Skip to Main content Skip to Navigation
Conference papers

Type inference in intuitionistic linear logic

Abstract : We study the type checking and type inference problems for intuitionistic linear logic: given a System F typed -term, (i) for an alleged linear logic type, determine whether there exists a corresponding typing derivation in linear logic (type checking) (ii) provide a concise description of all possible corresponding linear logic typings (type inference). We solve these problems using a novel algorithmic type system for linear logic whose typing rules carry arithmetic side conditions describing essentially the nesting depth of (proof-net) boxes. By understanding these side conditions as unknowns we then reduce type inference to solving a system of arithmetic constraints. We show that these constraint systems fall into a tractable class hence leading to an efficient (polynomial-time) solution. There are two important restrictions: first, our source language is typed System F rather than untyped lambda calculus; this is necessary because type inference for System F is known to be undecidable. Second, we assume that sharing is made explicit in the input, thus we do not try to automatically infer opportunities for sharing identical subterms. Relieving the latter restriction is left as a challenge for future work.
Document type :
Conference papers
Complete list of metadatas

https://hal.archives-ouvertes.fr/hal-00543120
Contributor : Patrick Baillot <>
Submitted on : Sunday, December 5, 2010 - 7:27:47 PM
Last modification on : Wednesday, November 20, 2019 - 3:02:31 AM

Identifiers

Collections

Citation

Patrick Baillot, Martin Hofmann. Type inference in intuitionistic linear logic. 12th International ACM SIGPLAN Conference on Principles and Practice of Declarative Programming (PPDP 2010), Jul 2010, Hagenberg, Austria. pp.219-230, ⟨10.1145/1836089.1836118⟩. ⟨hal-00543120⟩

Share

Metrics

Record views

140