# Mathieu Acher

Full Professor of Computer Science · INSA Rennes · IRISA · Inria · DiverSE team

Member of the [Institut Universitaire de France](https://www.iufrance.fr/les-membres-de-liuf/membre/2282-mathieu-acher.html) (IUF)

Website: https://www.mathieuacher.com/ · Email: mathieu.acher@irisa.fr · Last updated: 2026-10-19

I am a Full Professor of Computer Science at [INSA Rennes](https://www.insa-rennes.fr/), a member (*sociétaire*) of the [Institut Universitaire de France](https://www.iufrance.fr/les-membres-de-liuf/membre/2282-mathieu-acher.html) (IUF), which each year recognizes a small number of French academics for the excellence of their research, and a researcher in the [DiverSE](https://www.diverse-team.fr/) team (Inria, IRISA, CNRS, Université de Rennes). I was appointed junior member of the IUF in 2021, on its then-new chair in science outreach (*Médiation scientifique*). Before joining INSA, I was Associate Professor at the University of Rennes 1 (2012–2022). In DiverSE, I lead the research axis on variability engineering. I co-lead the Inria challenge [LLM4Code](https://project.inria.fr/llm4code/) on reliable code assistants based on large language models, and the Inria–Simula associate team [RIPOST](https://diverse-project.github.io/ripost_ea/) on resilient and reproducible software. I am also strongly involved in [FRAIME](https://www.inria.fr/fr/fraime), the new Inria–Mitsubishi Electric challenge on trustworthy AI with formal methods, and in [CodeCommons](https://codecommons.org/), a France 2030 project around Software Heritage, where I lead work package 2 and a team of nine engineers.

My research is about **variability**: software that comes in many variants and configurations, from the Linux kernel and video encoders to scientific pipelines and machine learning systems. I work on modelling, reverse engineering, testing, and learning these huge configuration spaces, and on what they mean for the reproducibility of science. I also study how large language models and coding agents change the way we build, migrate, and vary software, from legacy COBOL to chess engines written in a dozen programming languages.

I am also a chess player (FIDE Master, three international master norms), always up for a game of blitz. On my YouTube channel [CodeplAI](https://www.youtube.com/@CodeplAI) (in French), I explore AI, programming, and games: puzzles, algorithms, variants, and playing with and against AI.

Profiles: [Email](mailto:mathieu.acher@irisa.fr) · [Blog](https://blog.mathieuacher.com) · [Google Scholar](https://scholar.google.com/citations?user=cC6xt1YAAAAJ) · [DBLP](https://dblp.org/pid/57/2898.html) · [HAL](https://cv.hal.science/mathieu-acher) · [ORCID](https://orcid.org/0000-0003-1483-3858) · [GitHub](https://github.com/acherm) · [LinkedIn](https://www.linkedin.com/in/mathieuacher) · [Slides](https://speakerdeck.com/acherm) · [YouTube](https://www.youtube.com/@CodeplAI) · [Bluesky](https://bsky.app/profile/macher.bsky.social) · [X / Twitter](https://x.com/acherm)

## News

- Oct 2026: Paper at ASE 2026: *Code Transformation Rule Synthesis Using LLMs: Potential and Limits*, with Axel Allain, Aymeric Blot, and Djamel Eddine Khelladi ([arXiv](https://arxiv.org/abs/2609.03592)).
- Oct 2026: **Most Influential Paper Award** at [VARIABILITY 2026](https://conf.researchr.org/track/variability-2026/variability-2026-mip-award), the new conference merging SPLC, VaMoS, and ICSR, for *Using Machine Learning to Infer Constraints for Product Lines* (SPLC 2016), with Paul Temple, José A. Galindo, and Jean-Marc Jézéquel. I was also Journal-First co-chair.
- Sep 2026: **Paper accepted at NeurIPS 2026**: *Externalized CPDAG Summaries Improve LLM Causal Deduction*, with my PhD student Wentao Sun (CIFRE with Nokia), João Paulo Nogueira, Dominique Verchère, and Alonso Silva ([arXiv](https://arxiv.org/abs/2609.31071)).
- Sep 2026: **Distinguished Paper Award** at SCAM 2026 for *CRACS: Question-Driven Assessment of Legacy Code Summaries in Industrial IBM-i Modernization*, with N'Guessan Hermann Kouadio, Olivier Barais, and Jean-Marc Jézéquel, in collaboration with CGI France ([HAL](https://hal.science/hal-05757598)).
- Sep 2026: Talk at the Semaine de l'IA of Université de Rennes: *Langages de programmation à l'ère de l'IA générative : COBOL, M&M's, français… ou peu importe ?* ([video](https://www.youtube.com/watch?v=Aj5arTQzMQ4), [slides](https://blog.mathieuacher.com/languages-codingagents-semaineIA2026/)).
- Jun 2026: Georges Aaron Randrianaina received the **GDR SciLog 2026 thesis prize** for *Incremental, Reproducible Builds of Software Variants* ([news](https://www.irisa.fr/en/news/2026-06/congratulations-georges-aaron-randrianaina-who-has-been-awarded-gdr-scilog-software)). Congratulations!
- Jun 2026: Preprint: *Do Programming Languages Still Matter to Your AI Coding Agent Teammate? Evidence at Scale from Chess Engines*, with Jean-Marc Jézéquel: 34 engines, 17 languages, 2 coding agents ([arXiv](https://arxiv.org/abs/2606.13763)).
- Mar 2026: Blog series on coding agents writing chess engines from scratch in Rust, C++, COBOL, Rocq, LaTeX ([TeXCCChess](https://blog.mathieuacher.com/TeXCCChessEngine/)), Brainfuck ([BFChess](https://blog.mathieuacher.com/BFChessChessEngineBrainfuck/)), and more ([overview](https://blog.mathieuacher.com/FromScratchChessEnginesPolyglot/)).
- Jan 2026: My panel [*The Constant*](https://www.sourcecode-exhibition.softwareheritage.org/the-constant/), on the number 1070372 that has stayed unchanged in Stockfish's source code since 2014, is part of the [Source Code Exhibition](https://www.sourcecode-exhibition.softwareheritage.org/) for the 10th anniversary of Software Heritage, unveiled at UNESCO in Paris and since shown at the Cité des sciences and Sorbonne Université.
- Dec 2025: JSS article: *Piloting Copilot, Codex, and StarCoder2: Hot Temperature, Cold Prompts, or Black Magic?* ([DOI](https://doi.org/10.1016/j.jss.2025.112562)).
- Sep 2025: Keynote at SBCARS (CBSoft 2025, Brazil): *Software Variability for Replicable Science*. Three papers at SPLC 2025 on neuroimaging pipelines, feature reduction and energy, and uniform sampling.
- Sep 2025: Start of **[FRAIME](https://www.inria.fr/fr/fraime)** (*Formal Reasoning applied to AI for Methodological Engineering*), a four-year Inria challenge with Mitsubishi Electric R&D Centre Europe on trustworthy AI with formal methods, in which I am strongly involved ([press release](https://www.mitsubishielectric.com/en/pr/2025/pdf/0918.pdf)).
- Jul 2025: *Teaching Reproducibility and Embracing Variability* at ACM REP 2025, about my reproducibility course at INSA Rennes ([blog post](https://blog.mathieuacher.com/TeachingReproducibilityVariability/)).
- Jun 2025: Three papers at EASE 2025: a dataset of Linux kernel configurations at scale, LLM-based code customization, and JHipster performance.
- May 2025: Two PhD defenses: Zohra Kaouter Kebaili on metamodel and code co-evolution, and Georges Aaron Randrianaina on incremental and reproducible builds of software variants.
- Feb 2025: General chair of [VaMoS 2025](https://familiar-project.github.io/VaMoS2025/), the 19th International Working Conference on Variability Modelling of Software-Intensive Systems, in Rennes (4–6 February 2025).
- Sep 2024: **Distinguished Paper Award** at ISSTA 2024 for our replicability study of MDPFuzz, with the Inria–Simula associate team RIPOST.
- May 2024: Our experiments on large language models playing chess are featured in Monsieur Phi's video *ChatGPT rêve-t-il de cavaliers électriques ?* ([video](https://www.youtube.com/watch?v=6D1XIbkm4JE), in French).
- Apr 2024: **Best Paper Award** at AST 2024 for *Testing for Fault Diversity in Reinforcement Learning* ([press release](https://www.linkedin.com/posts/simula-research-laboratory_ast2024-simula-research-activity-7187052083521458177-dsHW/)).
- Nov 2023: Recommended for the **CNRS bronze medal** by Section 6 of the CNRS National Committee (Comité national de la recherche scientifique).
- Oct 2023: **Best Paper Award** at MODELS 2023 for *Experience in Specializing a Generic Realization Language for SPL Engineering at Airbus*, with Airbus and McGill University.
- Apr 2023: Luc Lesoil defended his PhD thesis, *Deep Software Variability for Resilient Performance Models of Configurable Systems*.
- Jan 2023: The *Handbook of Re-Engineering Software Intensive Systems into Software Product Lines* is out ([Springer](https://link.springer.com/book/10.1007/978-3-031-11686-5)): 20 chapters and 78 contributors on reverse engineering variability and re-engineering product lines and configurable systems.
- Sep 2022: I am now Full Professor at [INSA Rennes](https://www.insa-rennes.fr/), still in Rennes and in the DiverSE team.
- Jun 2022: **Best Paper Award** at ICSR 2022 ([press release](https://www.irisa.fr/article/2022-06/felicitations-lequipe-diverse-pour-le-prix-obtenu-icsr-2022)).
- May 2022: Three papers at [ICSE 2022](https://conf.researchr.org/home/icse-2022): incremental build (research and NIER tracks) and transfer learning across Linux versions (journal first).
- Feb 2022: Keynote on reproducible science and deep software variability, and two papers, at [VaMoS 2022](https://vamos2022.isti.cnr.it/).
- Dec 2021: Hugo Martin defended his PhD thesis, *Machine Learning for Performance Modelling on Colossal Software Configuration Spaces*.
- Nov 2021: I defended my HDR, *Modelling, Reverse Engineering, and Learning Software Variability* ([manuscript](https://hal.science/tel-03521806), [video](https://www.youtube.com/watch?v=w4qBwU7aavI)).
- Sep 2021: **Best Paper Award** at SPLC 2021 ([press release](https://www.irisa.fr/article/2021-09/best-research-paper-award-la-conference-acm-international-systems-and-software)).
- Jun 2021: Appointed junior member of the Institut Universitaire de France, starting October 2021 ([announcement](https://www.iufrance.fr/detail-de-lactualite/247.html)).

## Research

Software comes in many variants. Configuration options, conditional compilation, plugins, generators, input data, and now prompts multiply the possible versions of a system, and each version can behave and perform differently. Variability makes it possible to tailor software to many needs, but it also creates combinatorial spaces that no one can fully test, configure, or understand by hand. My work is about modelling, reverse engineering, testing, and learning these spaces, and about the new forms of variability that large language models bring to software engineering.

### AI and formal methods

Generative AI makes code cheap to produce but hard to trust. In [FRAIME](https://www.inria.fr/fr/fraime), a four-year Inria challenge with Mitsubishi Electric R&D Centre Europe started in 2025, we combine formal reasoning with machine learning: checking what AI produces, and letting AI help practitioners use formal methods. Two PhD theses started in DiverSE in 2026.

- [FRAIME (Inria)](https://www.inria.fr/fr/fraime)
- [Mitsubishi Electric press release (2025)](https://www.mitsubishielectric.com/en/pr/2025/pdf/0918.pdf)

### Coding agents and programming languages

Do programming languages still matter when a coding agent writes the code? I run large-scale experiments where agents build complete systems, such as chess engines in 17 languages or COBOL systems, and measure what they get right, what it costs, and where they fail.

- [Do Programming Languages Still Matter to Your AI Coding Agent Teammate? (2026)](https://arxiv.org/abs/2606.13763)
- [Coding Agents Can Develop COBOL Systems (2026)](https://inria.hal.science/hal-05743207)

### LLMs for software evolution and modernization

Legacy systems have to evolve, and LLMs are a new tool for migrating, documenting, and transforming them. With CGI France, Sopra Steria, and the Inria LLM4Code challenge, I study how reliable these tools are on real industrial code.

- [CRACS, SCAM 2026 Distinguished Paper](https://hal.science/hal-05757598)
- [Code Transformation Rule Synthesis Using LLMs (ASE 2026)](https://arxiv.org/abs/2609.03592)

### Deep variability and reproducible science

Results depend not only on software options but on inputs, compilers, libraries, operating systems, and hardware. This *deep* variability threatens the generalization of empirical results and the reproducibility of computational science, from neuroimaging pipelines to machine learning. I also teach reproducibility hands-on.

- [Embracing Deep Variability for Reproducibility and Replicability (REP 2024)](https://hal.science/hal-04582287)
- [Navigating Variability in Neuroimaging Software Pipelines (SPLC 2025)](https://hal.science/hal-05158426)

### Learning configurable systems

How fast, how big, or how buggy is a configuration? I study how to sample, measure, and learn the behaviour of configurable systems with statistical machine learning, up to the 15,000+ options of the Linux kernel, including transfer learning across versions and incremental, reproducible builds.

- [Linux Kernel Configurations at Scale (EASE 2025)](https://hal.science/hal-05063560)
- [PyroBuildS: incremental build of configurations (JSS 2026)](https://hal.science/hal-05213625)

### Testing AI-based and configurable systems

I work on testing systems whose behaviour is learned, such as reinforcement learning policies, and on testing the tools we rely on to explore configuration spaces, such as uniform random samplers. Replication studies are part of the method.

- [Fault Diversity in Reinforcement Learning Policy Testing (TOSEM 2026)](https://doi.org/10.1145/3821533)
- [Testing Uniform Random Samplers (TOSEM 2026)](https://doi.org/10.1145/3797477)

### Modelling and reverse engineering variability

Feature models are the de facto standard for documenting what can vary. I have worked on their foundations, on languages and tools to compose, slice, compare, and reason about them (FAMILIAR, UVL), and on extracting them from code, configurators, and tables. This is the topic of my PhD and HDR.

- [24 Reasons Why Variability Models Are Not Yet Universal (keynote, 2024)](https://blog.mathieuacher.com/VaMoS2024/)
- [Handbook of Re-Engineering Software Intensive Systems into Software Product Lines (2023)](https://link.springer.com/book/10.1007/978-3-031-11686-5)

## Selected publications

Full list with abstracts: https://www.mathieuacher.com/publications.md · BibTeX: https://www.mathieuacher.com/acher.bib

- **Code Transformation Rule Synthesis Using LLMs: Potential and Limits.** Axel Allain, Aymeric Blot, Djamel Eddine Khelladi, and Mathieu Acher. *41st IEEE/ACM International Conference on Automated Software Engineering (ASE 2026)*, 2026. [PDF](https://hal.science/hal-05607247v2/file/Code_Transformation_Rule_Synthesis_Using_LLMs__Potential_and_Limits-3.pdf) · [HAL](https://hal.science/hal-05607247) · [DOI](https://doi.org/10.1145/3832783.3837504) · [arXiv](https://arxiv.org/abs/2609.03592)

- **CRACS: Question-Driven Assessment of Legacy Code Summaries in Industrial IBM-i Modernization.** N'Guessan Hermann Kouadio, Mathieu Acher, Olivier Barais, and Jean-Marc Jézéquel. *SCAM 2026 - 26th IEEE International Conference on Source Code Analysis and Manipulation (Engineering Track), co-located with ICSME 2026*, 2026. **Distinguished Paper Award.** [PDF](https://hal.science/hal-05757598v1/file/CRACS_scam2026_engineering_track_paper.pdf) · [HAL](https://hal.science/hal-05757598)

- **Externalized CPDAG Summaries Improve LLM Causal Deduction.** Wentao Sun, João Paulo Nogueira, Dominique Verchère, Mathieu Acher, and Alonso Silva. *NeurIPS 2026 - 40th Annual Conference on Neural Information Processing Systems*, 2026. [arXiv](https://arxiv.org/abs/2609.31071)

- **Policy Testing with MDPFuzz (Replicability Study).** Quentin Mazouni, Helge Spieker, Arnaud Gotlieb, and Mathieu Acher. *Proceedings of the 33rd ACM SIGSOFT International Symposium on Software Testing and Analysis*, 2024. **Distinguished Paper Award.** [DOI](https://doi.org/10.1145/3650212.3680382) · [arXiv](https://arxiv.org/abs/2502.19116)

- **Testing for Fault Diversity in Reinforcement Learning.** Quentin Mazouni, Helge Spieker, Arnaud Gotlieb, and Mathieu Acher. *Proceedings of the 5th ACM/IEEE International Conference on Automation of Software Test (AST 2024), Lisbon, Portugal, April 15-16, 2024*, 2024. **Best Paper Award.** [DOI](https://doi.org/10.1145/3644032.3644458) · [arXiv](https://arxiv.org/abs/2403.15065)

- **Experience in Specializing a Generic Realization Language for SPL Engineering at Airbus.** Damien Foures, Mathieu Acher, Olivier Barais, Benoit Combemale, Jean-Marc Jézéquel, and Jörg Kienzle. *MODELS 2023 - 26th International Conference on Model-Driven Engineering Languages and Systems*, 2023. **Best Paper Award.** [PDF](https://inria.hal.science/hal-04216627/file/Experience_in_Specializing_a_Generic_Realization_Language_for_SPL_Engineering_at_Airbus%20%285%29.pdf) · [HAL](https://hal.science/hal-04216627) · [DOI](https://doi.org/10.1109/models58315.2023.00035)

- **Input Sensitivity on the Performance of Configurable Systems: An Empirical Study.** Luc Lesoil, Mathieu Acher, Arnaud Blouin, and Jean-Marc Jézéquel. *Journal of Systems and Software, 201*, 2023. [PDF](https://inria.hal.science/hal-03476464v2/file/main.pdf) · [HAL](https://hal.science/hal-03476464) · [DOI](https://doi.org/10.1016/j.jss.2023.111671)

- **Handbook of Re-Engineering Software Intensive Systems into Software Product Lines.** Roberto Lopez-Herrejon, Jabier Martinez, Tewfik Ziadi, Mathieu Acher, K.G. Wesley Assunção, and Silvia Vergilio (eds.). *Springer International Publishing*, 2023. [HAL](https://hal.science/hal-03885628) · [DOI](https://doi.org/10.1007/978-3-031-11686-5) · [Thread](https://twitter.com/wesleyklewerton/status/1597174694259888128)

- **Scratching the Surface of ./configure: Learning the Effects of Compile-Time Options on Binary Size and Gadgets.** Xhevahire Tërnava, Mathieu Acher, Luc Lesoil, Arnaud Blouin, and Jean-Marc Jézéquel. *ICSR 2022 - 20th International Conference on Software and Systems Reuse*, 2022. **Best Paper Award.** [PDF](https://hal.science/hal-03627246/file/paper32-ICSR2022_v03.pdf) · [HAL](https://hal.science/hal-03627246) · [DOI](https://doi.org/10.1007/978-3-031-08129-3_3)

- **On the Benefits and Limits of Incremental Build of Software Configurations: An Exploratory Study.** Georges Aaron Randrianaina, Xhevahire Tërnava, Djamel Eddine Khelladi, and Mathieu Acher. *ICSE 2022 - 44th International Conference on Software Engineering*, 2022. [PDF](https://hal.science/hal-03547219v2/file/ICSE22_Incremental_Build_HAL.pdf) · [HAL](https://hal.science/hal-03547219) · [DOI](https://doi.org/10.1145/3510003.3510190)

- **Towards Incremental Build of Software Configurations.** Georges Aaron Randrianaina, Djamel Eddine Khelladi, Olivier Zendra, and Mathieu Acher. *ICSE-NIER 2022 - 44th International Conference on Software Engineering – New Ideas and Emerging Results*, 2022. [PDF](https://hal.science/hal-03558479/file/ICSE22_NIER_HAL.pdf) · [HAL](https://hal.science/hal-03558479) · [DOI](https://doi.org/10.1145/3510455.3512792)

- **Transfer Learning Across Variants and Versions: The Case of Linux Kernel Size.** Hugo Martin, Mathieu Acher, Juliana Alves Pereira, Luc Lesoil, Jean-Marc Jézéquel, and Djamel Eddine Khelladi. *IEEE Transactions on Software Engineering, 48(11)*, 2022. [PDF](https://hal.science/hal-03358817/file/TransferLinux__hal.pdf) · [HAL](https://hal.science/hal-03358817) · [DOI](https://doi.org/10.1109/tse.2021.3116768) · [Thread](https://twitter.com/acherm/status/1446399342236078080)

- **A comparison of performance specialization learning for configurable systems.** Hugo Martin, Mathieu Acher, Juliana Alves Pereira, and Jean-Marc Jézéquel. *SPLC 2021 - 25th ACM International Systems and Software Product Line Conference*, 2021. **Best Paper Award.** [PDF](https://hal.science/hal-03335263/file/PerformanceSpecialization.pdf) · [HAL](https://hal.science/hal-03335263) · [DOI](https://doi.org/10.1145/3461001.3471155) · [Video](https://www.youtube.com/watch?v=DC6GeqISf6E) · [Thread](https://twitter.com/acherm/status/1435611528154189828)

- **Learning Software Configuration Spaces: A Systematic Literature Review.** Juliana Alves Pereira, Hugo Martin, Mathieu Acher, Jean-Marc Jézéquel, Goetz Botterweck, and Anthony Ventresque. *Journal of Systems and Software, 182*, 2021. [PDF](https://hal.science/hal-02148791v2/file/Survey_MachineLearningConfiguration_RevisedJSS.pdf) · [HAL](https://hal.science/hal-02148791) · [DOI](https://doi.org/10.1016/j.jss.2021.111044) · [arXiv](https://arxiv.org/abs/1906.03018) · [Thread](https://twitter.com/acherm/status/1445731364888666117)

- **Empirical Assessment of Multimorphic Testing.** Paul Temple, Mathieu Acher, and Jean-Marc Jézéquel. *IEEE Transactions on Software Engineering (TSE), 47(7)*, 2021. [PDF](https://hal.science/hal-02177158/file/Leveraging_performance_variations_with_Multimorphic_Testing-TempleAcherJezequel-TSE.pdf) · [HAL](https://hal.science/hal-02177158) · [DOI](https://doi.org/10.1109/TSE.2019.2926971)

- **Deep Software Variability: Towards Handling Cross-Layer Configuration.** Luc Lesoil, Mathieu Acher, Arnaud Blouin, and Jean-Marc Jézéquel. *VaMoS 2021 - 15th International Working Conference on Variability Modelling of Software-Intensive Systems*, 2021. [PDF](https://hal.science/hal-03084276v2/file/vision_paper_deep_software_variability.pdf) · [HAL](https://hal.science/hal-03084276) · [DOI](https://doi.org/10.1145/3442391.3442402) · [Thread](https://twitter.com/lesoil_l/status/1489062896977989635)

- **Co-evolving code with evolving metamodels.** Djamel Eddine Khelladi, Benoit Combemale, Mathieu Acher, Olivier Barais, and Jean-Marc Jézéquel. *ICSE '20: 42nd International Conference on Software Engineering, Seoul, South Korea, 27 June - 19 July, 2020*, 2020. [PDF](https://inria.hal.science/hal-03029429v1/file/temp_icse.pdf) · [HAL](https://hal.science/hal-03029429) · [DOI](https://doi.org/10.1145/3377811.3380324) · [Video](https://www.youtube.com/watch?v=zqiG-tEDJKk)

- **Sampling Effect on Performance Prediction of Configurable Systems: A Case Study.** Juliana Alves Pereira, Mathieu Acher, Hugo Martin, and Jean-Marc Jézéquel. *International Conference on Performance Engineering (ICPE 2020)*, 2020. **Best Paper Award.** [PDF](https://inria.hal.science/hal-02356290v3/file/ICPE2020_x264_withAppendix.pdf) · [HAL](https://hal.science/hal-02356290) · [DOI](https://doi.org/10.1145/3358960.3379137) · [Video](https://www.youtube.com/watch?v=jtXoYdHhVtI)

- **Test them all, is it worth it? Assessing configuration sampling on the JHipster Web development stack.** Axel Halin, Alexandre Nuttinck, Mathieu Acher, Xavier Devroey, Gilles Perrouin, and Benoit Baudry. *Empirical Software Engineering (ESE), 24(2)*, 2019. [PDF](https://inria.hal.science/hal-01829928v1/file/10.1007%252Fs10664-018-9635-4.pdf) · [HAL](https://hal.science/hal-01829928) · [DOI](https://doi.org/10.1007/s10664-018-9635-4) · [arXiv](https://arxiv.org/abs/1710.07980) · [Video](https://www.youtube.com/watch?v=mjBmarVDBBo)

- **Learning Contextual-Variability Models.** Paul Temple, Mathieu Acher, Jean-Marc Jézéquel, and Olivier Barais. *IEEE Software, 34(6)*, 2017. [PDF](https://inria.hal.science/hal-01659137v1/file/Learning_Contextual_Variability_Models.pdf) · [HAL](https://hal.science/hal-01659137) · [DOI](https://doi.org/10.1109/ms.2017.4121211)

- **Efficient and Complete FD-Solving for Extended Array Constraints.** Quentin Plazar, Mathieu Acher, Sébastien Bardin, and Arnaud Gotlieb. *IJCAI 2017*, 2017. [PDF](https://hal.science/hal-01545557/file/camera_ready.pdf) · [HAL](https://hal.science/hal-01545557) · [DOI](https://doi.org/10.24963/ijcai.2017/171)

- **Learning-Based Performance Specialization of Configurable Systems.** Paul Temple, Mathieu Acher, Jean-Marc Jézéquel, Léo Noel-Baron, and José A. Galindo. *Research Report, IRISA, Inria Rennes ; University of Rennes 1*, 2017. [PDF](https://hal.science/hal-01467299/file/TSE-ML.pdf) · [HAL](https://hal.science/hal-01467299)

- **Using Machine Learning to Infer Constraints for Product Lines.** Paul Temple, José Angel Galindo Duarte, Mathieu Acher, and Jean-Marc Jézéquel. *Software Product Line Conference (SPLC'16)*, 2016. **Most Influential Paper Award at VARIABILITY 2026.** [PDF](https://hal.science/hal-01323446/file/SPLC_2016_Paper.pdf) · [HAL](https://hal.science/hal-01323446) · [DOI](https://doi.org/10.1145/2934466.2934472)

- **Large-scale Analysis of Chess Games with Chess Engines: A Preliminary Report.** Mathieu Acher and François Esnault. *Technical Report RT-0479, Inria Rennes Bretagne Atlantique*, 2016. [PDF](https://hal.science/hal-01307091/file/RT-479%20%281%29.pdf) · [HAL](https://hal.science/hal-01307091)

- **Breathing Ontological Knowledge Into Feature Model Synthesis: An Empirical Study.** Guillaume Bécan, Mathieu Acher, Benoit Baudry, and Sana Ben Nasr. *Empirical Software Engineering (ESE), 21(4)*, 2016. [PDF](https://inria.hal.science/hal-01096969v1/file/ESE-KSynthesis.pdf) · [HAL](https://hal.science/hal-01096969) · [DOI](https://doi.org/10.1007/s10664-014-9357-1)

- **Synthesis of Attributed Feature Models From Product Descriptions.** Guillaume Bécan, Razieh Behjati, Arnaud Gotlieb, and Mathieu Acher. *19th International Software Product Line Conference (SPLC'15)*, 2015. [PDF](https://inria.hal.science/hal-01178454v1/file/SPLC15-AFM.pdf) · [HAL](https://hal.science/hal-01178454) · [DOI](https://doi.org/10.1145/2791060.2791068) · [Slides](https://fr.slideshare.net/acher/synthesis-of-attributed-feature-models-from-product-descriptions)

- **Assessing Product Line Derivation Operators Applied to Java Source Code: An Empirical Study.** João Bosco Ferreira Filho, Simon Allier, Mathieu Acher, Olivier Barais, and Benoit Baudry. *19th International Software Product Line Conference (SPLC'15)*, 2015. [PDF](https://inria.hal.science/hal-01163423v1/file/VaryJavaSPLC2015.pdf) · [HAL](https://hal.science/hal-01163423) · [DOI](https://doi.org/10.1145/2791060.2791099) · [Slides](https://fr.slideshare.net/acher/assessing-product-line-derivation-operators-applied-to-java-source-code-an-empirical-study)

- **Automating the Formalization of Product Comparison Matrices.** Guillaume Bécan, Nicolas Sannier, Mathieu Acher, Olivier Barais, Arnaud Blouin, and Benoit Baudry. *29th IEEE/ACM International Conference on Automated Software Engineering (ASE'14)*, 2014. [PDF](https://inria.hal.science/hal-01058440v1/file/FromData2Models.pdf) · [HAL](https://hal.science/hal-01058440) · [DOI](https://doi.org/10.1145/2642937.2643000) · [Slides](https://fr.slideshare.net/gbecan/ase2014-presentation)

- **Generating Counterexamples of Model-based Software Product Lines.** Joao Bosco Ferreira Filho, Olivier Barais, Mathieu Acher, Jérôme Le Noir, Axel Legay, and Benoit Baudry. *International Journal on Software Tools for Technology Transfer (STTT), 17(5)*, 2014. [PDF](https://inria.hal.science/hal-01026581v1/file/STTT2014.pdf) · [HAL](https://hal.science/hal-01026581) · [DOI](https://doi.org/10.1007/s10009-014-0341-2)

- **Sound Merging and Differencing for Class Diagrams.** Uli Fahrenberg, Mathieu Acher, Axel Legay, and Andrzej Wasowski. *17th International Conference on Fundamental Approaches to Software Engineering (FASE'14)*, 2014. [PDF](https://inria.hal.science/hal-01087323v1/file/fase-final.pdf) · [HAL](https://hal.science/hal-01087323) · [DOI](https://doi.org/10.1007/978-3-642-54804-8_5) · [Link](https://hal.science/hal-00927310)

- **Software Architectures and Multiple Variability.** Mathieu Acher, Philippe Collet, and Philippe Lahire. *Chapter in Software Architecture 2, Wiley*, 2014. [HAL](https://hal.science/hal-01098107)

- **A Variability-based Testing Approach for Synthesizing Video Sequences.** José A. Galindo, Mauricio Alferez, Mathieu Acher, Benoit Baudry, and David Benavides. *International Symposium on Software Testing and Analysis (ISSTA'14)*, 2014. [PDF](https://inria.hal.science/hal-01003148v3/file/paper.pdf) · [HAL](https://hal.science/hal-01003148) · [DOI](https://doi.org/10.1145/2610384.2610411)

- **The Anatomy of a Sales Configurator: An Empirical Study of 111 Cases.** Ebrahim Khalil Abbasi, Arnaud Hubaux, Mathieu Acher, Quentin Boucher, and Patrick Heymans. *25th International Conference on Advanced Information Systems Engineering (CAiSE'13)*, 2013. [PDF](https://inria.hal.science/hal-00796555v1/file/ConfiguratorsCAISE13.pdf) · [HAL](https://hal.science/hal-00796555) · [DOI](https://doi.org/10.1007/978-3-642-38709-8_11)

- **From Comparison Matrix to Variability Model: The Wikipedia Case Study.** Nicolas Sannier, Mathieu Acher, and Benoit Baudry. *28th IEEE/ACM International Conference on Automated Software Engineering (ASE'13)*, 2013. [PDF](https://inria.hal.science/hal-00858491v1/file/ASE2013_Sannier_Acher_Baudry.pdf) · [HAL](https://hal.science/hal-00858491) · [DOI](https://doi.org/10.1109/ase.2013.6693116) · [Slides](https://fr.slideshare.net/acher/product-comparison-matrix-pcm-and-wikipedia)

- **Composing your Compositions of Variability Models.** Mathieu Acher, Benoit Combemale, Philippe Collet, Olivier Barais, Philippe Lahire, and Robert B. France. *ACM/IEEE 16th International Conference on Model Driven Engineering Languages and Systems (MODELS'13)*, 2013. [PDF](https://inria.hal.science/hal-00859473v1/file/models13-compofm.pdf) · [HAL](https://hal.science/hal-00859473) · [DOI](https://doi.org/10.1007/978-3-642-41533-3_22)

- **Feature Model Extraction from Large Collections of Informal Product Descriptions.** Jean-Marc Davril, Edouard Delfosse, Negar Hariri, Mathieu Acher, Jane Cleland-Huang, and Patrick Heymans. *European Software Engineering Conference and the ACM SIGSOFT Symposium on the Foundations of Software Engineering (ESEC/FSE'13)*, 2013. [PDF](https://inria.hal.science/hal-00859475v1/file/fse13main-id221-p-18686-final.pdf) · [HAL](https://hal.science/hal-00859475) · [DOI](https://doi.org/10.1145/2491411.2491455)

- **Generating Counterexamples of Model-based Software Product Lines: An Exploratory Study.** João Bosco Ferreira Filho, Olivier Barais, Mathieu Acher, Jérôme Le Noir, and Benoit Baudry. *17th International Software Product Line Conference (SPLC'13)*, 2013. **Best Student Paper Award.** [PDF](https://inria.hal.science/hal-00837523v1/file/SPLC2013.pdf) · [HAL](https://hal.science/hal-00837523) · [DOI](https://doi.org/10.1145/2491627.2491639)

- **Extraction and Evolution of Architectural Variability Models in Plugin-based Systems.** Mathieu Acher, Anthony Cleve, Philippe Collet, Philippe Merle, Laurence Duchien, and Philippe Lahire. *Software and Systems Modeling (SoSyM), 13(4)*, 2013. [PDF](https://inria.hal.science/hal-00859472v1/file/FraSCAtiSoSyM12.pdf) · [HAL](https://hal.science/hal-00859472) · [DOI](https://doi.org/10.1007/s10270-013-0364-2)

- **FAMILIAR: A Domain-Specific Language for Large Scale Management of Feature Models.** Mathieu Acher, Philippe Collet, Philippe Lahire, and Robert France. *Science of Computer Programming (SCP) Special issue on programming languages, 78(6)*, 2013. [PDF](https://inria.hal.science/hal-00767175v1/file/SCP-FAMILIAR.pdf) · [HAL](https://hal.science/hal-00767175) · [DOI](https://doi.org/10.1016/j.scico.2012.12.004)

- **Separating Concerns in Feature Models: Retrospective and Support for Multi-Views.** Arnaud Hubaux, Mathieu Acher, T. T. Tun, Patrick Heymans, Philippe Collet, and Philippe Lahire. *Chapter in Domain Engineering: Product Lines, Conceptual Models, and Languages (editors: Reinhartz-Berger,I. and Sturm, A. and Clark, T. and Bettin, J. and Cohen, S.), Springer*, 2013. [PDF](https://inria.hal.science/hal-00767213v1/file/DESoCFMBook.pdf) · [HAL](https://hal.science/hal-00767213) · [DOI](https://doi.org/10.1007/978-3-642-36654-3_1)

- **Composing Multiple Variability Artifacts to Assemble Coherent Workflows.** Mathieu Acher, Philippe Collet, Alban Gaignard, Philippe Lahire, Johan Montagnat, and Robert France. *Software Quality Journal Special issue on Quality Engineering for Software Product Lines, 20(3-4)*, 2012. [PDF](https://hal.science/hal-00733556v1/file/SQJO-SPL-2011-final.pdf) · [HAL](https://hal.science/hal-00733556) · [DOI](https://doi.org/10.1007/s11219-011-9170-7)

- **Feature Model Differences.** Mathieu Acher, Patrick Heymans, Philippe Collet, Clément Quinton, Philippe Lahire, and Philippe Merle. *24th International Conference on Advanced Information Systems Engineering (CAiSE'12)*, 2012. [PDF](https://inria.hal.science/hal-00713849v2/file/CAiSE2012Diff.pdf) · [HAL](https://hal.science/hal-00713849) · [DOI](https://doi.org/10.1007/978-3-642-31095-9_41)

- **Separation of Concerns in Feature Modeling: Support and Applications.** Mathieu Acher, Philippe Collet, Philippe Lahire, and Robert France. *Aspect-Oriented Software Development (AOSD'12)*, 2012. [PDF](https://inria.hal.science/hal-00767423v1/file/acher-collet-etal-2012.pdf) · [HAL](https://hal.science/hal-00767423) · [DOI](https://doi.org/10.1145/2162049.2162051)

- **Slicing Feature Models.** Mathieu Acher, Philippe Collet, Philippe Lahire, and Robert France. *26th IEEE/ACM International Conference On Automated Software Engineering (ASE'11), short paper*, 2011. [DOI](https://doi.org/10.1109/ASE.2011.6100089) · [Link](https://dl.acm.org/citation.cfm?id=2190091) · [Slides](https://fr.slideshare.net/acher/ase11-short-paper)

- **Reverse Engineering Architectural Feature Models.** Mathieu Acher, Anthony Cleve, Philippe Collet, Philippe Merle, Laurence Duchien, and Philippe Lahire. *5th European Conference on Software Architecture (ECSA'11), long paper*, 2011. [PDF](https://inria.hal.science/inria-00614984v1/file/ecsa2011_submission_38_1_.pdf) · [HAL](https://hal.science/inria-00614984) · [DOI](https://doi.org/10.1007/978-3-642-23798-0_25) · [Slides](https://fr.slideshare.net/acher/reverse-engineering-architectural-feature-models)

- **Composing Feature Models.** Mathieu Acher, Philippe Collet, Philippe Lahire, and Robert France. *2nd International Conference on Software Language Engineering (SLE'09)*, 2009. **Most Influential Paper Award at SLE 2019.** [PDF](https://hal.science/hal-00415767v1/file/SLE2009-postproceedings.pdf) · [HAL](https://hal.science/hal-00415767) · [DOI](https://doi.org/10.1007/978-3-642-12107-4_6) · [Slides](https://fr.slideshare.net/acher/composing-feature-models)

## Software

- **[Agentic chess engines](https://github.com/acherm/agentic-chessengine-metaanalysis)** (active): Chess engines written from scratch by coding agents (Claude Code, Codex) in 17+ languages, from Rust and Java to COBOL, Rocq, Lean, SQL, LaTeX ([TeXCCChess](https://github.com/acherm/agentic-chessengine-latex-TeXCCChess)), and Brainfuck ([BFChess](https://github.com/acherm/agentic-chessengine-brainfuck)), with an Elo evaluation harness. 
- **[UVL Studio](https://blog.mathieuacher.com/uvl-studio/)** (active): A self-contained web workbench for the Universal Variability Language: text editor, live checks, feature diagram, and SAT-based interactive configuration ([source](https://github.com/acherm/uvl-studio)). 
- **[PlantUML Studio](https://blog.mathieuacher.com/plantuml-studio/)** (active): An offline, single-file PlantUML editor for teaching UML, with live well-formedness checks and "did you mean" diagnostics ([source](https://github.com/acherm/plantuml-studio)). 
- **[Programming languages catalog](https://blog.mathieuacher.com/PL-ultimate-llm/)** (active): A catalog of 14,000+ programming languages merged from PLDB, Linguist, Pygments, Wikipedia, and Rosetta Code, with evidence from Software Heritage (CodeCommons project). 
- **[gptchess](https://github.com/acherm/gptchess)**: Experiments pitting GPT models against Stockfish: Elo estimates and legal-move abilities of large language models ([blog post](https://blog.mathieuacher.com/GPTsChessEloRatingLegalMoves/)). 
- **[FAMILIAR](https://familiar-project.github.io/)**: A domain-specific language for managing feature models: compose, decompose, diff, refactor, configure, and reason about several feature models with SAT solvers and BDDs. Migrated to modern Java with a coding agent in 2026. 
- **[TuxML](https://github.com/TuxML)** (archived): Sampling, building, measuring, and learning Linux kernel configurations at scale (build failures, kernel size), including transfer learning across kernel versions ([talk at ELC Europe 2019](https://youtu.be/UBghs-cwQX4)). 
- **[OpenCompare](https://github.com/OpenCompare)** (archived): Tools for product comparison matrices: editors, miners (MatrixMiner), visualizers, and configurators. No longer maintained, but we learned a lot. 

## Teaching

I teach software engineering, mostly at [INSA Rennes](https://www.insa-rennes.fr/) in the computer science curriculum, and since 2024 I co-head the research Master SIF in Rennes (with Alexandre Termier). Recent courses:

- **Modelling, verification, advanced programming, and variability** (CPOO1, 3rd year; CPOO2, 4th year): several formalisms to specify software (from class diagrams to feature models), to test it (JUnit, Mockito, mutation testing), and to vary it through time and space. Students apply them in Java, with design patterns, on the ChessBall project (student workspaces: [CPOO1](https://github.com/acherm/cpoo1-etd), [CPOO2](https://github.com/acherm/cpoo2-etd)). Redesigned in 2026–27 along the lines of Carnegie Mellon's 17-214 course, with a guided and critical use of AI coding assistants.
- **Reproducibility of computational experiments** (4th and 5th year, since 2024): a hands-on course going from floating-point non-associativity, Docker, and continuous integration to reproducing and replicating published studies (LLMs playing chess, home advantage in football, energy of programming languages). See the [ACM REP 2025 paper](https://hal.science/hal-05190848) and the [course repository](https://github.com/acherm/REP-INSA2526).
- **Domain-specific languages** (research Master SIF, Rennes; INSA Rennes since 2022): *The Art of Domain-Specific Languages: Let's Hack Our Own Languages!* ([materials](https://github.com/FAMILIAR-project/HackOurLanguages-SIF), [Langium demonstrators](https://github.com/acherm/dsl-langium)). At INSA, students develop DSLs for families of games, including services to play with AI and LLMs.

From 2012 to 2022, I taught at the University of Rennes 1 ([ISTIC](https://istic.univ-rennes.fr/) and ESIR): model-driven engineering, domain-specific languages, software development projects, software design and modelling, and Scala programming, and also at [ENSAI](https://ensai.fr/). I was head of the Master 1 in Software Engineering at ISTIC, coordinator of the Bachelor (L3) internships and projects (2018–2021, 100+ students a year), in charge of teaching services at ISTIC (2014–2017), and a jury member of the *Agrégation d'informatique* (2022).

Beyond the classroom: lectures at the EJCP summer school for PhD students (2018, 2019, 2021, and 2023 on reproducible science and software engineering), a course on AI and sport for the M2 APAS and EOPS at the Université de Toulon (UFR STAPS, 2024–25), tutorials at ASE, SPLC, MODELS, and ECOOP/ECSA, and [teaching.variability.io](https://teaching.variability.io/), a shared repository of teaching material on variability and configurable systems (see also the SPLTea workshops).

## Outreach

I joined the IUF in 2021 on its then-new chair in science outreach (*Médiation scientifique*). Chess is my main way in: everyone knows the game, its rules give an objective referee, and it lets anyone check, rather than take on faith, what generative AI can and cannot do. My [blog](https://blog.mathieuacher.com) publishes these experiments with code and data.

- **ChatGPT rêve-t-il de cavaliers électriques ?** (Monsieur Phi (YouTube, in French), 2024-05): Our experiments on large language models playing chess ([blog post](https://blog.mathieuacher.com/GPTsChessEloRatingLegalMoves/)). https://www.youtube.com/watch?v=6D1XIbkm4JE
- **The Constant** (Source Code Exhibition, Software Heritage and Inria, 2026-01): A panel on the number 1070372, unchanged in Stockfish's source code since 2014. The exhibition opened at UNESCO in Paris for Software Heritage's 10th anniversary and now travels to other cities and countries (Cité des sciences and Sorbonne Université in Paris, Luxembourg, and more). https://www.sourcecode-exhibition.softwareheritage.org/the-constant/
- **Langages de programmation à l'ère de l'IA générative : COBOL, M&M's, français… ou peu importe ?** (Semaine de l'IA, Université de Rennes, 2026-09): What coding agents change for programming languages, for about 230 students ([slides](https://blog.mathieuacher.com/languages-codingagents-semaineIA2026/)). https://www.youtube.com/watch?v=Aj5arTQzMQ4
- **Un ordinateur, ça ne calcule jamais juste** (Fête de la science, INSA Rennes, 2024-10): Why computers never compute exactly, and why it matters; also a chapter with Jean-Marc Jézéquel in [*Le calcul à découvert*](/publications/#acher2025ordinateur) (CNRS Éditions, 2025). https://hal.science/hal-04736121
- **CodeplAI: IA, Code & Jeux** (YouTube (in French), 2026): AI, programming, and games: puzzles, algorithms, variants, playing with and against AI. https://www.youtube.com/@CodeplAI
- **In the media** (interviews and articles): *Le Parisien* (2026), *Science & Vie* (2023), [CNRS News](https://news.cnrs.fr/articles/one-software-billions-of-possibilities) (2022).

## Students, projects & service

I supervise PhD students, postdocs, and engineers in the DiverSE team, on variability, machine learning and LLMs for software engineering, reproducibility, and testing, often with industrial partners (CGI France, Sopra Steria, Nokia, Mitsubishi Electric, Airbus). I serve the software product line and variability community (SPLC, VaMoS, and VARIABILITY steering committees, VaMoS 2025 general chair), co-lead the Inria challenge LLM4Code and the Inria–Simula associate team RIPOST, and am strongly involved in the Inria–Mitsubishi Electric challenge FRAIME and the France 2030 project CodeCommons.

Details: https://www.mathieuacher.com/service.md

## Theses

### HDR (Habilitation à diriger des recherches) (2021): Modelling, Reverse Engineering, and Learning Software Variability

University of Rennes 1, defended on 16 November 2021
[Manuscript (HAL)](https://hal.science/tel-03521806) · [Video of the defense](https://www.youtube.com/watch?v=w4qBwU7aavI)

Committee: Krzysztof Czarnecki (University of Waterloo, Canada), reviewer; Ina Schaefer (Technische Universität Braunschweig, Germany), reviewer; Romain Rouvoy (University of Lille, IUF, France), reviewer; Julia Lawall (Inria, France), examiner; Christian Kästner (Carnegie Mellon University, USA), examiner; Jean-Marc Jézéquel (University of Rennes 1, France), examiner.

Abstract: Society expects software to deliver the right functionality, in a short amount of time and with fewer resources, in every possible circumstance whatever are the hardware, the operating systems, the compilers, or the data fed as input. For fitting such a diversity of needs, it is common that software comes in many variants and is highly configurable through configuration options, runtime parameters, conditional compilation directives, menu preferences, configuration files, plugins, etc. As there is no one-size-fits-all solution, software variability ("the ability of a software system or artifact to be efficiently extended, changed, customized or configured for use in a particular context") has been studied the last two decades and is a discipline of its own. Though highly desirable, software variability also introduces an enormous complexity due to the combinatorial explosion of possible variants. For example, the Linux kernel has 15000+ options and most of them can have 3 values: "yes", "no", or "module". Variability is challenging for maintaining, verifying, and configuring software systems (Web applications, Web browsers, video tools, etc.). It is also a source of opportunities to better understand a domain, create reusable artefacts, deploy performance-wise optimal systems, or find specialized solutions to many kinds of problems. In many scenarios, a model of variability is either beneficial or mandatory to explore, observe, and reason about the space of possible variants. For instance, without a variability model, it is impossible to establish a sampling strategy that would satisfy the constraints among options and meet coverage or testing criteria. I address a central question in this HDR manuscript: How to model software variability? I detail several contributions related to modelling, reverse engineering, and learning software variability.

I first contribute to support the persons in charge of manually specifying feature models, the de facto standard for modeling variability. I develop an algebra together with a language for supporting the composition, decomposition, diff, refactoring, and reasoning of feature models. I further establish the syntactic and semantic relationships between feature models and product comparison matrices, a large class of tabular data. I then empirically investigate how these feature models can be used to test in the large configurable systems with different sampling strategies. Along with this effort, I report on the attempts and lessons learned when defining the "right" variability language. From a reverse engineering perspective, I contribute to synthesize variability information into models and from various kinds of artefacts. I develop foundations and methods for reverse engineering feature models from satisfiability formulae, product comparison matrices, dependencies files and architectural information, and from Web configurators. I also report on the degree of automation and show that the involvement of developers and domain experts is beneficial to obtain high-quality models. Thirdly, I contribute to learning constraints and non-functional properties (performance) of a variability-intensive system. I describe a systematic process "sampling, measuring, learning" that aims to enforce or augment a variability model, capturing variability knowledge that domain experts can hardly express. I show that supervised, statistical machine learning can be used to synthesize rules or build prediction models in an accurate and interpretable way. This process can even be applied to huge configuration space, such as the Linux kernel one.

Despite wide applicability and observed benefits, I show that each individual line of contributions has limitations. I defend the following answer: a supervised, iterative process (1) based on the combination of reverse engineering, modelling, and learning techniques; (2) capable of integrating multiple variability information (eg expert knowledge, legacy artefacts, dynamic observations).

Finally, this work opens different perspectives related to so-called deep software variability, security, smart build of configurations, and (threats to) science.

### PhD thesis (2011): Managing Multiple Feature Models: Foundations, Language, and Applications

University of Nice Sophia Antipolis, I3S laboratory (Modalis team). Supervised by Philippe Collet and Philippe Lahire
[Manuscript (PDF)](/PhDAcher2011-revised.pdf) · [theses.fr](https://theses.fr/2011NICE4052) · [FAMILIAR](https://familiar-project.github.io/)

Committee: Krzysztof Czarnecki, Associate Professor, University of Waterloo (Canada), reviewer; Jean-Marc Jézéquel, Professor, University of Rennes 1 (France), reviewer; Robert De Simone, Research Director, INRIA Sophia Antipolis (France), examiner; Patrick Heymans, Professor, University of Namur (Belgium), examiner; Robert B. France, Professor, Colorado State University (USA), examiner; Philippe Lahire, Professor, University of Nice Sophia Antipolis, supervisor; Philippe Collet, Associate Professor, University of Nice Sophia Antipolis, supervisor.

Abstract: **Feature models** (FMs) are a fundamental formalism for specifying and reasoning about commonality and variability of software product lines (SPLs). FMs are becoming increasingly complex, handled by several stakeholders or organizations, used to describe features at various levels of abstraction and related in a variety of ways. In different contexts and application domains, maintaining a single large FM is neither feasible nor desirable. Instead, multiple FMs are now used. In this thesis, we develop theoretical foundations and practical support for managing multiple FMs. We design and develop a set of **composition** and **decomposition** operators (aggregate, merge, slice) for supporting **separation of concerns**. The operators are formally defined, implemented with a fully automated algorithm and guarantee properties in terms of sets of configurations. We show how the composition and decomposition operators can be combined together or with other reasoning and editing operators to realize complex tasks. We propose a textual language, **FAMILIAR** (for FeAture Model scrIpt Language for manIpulation and Automatic Reasoning), which provides a practical solution for managing FMs on a large scale. An SPL practitioner can combine the different operators and manipulate a restricted set of concepts (FMs, features, configurations, etc.) using a concise notation and language facilities. FAMILIAR hides implementation details (e.g., solvers) and comes with a development environment. We report various applications of the operators and usages of FAMILIAR in different domains (medical imaging, video surveillance) and for different purposes (scientific workflow design, variability modeling from requirements to runtime, reverse engineering), showing the applicability of both the operators and the supporting language. Without the new capabilities brought by the operators and FAMILIAR, some analysis and reasoning operations would not be made possible in the different case studies.

## Chess

I have been playing chess since I was 8. I discovered the game by chance in my home town, joined the Cannes chess club at 11, and had the chance to be trained by strong grandmasters there.

- [FIDE Master](https://ratings.fide.com/profile/618497) with three international master norms (2007, 2008, 2010), FIDE-rated since 1998
- Twice French youth champion (1998, 2002)
- Three World Youth Chess Championships (1998, 1999, 2002) and one European Youth Chess Championship (1998)
- Second place in the French first league with the Cannes chess club (2003)
- Gold medal in the French first league (board 6, 2009)

I no longer play at a high level, but I am always up for a [blitz](https://en.wikipedia.org/wiki/Blitz_chess) game online. Three of [my games](https://www.365chess.com/players/Mathieu_Acher) I like: a [win against Santo-Roman](https://www.365chess.com/view_game.php?g=500307) (French first league, 2000), a [win against Zhigalko](https://www.365chess.com/view_game.php?g=291633) (World Youth Championship, Greece, 2002), and a [draw against Vachier-Lagrave](https://www.365chess.com/view_game.php?g=3473459) (French first league, 2007).

Chess also shows up in my research and teaching:

- How well do large language models play chess? Elo estimates and illegal moves of GPT models against Stockfish ([gptchess](https://github.com/acherm/gptchess), [blog post](https://blog.mathieuacher.com/GPTsChessEloRatingLegalMoves/)), featured in Monsieur Phi's video *ChatGPT rêve-t-il de cavaliers électriques ?* ([YouTube](https://www.youtube.com/watch?v=6D1XIbkm4JE), in French) and in a talk for school pupils at the Fête de la science 2025 ([video](https://www.youtube.com/watch?v=TtGiT-tWdmE), in French).
- Chess engines written from scratch by coding agents in 17+ programming languages ([overview](https://blog.mathieuacher.com/FromScratchChessEnginesPolyglot/), [paper](https://arxiv.org/abs/2606.13763)).
- A replication study of metamorphic testing of chess engines ([IST 2025](https://doi.org/10.1016/j.infsof.2025.107679)) and a [large-scale analysis of chess games with Stockfish](https://hal.science/hal-01307091) (5 million games, 270 million positions).
- Object-oriented design and testing courses at INSA Rennes built around the ChessBall project, and discussions about AlphaZero ([video](https://www.youtube.com/watch?v=PH06mEOVVyY), in French).

## Contact

The best way to reach me is by email: [mathieu.acher@irisa.fr](mailto:mathieu.acher@irisa.fr).

IRISA / Inria Rennes, Campus de Beaulieu, 35042 Rennes, France.