We want to concentrate on the development of mathematical libraries for theorem proving tools. This objective contributes to two main areas of application: tools for mathematicians and correctness verification tools for software dealing with numerical computation.
In the short term, we aim for mathematical libraries that concern polynomials, algebra, group theory, floating point numbers, real numbers, big integers, probabilities and geometrical objects. In the long run, we think that this will involve any function that may be of use in embedded software for control or robotics (in what is called hybrid systems, systems that contain both software and physical components) and in cryptographical systems. We want to integrate these libraries in theorem proving tools because we believe they will become important tools for mathematical practice and for engineers who need to prove the correctness of their algorithms and software.
We believe that theorem proving tools are good tools to produce highly dependable software, because they provide a framework where algorithms and specifications can be studied uniformly and often provide means to mechanically derive programs that are correct by construction.
We also study the extensibility of interactive theorem proving tools based on decision procedures that free designers from the burden of verifying some of the required properties. We often rely on “satisfiability modulo theory” procedures, which can be connected to theorem proving tools in a way that preserves the trustability of the final results.
The calculus of inductive constructions is a branch of type theory that serves as a foundation for theorem proving tools, especially the Coq proof assistant. It is powerful enough to formalize complex mathematics, based on algebraic structures and operations. This is especially important as we want to produce proofs of logical properties for these algebraic structures, a goal that is only marginally addressed in most scientific computation systems.
The calculus of inductive constructions also makes it possible to write algorithms as recursive functional programs which manipulate tree-like data structures. A third important characteristic of this calculus is that it is also a language for manipulating proofs. All this makes this calculus a tool of choice for our investigations. However, this language still is the object of improvements and part of our work focusses on these improvements.
To produce certified algorithms, we use the following approach: instead of attempting to prove properties of an existing program written in a conventional programming language such as C or Java, we produce new programs in the calculus of constructions whose correctness is an immediate consequence of their construction. This has several advantages. First, we work at a high level of abstraction, independently of the target implementation language. Secondly, we concentrate on specific characteristics of the algorithm, and abstract away from the rest (for instance, we abstract away from memory management or data implementation strategies). Therefore, we are able to address more high-level mathematics and to express more general properties without being overwhelmed by implementation details.
However, this approach also presents a few drawbacks. For instance, the calculus of constructions usually imposes that recursive programs should explicitly terminate for all inputs. For some algorithms, we need to use advanced concepts (for instance, well-founded relations) to make the property of termination explicit, and proofs of correctness become especially difficult in this setting.
To bridge the gap between our high-level descriptions of algorithms and conventional programming languages, we investigate the algorithms that are present in programming language implementations, for instance algorithms that are used in a compiler or a static analysis tool. When working on these algorithms, we usually base our work on the semantic description of the programming language. The properties that we attempt to prove for an algorithm are, for example, that an optimization respects the meaning of programs or that the programs produced are free of some unwanted behavior. In practice, we rely on this study of programming language semantics to propose extensions to theorem proving tools or to verify that compilers for conventional programming languages are exempt from bugs.
The Coq Proof Assistant
Keywords: Proof - Certification - Formalisation
Functional Description
Coq provides both a dependently-typed functional programming language and a logical formalism, which, altogether, support the formalisation of mathematical theories and the specification and certification of properties of programs. Coq also provides a large and extensible set of automatic or semi-automatic proof methods. Coq's programs are extractible to OCaml, Haskell, Scheme, ...
Participants: Benjamin Gregoire, Enrico Tassi, Bruno Barras, Yves Bertot, Pierre Boutillier, Xavier Clerc, Pierre Courtieu, Maxime Dénès, Stephane Glondu, Vincent Gross, Hugo Herbelin, Pierre Letouzey, Assia Mahboubi, Julien Narboux, Jean-Marc Notin, Christine Paulin-Mohring, Pierre-Marie Pedrot, Loic Pottier, Matthias Puech, Yann Regis-Gianas, François Ripault, Matthieu Sozeau, Arnaud Spiwack, Pierre-Yves Strub, Benjamin Werner, Guillaume Melquiond and Jean-Christophe Filliatre
Partners: CNRS - ENS Lyon - Université Paris-Diderot - Université Paris-Sud
Contact: Matthieu Sozeau
URL: http://
The Marelle team, in collaboration with the pi.r2 team, plays an important role in the development of Coq. During this year, we contributed to the 8.6 version of Coq, released in December. As the release manager, Maxime Dénès led the implementation of a time-based release process, aiming at shorter and more predicitible release cycles. We successfully transitioned to 10-month cycles and hope to soon move to 6-month cycles, making it easier for users to benefit from the latest improvements.
At a more detailed level, members of the Marelle team attended the Coq developer meetings (organized in Paris by Maxime Dénès and Matthieu Sozeau) and contributed to the development of Coq concerning bug fixes for virtual machine execution (Benjamin Grégoire and Maxime Dénès), cleaning up the API for plug-in developers (Matej Košík), improving the State Transaction Machine (Enrico Tassi), setting up a package index based on OPAM (Enrico Tassi), introducing a system to discuss Coq Enhancement Proposals (Enrico Tassi), and implementing a new configurable system of warnings (Maxime Dénès).
We supervise of an engineer working at MIT on questions related to efficient proof construction and proof development environments, in cooperation with researchers from the pi.r2 team. The collaboration with MIT was also an occasion to reflect on the licence framework governing collaborations around the Coq system.
We also prepared the set-up of a consortium to gather intensive users and contributors to the development of Coq. This was an occasion to work with the promotors of the InriaSoft structure which is expected to host the consortium in the long run.
Functional Description
EasyCrypt is a toolset for reasoning about relational properties of probabilistic computations with adversarial code. Its main application is the construction and verification of game-based cryptographic proofs. EasyCrypt can also be used for reasoning about differential privacy.
Participants: Gilles Barthe, Benjamin Gregoire and Pierre-Yves Strub
Contact: Benjamin Grégoire
This year, development on this software system concerned the development of new logical settings to work on differential privay problems: a Hoare logic based on union bound and a logic based on probabilistic couplings.
Mathematical Components library
Functional Description
The Mathematical Components library is a set of Coq libraries that cover the mechanization of the proof of the Odd Order Theorem.
Participants: Andrea Asperti, Jeremy Avigad, Yves Bertot, Cyril Cohen, Francois Garillot, Georges Gonthier, Stéphane Le Roux, Assia Mahboubi, Sidi Ould Biha, Ioana Pasca, Laurence Rideau, Alexey Solovyev, Enrico Tassi, Laurent Théry and Russell O'Connor
Contact: Assia Mahboubi
URL: http://
This year we contributed to the library by adding a new module to cover finite sets within potentially infinite finite types, organizing tutorials and schools to teach its usage:
in January in Sophia Antipolis (one-week format) https://
in August in Nancy (one-day tutorial format, colocated with the ITP conference, organized by Assia Mahboubi
and Enrico Tassi, with contributions by Yves Bertot, Cyril Cohen, and Laurent Théry)
https://
in November in Sophia Antipols https://
Functional Description
Ssreflect is a tactic language extension to the Coq system, developed by the Mathematical Components team.
Participants: Cyril Cohen, Yves Bertot, Laurence Rideau, Enrico Tassi, Laurent Thery, Assia Mahboubi and Georges Gonthier
Contact: Yves Bertot
This year we mainly performed maintenance operations on this software extension to the Coq system (Enrico Tassi).
Functional Description
ZooCrypt is an automated tool for analyzing the security of padding-based public-key encryption schemes (i.e. schemes built from trapdoor permutations and hash functions). This year, we extended the tool to be able to deal with schemes based on cyclic groups and bilinear maps.
Participants: Benjamin Gregoire, Gilles Barthe and Pierre-Yves Strub
Contact: Benjamin Grégoire
We carried on our experiments with extensions of
We carried on our experiments with the Coqoon integrated development environment. This lead to a preliminary report submitted for publication
As an effort to lower the entry barrier to use a structured library of formalized mathematics, we wrote a book explaining the principles of ssreflect and mathematical components. This book-in-the-making is available on github at https://
In the previous year, we developed formally verified proofs that
This work mainly concerns Univalent Foundations and Homotopy Type Theory
which builds on recently discovered connections between type theory
and abstract homotopy theory. The main question we have been working on
lately is finding a computational interpretation for the univalence
axiom, the main fruit of this work is a recent paper on, and
implementation of, cubical type theory which provides a
constructive justification for this axiom. The code is visible at
https://
Anders Mörtberg also recently visited Thierry Coquand to start a collaboration on the formalization of this model in the UniMath system implemented in Coq. Together with Benedikt Ahrens in the Ascola team at Inria Nantes and Ralph Matthes at IRIT in Toulouse, Anders Mörtberg also worked on the formalization of a translation from binding signatures to monads for representing languages with binders in UniMath . This work uses the new possibilities for representing category theory in type theory that univalence provides.
As part of the ANR Fastrelax project, we have started to formalize double-word arithmetic algorithms, in particular the sum of a double-word and a floating point number and the sum of two double-word numbers described in the article " Tight and rigourous error bounds for basic building blocks of double-word arithmetic" .
We formalized the 3D geometry concepts used in the description of kinematics chains, in particular: rotations, rigid body tranformations, screw motions, frame changes, and the Denavit-Hartenberg Convention. This lead to a publication to appear in the internaltional conference CPP 2017 .
We extend the Mathematical Components library with a module concerning finite sets (in potentially infinite types), finite maps and multisets. This module plays a crucial role in the formalization of nominal sets, multinomials, semi-algebraic sets, and many experimental developments.
We also extend the Mathematical Components library with a module concerning orders, lattices, and sets. This serves as an abstraction on various libraries, including the finite set library, semi-algebraic sets, finite reunions of intervals, and boolean predicates (in classical theories).
Extending work by Guillaume Cano, Cyril Cohen, Maxime Dénès, Anders Mörtberg and Vincent Silès, we reimplemented the foundations of the CoqEAL library on Keller and Lasson's parametricity plug-in and provided a more robust translation mechanism. We illustrated the use of this enhanced version of CoqEAL on a new version of the traditional ring tactic. This lead to a publication at JFLA 2017 (Journées Francophones des Langages Applicatifs, the article actually is in English) .
We developed the necessary results about first-order logical formulae to be able to define semi-algebraic sets and semi-algebraic functions in Coq. This required that we provide elements of language to describe quantification over blocks of variables. We show that the equality of semi-algebraic sets is decidable, thanks to the already formalized decision procedure based on quantifier elimination. We then show that our formalized semi-algebraic sets do satisfy general abstract interfaces for sets, as seen in section
In the long run this work will be instrumental to describe the output of cylindrical algebraic decomposition algorithms. Indeed, this output is usually made of semi-algebraic sets.
We formalize the spectral theorem for normal, hermitian and unitary
matrices (this work in progress is available at https://
We started formalizing the proof of La Salle's invariance principle using the Coquelicot library, with the goal of using it to formalize the proof of stability of a control function for the inverted pendulum (a basic exercise that can serve as an introduction to problems in robotics). For now, I have proven a few properties of the set of limit points of a function.
We studied the applicability of the mathematical component library to describe Delaunay triangulation algorithms in the most abstract way. We also formalized a theorem on convex functions known as Jensen's inequality.
We have formalized an algorithm proposed by Peter Selinger to synthesize quantum gates. His approach mixes number theoretical notions and linear algebra, two aspects that are well covered by the Mathematical Components Library.
De Bruijn sequences are combinatorial objects. We have shown how they can be generated by exhibiting a link with irreducible polynomials in finite fields, with a formal proof in Coq.
The problem of Hanoi towers is a standard example to explain recursion. While trying to write a formalization, we discovered that there exists an interesting generalisation. Starting with two arbitrary valid positions, the problem is to find an optimal solution to go from one to the other. The solution is somewhat counter-intuitive, and not always unique. We formalized it in Coq.
A paper describing our implementation of the sets of natural numbers, of rational numbers and of real numbers has been published by the Journal of Formalized Reasoning .
We implemented Chapter 3, Section 7 (Inverse Limits and Direct Limits) and the start of Chapter 4 (Structures) of the Theory of Sets of Bourbaki, details are found in the Research Report
Ordinal numbers have been designed at approximately the same time that the foundations of mathematics were being revisited, in the beginning of the 20th century. These objects cross the boundaries of set theory and pose especially difficult challenges when considering the task of formalizing mathematics. This is the reason why we concentrate on formal proofs concerning these objects.
An ordinal number
.
We proposed new logics to work on examples from the differential privacy literature, a hoare logic based on the union bound and a logic based on the deep connection between differential privacy and probabilistic couplings , .
Differential power analysis (DPA) is a side-channel attack in which an adversary retrieves cryptographic material by measuring and analyzing the power consumption of the device on which the cryptographic algorithm under attack executes. We introduced new notions and models allowing to check the correctness of counter measures (known as masking schemes) , . Based on this idea we have developed a compiler to transform an unmasked program into it masked version.
We are currently members of two projects funded by the French national agency for research funding.
BRUTUS "Chiffrements authentifiés et résistants aux attaques par canaux auxiliaires", started on October 1st, 2014, for 60 months, with a grant of 41 kEuros for Marelle. Other partners are Université de Rennes 1, CNRS, secrétariat Général de la défense et de la sécurité nationale, and Université des Sciences et Technologies de Lille 1. The corresponding researcher for this contract is Benjamin Grégoire.
FastRelax, "Fast and Reliable Approximations", started on October 1st, 2014, for 60 months, with a grant of 75 kEuros for Marelle. Other partners are Inria Grenoble (ARIC project-team), LAAS-CNRS (Toulouse), Inria Saclay (Toccata and Specfun project-teams), and LIP6-CNRS (Paris). The corresponding researcher for this contract is Laurence Rideau.
We work with the team of Adam Chlipala at MIT, in particular the engineer Paul Steckler, with whom we have regular meetings concerning the optimization of parts of the Coq system with respect to use cases provided by the MIT team, and the design of user-interface tools. This engineer had a visit of 6 weeks in France in April, three weeks in the pi.r2 team (mostly hosted by Matthieu Sozeau) and three weeks in the Marelle team, mostly hosted by Enrico Tassi and Maxime Dénès. The collaboration continues since that visit with a weekly phone conference.
We had visits by Gilles Barthe (IMDEA, Madrid, Spain) for 2 weeks, Benedikt Schmidt (IMDEA), for 2 weeks, François-Xavier Standaert (Université Catholique de Louvain, Crypto Group, Belgium), for 1 week, Sebastian Faust (Ruhr-University Bochum, Germany) for 1 week, François Dupressoir (IMDEA) for 1 week, Pierre-Yves Strub (IMDEA), for 1 week, and Peter Schwabe (Radboud University, Nijmegen, the Netherlands) for 3 days.
Benjamin Grégoire visited IMDEA (Madrid, Spain) for two one-week trips.
Yves Bertot, Maxime Dénès, and Enrico Tassi visited Princeton University in June for the kick-off meeting of the Expedition in Computing entitled “the science of deep specification” funded by the NSF foundation.
Enrico Tassi visited the team of Jesper Bengtson at the IT-University of Copenhagen, Denmark.
Anders Mörtberg visited the team of Thierry Coquand at Chalmers and University of Göteborg in Sweden.
Yves Bertot is program co-chair, with Viktor Vafeiadis from MPI-SWS in Germany for the ACM conference Certified Programs and Proofs (CPP) to be held in Paris in January 2017. Most of the editorial activities took place in 2016.
Yves Bertot and Laurent Théry were members of the conference program committee for the conference Interactive Theorem Proving (ITP) and User-Interfaces for Theorem Provers (UITP).
Cyril Cohen was a member ot the program committee for the 8th Coq workshop.
Cyril Cohen was reviewer for the conferences CSL 2016 and ITP 2016. Laurent Théry was a reviewer for the conferences TACAS'17 and CPP'17. Benjamin Grégoire was a reviewer for TACAS. Benjamin Grégoire was a reviewer for PoPL 2017.
Cyril Cohen was a reviewer for Journal of Automated Reasoning. Laurent Théry was a reviewer for Journal of Automated Reasoning and Journal of Symbolic Computation. Yves Bertot was a reviewer for Journal of Automated Reasoning and Computational Geometry: Theory and Applications.
Laurent Théry gave an invited talk at MAP'16 (Mathematics, Algorithms, and Proofs).
Cyril Cohen gave an invited talk at the ELFIC seminar on the Paris-Saclay campus (Elfic stands for Éléments finis formellement vérifiés).
Yves Bertot and Maxime Dénès have been working on setting up a Consortium of users for the Coq system. The consortium should start in the early days of 2017. Yves Bertot, Enrico Tassi, and Maxime Dénès were invited to the kick-off meeting of the Expedition in Computing entitled “the science of deep specification” funded by the NSF foundation, along with three other developers from the pi.r2 project-team, as expert developers of the Coq system. This kick-off meeting took place in June.
Laurent Théry evaluated projects for the French national agency for research funding (ANR),
José Grimm is a member of the local committee for Hygiene and Work safety,
Cyril Cohen served several times as secretary for the local committee of project-team leaders,
Benjamin Grégoire is a member of the committee on computer tools usage (CUMI) for the Sophia-Antipolis Méditerranée Inria center.
Licence : Cyril Cohen, mathematics oral exam, 30 hours, Classes préparatoires aux grandes écoles
Master : Laurent Théry gave a course at ENS Lyon (9 hours), a course at École des Mines (3 hours), and a course at University de Marseille (3 hours). Yves Bertot gave a one-week introductory course on Coq at University of Nice (21 hours). Enrico Tassi organized a one-week advanced course on Coq and Mathematical Components for students of ENS Lyon and University of Nice (30 hours). There were two instances of this school, in January and in November, teachers for this course were Enrico Tassi, Yves Bertot, Cyril Cohen, Laurence Rideau, and Laurent Théry.
PhD in progress : Boris Djalal, started in October 2015, supervised by Yves Bertot and Cyril Cohen
PhD in progress : Cécile Baritel-Ruet, started in October 2016, supervised by Yves Bertot and Benjamin Grégoire
PhD in progress : Sophie Bernard, started in October 2016, supervised by Yves Bertot and Laurence Rideau
PhD in progress : Damien Rouhling, started in October 2016, supervised by Yves Bertot and Cyril Cohen.
Yves Bertot was member of the defense committee for the thesis of Jacques-Henri Jourdan.
Laurent Théry gave talks in the context of “Fête de la science”.