Skip to Main content Skip to Navigation

Non-idempotent typing operators, beyond the lambda-calculus

Abstract : In this dissertation, we extend the methods of non-idempotent intersection type theory, pioneered by Gardner and de Carvalho, to some calculi beyond the lambda-calculus.- We first present a characterization of head and strong normalization in the lambda-mu calculus (classical natural deduction) by introducing non-idempotent union types. As in the intuitionistic case, non-idempotency allows us to extract quantitative information from the typing derivations and we obtain proofs of termination that are far more elementary than those in the idempotent case. These results leads us to define a small-step variant of the lambda-mu calculus, in which strong normalization is also characterized by means of quantitative methods.- In the second part of the dissertation, we extend the characterization of weak normalization in the pure lambda-calculus to an infinitary lambda-calculus narrowly related to Böhm trees, which was introduced by Klop et al. This gives a positive answer to a question known as Klop's problem. In that purpose, it is necessary to simultaneously introduce a system (system S) featuring infinite types and resorting to an intersection operator that we call sequential, and a validity criterion in order to discard unsound proofs that coinductive grammars give rise to. This also allows us to give a solution to TLCA problem #20 (type-theoretic characterization of hereditary permutations). It is to be noted that those two problem do not have a solution in the finite case (Tatsuta, 2007).- Finally, we study the expressive power of coinductive type grammars, without any validity criterion. We must once more resort to system S and we show that every term is typable in a non-trivial way with infinite types and that one can extract semantical information from those typings e.g. the order (arity) of any lambda-term. This leads us to introduce a method that allows typing totally unproductive terms (the so-called mute terms), which is inspired from first order logic. This result establishes that, in the coinductive extension of the relational model, every term has a non-empty interpretation. Using a similar method, we also prove that system S surjectively collapses on the set of points of this model
Document type :
Complete list of metadatas

Cited literature [253 references]  Display  Hide  Download
Contributor : Abes Star :  Contact
Submitted on : Monday, October 15, 2018 - 3:29:07 PM
Last modification on : Friday, October 23, 2020 - 5:02:34 PM
Long-term archiving on: : Wednesday, January 16, 2019 - 3:32:38 PM


Version validated by the jury (STAR)


  • HAL Id : tel-01895859, version 1


Pierre Vial. Non-idempotent typing operators, beyond the lambda-calculus. Computation and Language [cs.CL]. Université Sorbonne Paris Cité, 2017. English. ⟨NNT : 2017USPCC038⟩. ⟨tel-01895859⟩



Record views


Files downloads