Team algebra — pure Finset combinatorics for team semantics #
[Ant21] [Vaa07] [yang-vaananen-2017]
The combinatorial primitives underlying team-semantic logics — BSML (Aloni 2022), QBSML (Aloni & van Ormondt 2023), propositional team logic PT⁺ (Yang & Väänänen 2017), modal team logic MTL (Hella et al., Lück), modal dependence logic MDᵂ (Yang), and inquisitive variants.
This file is theory-neutral: nothing in it knows about evaluation, support, or formulas. It provides only the partition relations on teams and the frame conditions on accessibility.
Naming: "team" vs. "state" #
We use "team" (Hodges 1997, Väänänen 2007) — the foundational term in dependence/independence logic and propositional team logic. Aloni's "state-based modal logic" papers use "state" for the same object (a subset of evaluation points), inheriting via Anttila 2021. The two terms are interchangeable in this layer; "team" is preferred here because it generalizes cleanly: a team can be a set of worlds (BSML), a set of (world, assignment) pairs (QBSML), a set of assignments (dependence logic), etc., without the "state" connotation of "set of worlds".
Two layers in this file #
Team partition (
splitsAs,splitsAsNE): the structural condition underlying team-semantic split disjunction.Frame conditions on accessibility (
IsStateBased,IsIndisputable): Anttila Definition 2.2.10-equivalent properties of a relationR : W → Finset Wrelative to a base team. Used to distinguish epistemic and deontic modalities (Aloni 2022 §6.1).
Family roadmap #
Team-semantic logics form a family along two axes ([Ant25]):
a signature axis (propositional, modal, first-order) and a
closure-class axis over team properties (∅ ∈ ·, IsLowerSet,
SupClosed, Set.OrdConnected, with DC = convex + empty). Each logic
is pinned to a cell by an expressive-completeness theorem (⟦L⟧ = the
properties in that cell), so the cell is a theorem about a logic, not a
directory. This substrate provides the closure predicates those theorems
are stated in; formalised consumers so far are BSML, QBSML, MDL, MIL,
InqML under Core/Logic/Modal/.
The shared abstraction is a Definability plus uniform-definability
lemma layer, not a bundled TeamLogic class — no closure law is shared
across cells. Refactor backward from concrete instances; do not extract
the abstraction forward (cf. the ≥ 3-systems rule). Full long-run shape,
target tree, and dependency-ordered build phases:
Core/Logic/Modal/README.md.
Team partition (split disjunction primitive) #
Binary cover: team s is the union of teams t₁ and t₂. The two
sub-teams may overlap; only their union must equal s. This is the
structural condition under team-semantic split disjunction (also called
tensor disjunction in dependence logic; cf. Anttila 2021 Definition
2.1.5, Yang & Väänänen 2017).
Argument order s t₁ t₂ reads as "splits team s into t₁ and t₂".
Defined as abbrev so that t₁ ∪ t₂ = s reduces transparently in proofs
— consumers don't need unfold splitsAs and can pattern-match on the
underlying union equation directly.
Equations
- Core.Logic.Team.splitsAs s t₁ t₂ = (t₁ ∪ t₂ = s)
Instances For
Binary cover with both parts non-empty. Used by pragmatically-enriched
split disjunction in BSML+ / QBSML+ (Aloni 2022 §3.3): [φ ∨ ψ]⁺ requires
both sub-teams to be non-empty (forced by the NE conjunct propagating
through enrichment).
Equations
- Core.Logic.Team.splitsAsNE s t₁ t₂ = (t₁ ∪ t₂ = s ∧ t₁.Nonempty ∧ t₂.Nonempty)
Instances For
The trivial split: s = s ∪ ∅. Used by classical formulas, which are
supported on the empty team vacuously.
The reflexive split: s = s ∪ s (parts may overlap).
Substate property: if splitsAs s t₁ t₂, then t₁ ⊆ s.
Equations
- Core.Logic.Team.instDecidableSplitsAs s t₁ t₂ = decEq (t₁ ∪ t₂) s
Equations
- Core.Logic.Team.instDecidableSplitsAsNE s t₁ t₂ = instDecidableAnd
Frame conditions on accessibility #
R is state-based on team s iff every world in s is R-accessible
exactly to s. Strictly stronger than indisputability.
Aloni 2022 Definition 5; Anttila 2021 Definition 4.10-style.
Equations
- Core.Logic.Team.IsStateBased R s = ∀ w ∈ s, R w = s
Instances For
R is indisputable on team s iff all worlds in s see the same
set of accessible worlds. Equivalently: R is constant on s.
Aloni 2022 Definition 5 (indisputable ↔ deontic-with-knowledgeable-speaker).
Equations
- Core.Logic.Team.IsIndisputable R s = ∀ w₁ ∈ s, ∀ w₂ ∈ s, R w₁ = R w₂
Instances For
State-based implies indisputable.
Equations
- Core.Logic.Team.instDecidableIsStateBasedOfDecidableEqOfFintype R s = id inferInstance
Equations
- Core.Logic.Team.instDecidableIsIndisputableOfDecidableEqOfFintype R s = id inferInstance