natural deduction metalanguage, practical foundations
type theory (dependent, intensional, observational type theory, homotopy type theory)
computational trinitarianism = propositions as types +programs as proofs +relation type theory/category theory
These are a class of model structures on certain categories of cubical sets, due to Christian Sattler, motivated by cubical type theory.
Let $\mathbb{C}$ be a small category with finite products. Assume there is a (cartesian) interval object $\mathbb{I}$, i.e., an object $\mathbb{I}$ with two morphisms $0,1 : e \to \mathbb{I}$, where $e$ is the terminal object in $\mathbb{C}$.
Assume further that we have connection maps $\vee,\wedge : \mathbb{I}\times\mathbb{I} \to \mathbb{I}$ such that $x \vee 0 = 0 \vee x = x$ and $x \wedge 1 = 1 \wedge x = x$.
We also need a face lattice $\mathbb{F}$, that is, a sublattice of the subobject classifier $\Omega$ in the presheaf category $cSet = \mathbb{C}^{op} \to Set$ containing the endpoints of each cylinder $J^+ = J \times \mathbb{I}$ for each object $J$ of $\mathbb{C}$, and an operation $\forall : \mathbb{F}^\mathbb{I} \to \mathbb{F}$ right adjoint to the projection in the sense that $\psi \le \forall \delta$ iff $\psi p \le \delta$, where $p : \mathbb{F} \times \mathbb{I} \to \mathbb{F}$.
If we want the model to be effective, we also require that each proposition $\psi = 1$ with $\psi \in \mathbb{F}(I)$ is decidable. (This rules out taking $\mathbb{F} = \Omega$.)
A simple and central example is the full subcategory of the category of posets $Pos$ on powers of $\mathbb{I} = (0 \lt 1)$. This is also the Lawvere theory of distributive lattices. The idempotent completion is the full subcategory of $Pos$ on finite lattices.
Other prominent examples are the Lawvere theories of de Morgan, Kleene, and boolean algebras.
For computational purposes we can take $\mathbb{F}$ to be the smallest lattice containing the cylinder endpoints (the faces of cubes). To get Cisinski model structures we can take $\mathbb{F} = \Omega$.
One may wonder whether these models structures are equivalent to the model in simplicial sets. This is not the case for Cartisian cubes; see mailing-list. However, this model does form a Grothendieck (infinity,1)-topos, because it satisfies the Giraud axioms. For deMorgan cubical sets the geometric realization is not a Quillen equivalence (because of the reversal map); the counterexample is the unit interval quotient by the symmetry); see Sattler. Whether they are equivalent by another map is not yet excluded. The question whether the geometric realization for the cubical sets based on distribute lattices is an equivalence is still open.
Thierry Coquand, A model structure on some presheaf categories, pdf
Thierry Coquand, Some examples of complete Cisinski model structures, pdf
Christian Sattler, Do cubical models of type theory also model homotopy types, lecture at Hausdorff Trimester Program: Types, Sets and Constructions, youtube
Last revised on February 15, 2019 at 22:56:39. See the history of this page for a list of all contributions to it.