Gabriel Scherer

I am a PhD student in the Gallium team (see our shared research blog!) at INRIA Paris-Rocquencourt, France. My advisor is Didier Rémy.

My main interest is programming languages. I'm familiar with statically typed functional programming languages such as OCaml, or more generally of ML tradition (SML, Haskell...). I'm interested in theoretical aspects of type systems, programming language implementation, general programming language concepts, and even some syntactic aspects. I have a preference for the formalizable aspects, or formalizable approaches to programming language aspects, rather than the often subjective appeals to taste or intuition.

Contact information

Outside the personal sphere, I communicate by e-mail -- and interleaved forum, mailing-list or blog posts. Please feel free to send me an email; I mostly use my Gmail adress, gabriel dot scherer at gmail.

Research themes and documents

Types having a unique inhabitant

I'm currently working on the question of knowing whether a given type T has a unique inhabitant -- modulo a given notion of program equivalence. I wrote two blog posts (part 1, and part 2) on the idea, and gave in may 2013 a talk at the PLUME seminar in Lyon, whose slides are available.

As a part of this research, I have been looking at the work on multi-focusing for type/proof systems with sum types. You may be interested by a recent draft on a correspondence between maximally multi-focused proofs and an equivalence algorithm for simply-typed lambda-calculus by Sam Lindley. I gave an introductory talk on ((maximal) multi-)focusing whose slides are available.

Theoretical aspects of mixing GADTs and subtyping in OCaml

When GADTs were added by Jacques Garrigue and Jacques Le Normand in OCaml 4.00, we found out that their interaction with subtyping had not been studied theoretically, in particular we were not sure how to soundly assign variances to GADT parameters, and a very restrictive criterion was implement in OCaml. I worked in 2012 on finding the right variance check for GADTs, and this resulted in the article GADTs meet subtyping (arxiv), also available as a long version or slides.

Meta-theory of predicative dependent type theories, using logical relations

I worked with Andreas Abel on meta-theoretical aspects of predicative dependent type theories -- definitional-algorithmic correspondence with logical relations, and universe subtyping and polymorphism. I enjoyed collaborating to his article On Irrelevance and Algorithmic Equality in Predicative Type Theory (arxiv), and also wrote a much less rigorous internship report on the question of universe subtyping.

MLF type inference and type-level computation

In a former internship with Didier Rémy at Gallium, I worked on MLF, an System F-like type system with type inference; I did some prototyping and explored adding Fω-style type-level operators. A horribly-formated report is available, but this topic would need some more work because it raised more questions than it answered.

A type-safe database library using phantom types

In June-July 2009, I worked with Jérôme Vouillon on a statically typed domain-specific language for database queries embedded in OCaml, Macaque. The library is publicly available and in a minimal maintainance mode (if bugs are found, they get fixed). We wrote a french-language article for JFLA'10.