Linear Types and Locality

摘要

We introduce a system of linear dependent types, extended with quantifiers that ensure separation between distinct bound variables. Such variables may be interpreted as resources that can be accessed only locally. The main motivation for this system, is to make more manageable the logic encoding of specification formalisms based on graphs and state-transition models. The proof system is based on a sequent calculus presentation of quantified intuitionistic linear logic, relying on double-entry sequents. We prove the admissibility of cut, and show that this result can be used to prove subject reduction.