Linear type theories, semantics and action calculi
نویسنده
چکیده
In this thesis, we study linear type-theories and their semantics. We present a general framework for such type-theories, and prove certain decidability properties of its equality. We also present intuitionistic linear logic and Milner’s action calculi as instances of the framework, and use our results to show decidability of their respective equality judgements. Firstly, we motivate our development by giving a split-context logic and typetheory, called dual intuitionistic linear logic (DILL), which is equivalent at the level of term equality to the familiar type-theory derived from intuitionistic linear logic (ILL). We give a semantics for the type-theory DILL, and prove soundness and completeness for it; we can then deduce these results for the type-theory ILL by virtue of our translation. Secondly, we generalise DILL to obtain a general logic, type-theory and semantics based on an arbitrary set of operators, or general natural deduction rules. We again prove soundness and completeness results, augmented in this case by an initiality result. We introduce Milner’s action calculi, and present example instances of our framework which are isomorphic to them. We extend this isomorphism to three significant higher-order variants of the action calculi, having functional properties, and compare the induced semantics for these action calculi with those given previously. Thirdly, motivated by these functional extensions, we define higher-order instances of our general framework, which are equipped with functional structure, proceeding as before to give logic, type-theory and semantics. We show that the logic and type-theory DILL arise as a higher-order instance of our general framework. We then define the higher-order extension of any instance of our framework, and use a Yoneda lemma argument to show that the obvious embedding from an instance into its higher-order extension is conservative. This has the corollary that the embeddings from the action calculi into the higher-order action calculi are all conservative, extending a result of Milner. Finally, we introduce relations, a syntax derived from proof-nets, for our general framework, and use them to show that certain instances of our framework, including some higher-order instances, have decidable equality judgements. This immediately shows that the equalities of DILL, ILL, the action calculi and the higher-order action calculi are decidable.
منابع مشابه
A Complete Characterization of Complete Intersection-Type Theories
We characterize those intersection-type theories which yield complete intersection-type assignment systems for λ-calculi, with respect to the three canonical set-theoretical semantics for intersection-types: the inference semantics, the simple semantics and the F-semantics.
متن کاملA Lambek Calculus with Dependent Types∗
In this note, we discuss how to introduce dependent types into the Lambek calculus [7] (or, in general, an ordered calculus [17] and extensions such as those found in the studies of categorial grammars [12, 13]). One of the motivations to introduce dependent types in syntactical analysis is to facilitate a closer correspondence between syntax and semantics, especially when modern type theories ...
متن کاملLinear realizability and full completeness for typed lambda-calculi
We present the model construction technique called Linear Realizability. It consists in building a category of Partial Equivalence Relations over a Linear Combinatory Algebra. We illustrate how it can be used to provide models, which are fully complete for various typed λ-calculi. In particular, we focus on special Linear Combinatory Algebras of partial involutions, and we present PER models ov...
متن کاملStrong Normalisation in the π-Cal ulus
We introduce a typed π-calculus where strong normalisation is ensured by typability. Strong normalisation is a useful property in many computational contexts, including distributed systems. In spite of its simplicity, our type discipline captures a wide class of converging name-passing interactive behaviour. The proof of strong normalisability combines methods from typed λ-calculi and linear lo...
متن کاملStrong Normalisation in the π-Calculus
We introduce a typed π-calculus where strong normalisation is ensured by typability. Strong normalisation is a useful property in many computational contexts, including distributed systems. In spite of its simplicity, our type discipline captures a wide class of converging name-passing interactive behaviour. The proof of strong normalisability combines methods from typed λ-calculi and linear lo...
متن کامل