Postdoc in the dynamics of high-dimensional disordered systems (M/F)

CNRS; France; Sur place; EUR 42 000 - 52 000; Plein temps

Avantages offerts par ce poste

Conference funding

Computing equipment

Résumé du poste

The CNRS/LPTMS postdoctoral position at Université Paris-Saclay invites a researcher to study out-of-equilibrium dynamics in high-dimensional disordered systems, connecting analytical and numerical methods to identify metastable states and low-energy pathways.

You will publish results, present at seminars, and co-supervise a Master 2 intern, with dedicated funding for conferences and computing equipment available.

Qualifications

• PhD in theoretical/statistical physics or applied mathematics or a related field.
• Strong analytical and numerical skills.
• Interest in high-dimensional systems and machine learning theory.

Responsabilités

• Develop analytical and numerical approaches to out-of-equilibrium dynamics.
• Publish results and present at group meetings, workshops, and conferences.
• Co-supervise a Master 2 intern on related topics.

Connaissances

Analytical methods

Numerical simulations

Machine learning theory

Problem solving

Formation

PhD in theoretical/statistical physics or applied mathematics

Outils

Replica methods

Dynamical mean-field theory

Description du poste

Organisation/Company CNRS Department Laboratoire de physique théorique et modèles statistiques Research Field Physics Chemistry » Computational chemistry Researcher Profile Recognised Researcher (R2) Application Deadline 9 Oct 2026 - 23:59 (UTC) Country France Type of Contract Temporary Job Status Full-time Hours Per Week 35 Offer Starting Date 1 Dec 2026 Is the job funded through the EU Research Framework Programme? Not funded by a EU programme Is the Job related to staff position within a Research Infrastructure? No

Offer Description

The postdoctoral researcher will investigate out-of-equilibrium dynamics in high-dimensional disordered systems (including models relevant to machine learning and optimization) by characterizing the fixed points (metastable states, attractors) of the governing dynamical equations and the pathways connecting them. The focus will be on the geometry of high-dimensional random landscapes: identifying atypical, yet not rare, clusters of metastable states and low-energy pathways between them, and determining when and how constrained dynamical trajectories select these structures over more typical ones.

The postdoc will (a) contribute to the research tasks, using both analytical and numerical methods, contributing to numerical validation of analytical predictions where relevant; (b) write up results for publication and present them at group meetings, workshops, and conferences; (c) co-supervise a Master 2 intern on the research topics connected to the project.

The position is hosted at LPTMS (Laboratoire de Physique Théorique et Modèles Statistiques), CNRS / Université Paris-Saclay. Valentina Ros will directly supervise the postdoc. The postdoc will likely interact closely with the collaborators in the Paris Saclay area, such as P. Urbani (IPhT), A. Rosso (LPTMS), C. Furtlehner, S. Chibbaro (LISN). The postdoc will have dedicated funding for conferences and computing equipment.

Candidates should hold a recent PhD in theoretical/statistical physics, applied mathematics, or a related quantitative field. A background in the theory of high-dimensional systems and hands-on expertise in the technical toolkit of disordered systems (replica methods, dynamical mean-field theory) are appreciated. Experience with models and questions arising in machine learning theory (loss landscapes, generalization, high-dimensional optimization) is a strong asset, given the project's connections to that area.