β)]⁺ ⊨ ◇α ∧ ◇β (for NE-free α, β)

(Fact 4 from @cite{aloni-2022})

The enriched disjunction inside ◇ splits the accessible subteam into two non-empty parts (by enrichment + NE). Each part supports its respective disjunct (by enrichment_strengthens_support), and each is a subset of R[w] (via the union), yielding ◇α and ◇β.

theorem Semantics.BSML.wideScopeFC {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (α β : BSMLFormula Atom) (t : Finset W) ( : α.isNEFree = true) ( : β.isNEFree = true) (hInd : M.isIndisputable t) (h : support M (enrich (α.poss.disj β.poss)) t) :
support M α.poss t support M β.poss t

Wide-scope Free Choice: [◇α ∨ ◇β]⁺ ⊨ ◇α ∧ ◇β (for NE-free α, β with indisputable R)

(Fact 5 from @cite{aloni-2022})

The disjunction splits t = t₁ ∪ t₂. From t₁ pick w₁ — it has a subteam s_a ⊆ R[w₁] supporting α. By indisputability R[w] = R[w₁] for all w ∈ t, so s_a ⊆ R[w], yielding ◇α at every world. Symmetrically for β.

theorem Semantics.BSML.modalDisjunction {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (α β : BSMLFormula Atom) (t : Finset W) ( : α.isNEFree = true) ( : β.isNEFree = true) (hSB : M.isStateBased t) (h : support M (enrich (α.disj β)) t) :
support M α.poss t support M β.poss t

Modal Disjunction: [α ∨ β]⁺ ⊨_S ◇α ∧ ◇β (for NE-free α, β with state-based R)

(Fact 3 from @cite{aloni-2022})

Plain disjunction under enrichment, with state-based accessibility, derives possibility of each disjunct — without needing ◇ in the formula. The proof: enriched disjunction splits into two non-empty substates (by enrichment + NE). By state-basedness R[w] = t, so each substate is a non-empty subset of R[w], yielding ◇α and ◇β.

theorem Semantics.BSML.dualProhibition {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (α β : BSMLFormula Atom) (t : Finset W) ( : α.isNEFree = true) ( : β.isNEFree = true) (h : support M (enrich (α.disj β).poss.neg) t) :
support M α.poss.neg t support M β.poss.neg t

Dual Prohibition: [¬◇(α ∨ β)]⁺ ⊨ ¬◇α ∧ ¬◇β (for NE-free α, β)

(Fact 11 from @cite{aloni-2022})

Under negation, FC is correctly cancelled: the enriched prohibition of the disjunction entails prohibition of each disjunct. The proof strips enrichment from the anti-supported disjuncts using Fact 1 (enrichment_strengthens_antiSupport).

theorem Semantics.BSML.doubleNegationFC {W : Type u_1} [DecidableEq W] [Fintype W] {Atom : Type u_2} (M : BSMLModel W Atom) (α β : BSMLFormula Atom) (t : Finset W) ( : α.isNEFree = true) ( : β.isNEFree = true) (h : support M (enrich (α.disj β).poss.neg.neg) t) :
support M α.poss t support M β.poss t

Double Negation FC: [¬¬◇(α ∨ β)]⁺ ⊨ ◇α ∧ ◇β (for NE-free α, β)

(Fact 12 from @cite{aloni-2022})

Under double negation, FC effects re-emerge. Stripping the enrichment around ¬ reveals support M (enrich (◇(α∨β))) t (by bilateral negation swap), and narrow-scope FC (Fact 4) applies.

Fact 14 (◇ version): ◇¬(α ∧ β) ⊭_{BSML+} ◇¬α. Negative FC does not hold under pragmatic enrichment.

Fact 14 (□ version): ¬□(α ∧ β) ⊭_{BSML+} ¬□α. Negative FC with necessity also fails under pragmatic enrichment.