Marc Mézard

Marc Mézard
Name	Marc Mézard
Birth date	1964
Birth place	Lyon, France
Nationality	French
Fields	Statistical physics, theoretical physics, computer science
Workplaces	École Normale Supérieure, Université Paris-Sud, École Polytechnique, CNRS, CEA
Alma mater	École Normale Supérieure, Université Paris-Sud
Doctoral advisor	Giorgio Parisi
Known for	Replica method, spin glasses, message-passing algorithms

Marc Mézard is a French theoretical physicist and computational scientist noted for pioneering work at the interface of statistical physics, computer science, and information theory. He has made influential contributions to the theory of disordered systems, combinatorial optimization, and inference algorithms, and has held senior positions at major French institutions. Mézard's work connects strands from condensed matter physics to algorithmic aspects of artificial intelligence and error-correcting codes.

Early life and education

Born in Lyon, Mézard studied at the École Normale Supérieure and completed doctoral work at Université Paris-Sud under the supervision of Giorgio Parisi. During formative years he engaged with research groups at the Centre National de la Recherche Scientifique, collaborating with members of the Statistical Physics Group and visiting scholars from the University of Rome La Sapienza and Princeton University. His PhD training placed him in the lineage of researchers influenced by developments around the Sherrington–Kirkpatrick model, the Edwards–Anderson model, and the nascent cross-disciplinary dialogue with computer science research at institutions such as INRIA.

Academic career

Mézard held faculty and research positions at Université Pierre et Marie Curie, École Polytechnique, Université Paris-Sud, and research roles at the Commissariat à l'Énergie Atomique (CEA) and CNRS. He has been affiliated with the Laboratoire de Physique Théorique et Modèles Statistiques and participated in collaborative projects with Télécom Paris and INRIA. Mézard served on program committees for conferences organized by IEEE and ACM and lectured at summer schools sponsored by Les Houches and ICTP. His visiting appointments included periods at Cornell University and University of California, Berkeley where he interacted with researchers working on problems from the Ising model to probabilistic graphical models.

Research contributions

Mézard developed and applied advanced techniques in the statistical mechanics of disordered systems, prominently using the replica method and cavity methods to analyze spin glass models such as the Sherrington–Kirkpatrick model. He co-developed the theory of one-step replica symmetry breaking for constraint satisfaction problems related to the K-satisfiability problem and the XORSAT problem, linking phase transitions in combinatorial problems to algorithmic thresholds studied in theoretical computer science. Mézard was instrumental in adapting message-passing algorithms—belief propagation and survey propagation—from statistical physics to practical instances in error-correcting codes and compressive sensing, influencing designs at Turbo codes and LDPC codes research groups. His collaborations connected with scholars at David Richardson's community, researchers at Shannon Lab-style information theory groups, and groups working on Bayesian inference at Max Planck Institute for Intelligent Systems. Mézard's work also spans connections to neural networks theory, statistical learning theory at Massachusetts Institute of Technology, and applications in bioinformatics and computational neuroscience.

Awards and honors

Mézard's contributions have been recognized by fellowships and distinctions from institutions such as the CNRS, the École Normale Supérieure, and national research prizes awarded by French academies. He has been invited to deliver named lectures at venues including Les Houches and has been a plenary speaker at meetings of the European Physical Society and the American Physical Society. Mézard is a member of editorial boards for journals published by IOP Publishing and Elsevier and has chaired committees for awarding prizes in statistical physics and information theory.

Selected publications

Mézard authored and co-authored influential monographs and articles, often in collaboration with leading figures in statistical physics and computer science. Notable works include co-writing a renowned book on probabilistic inference and learning methods used widely in the communities around NeurIPS and ICML. He published seminal papers on survey propagation addressing the K-satisfiability problem and foundational articles linking spin glass theory to algorithmic performance in constraint satisfaction and coding theory. His articles appear in leading journals associated with Physical Review Letters, Journal of Statistical Physics, and conference proceedings of STOC and FOCS.

Teaching and mentorship

As a professor and research group leader, Mézard supervised doctoral students and postdoctoral researchers who have joined faculties and research labs at École Polytechnique, INRIA, CNRS, University of Cambridge, and Princeton University. He taught courses drawing from the literatures of statistical mechanics, information theory, and computer science at graduate programs hosted by ENS, Université Paris-Sud, and international summer schools tied to Les Houches and ICTP. His pedagogical contributions influenced curricula in probabilistic methods used at NeurIPS-oriented workshops and in training programs at CEA.

Personal life and affiliations

Mézard is active in collaborative networks spanning the European Research Council projects, French national laboratories including CEA and CNRS, and international consortia linking INRIA with industry partners. He participates in advisory panels for research funding bodies including panels convened by the European Commission and has served on committees for scientific societies such as the Société Française de Physique and the European Physical Society. Outside professional life he is known to engage with cultural institutions in Paris and to support outreach activities that connect scientific research to public audiences.

Category:French physicists Category:Theoretical physicists Category:Living people