High-performance computing in curved spacetime : toward new numerical methods for cosmology (M/F)

Job Description

Thesis Subject

The advent of precision cosmology, driven by a new generation of space
telescopes and observatories such as Euclid and the Vera Rubin Observatory's LSST, is creating new challenges for computational astrophysics. Phenomena that were previously considered negligible in cosmological models are becoming central to the simulation and analysis of the Universe's large-scale structure. This is particularly true of general relativistic effects, which may leave measurable signatures both in the dynamics of matter and in the propagation of light. Yet most cosmological simulations still rely on a Newtonian treatment of gravity in an expanding Universe. Moving beyond this framework requires a fundamental rethinking of the numerical methods and software architectures that underpin current simulation codes. This PhD project aims to address some of the associated numerical challenges through an interdisciplinary approach at the intersection of computer science, applied mathematics, and computational physics. In particular, it will investigate integration, differentiation, discretization, and interpolation schemes within a unified numerical framework suited to the differentiable manifolds on which cosmological spacetimes are defined. Relativistic raytracing, applied to the accurate simulation and analysis of gravitational lensing in next-generation surveys, will provide a natural testbed for developing, testing, and validating these new methods and their high-performance implementations.
From a computational perspective, this requires solving billions of systems of ordinary differential equations in parallel on adaptively refined meshes, while performing interpolation and tensor operations at every integration step. An additional challenge arises when attempting to reconstruct the curvature of spacetime and the associated mass distribution directly from observations. This leads to an inverse problem formulated as a non-convex optimization over the space of possible lensing configurations. The simultaneous emergence of next-generation telescopes and exascale computing systems creates an unprecedented opportunity to address this problem. The proposed PhD project will therefore develop the numerical methods and high-performance implementations required to meet this challenge, drawing on the latest advances in programming and fully exploiting modern high-performance computing architectures.

Qualifications:
Master's degree (MASTER 2) or is in the process of completing one.

Skills:
Ability to present one's work in English (C2 level).

Personal Qualities:
Passion for research and experimental physics.
Ability to work as part of a team.

Your Work Environment

The PhD thesis will be conducted within the cosmology/LSST group at the Annecy
Laboratory of Particle Physics (LAPP). It will be co-supervised by Vincent Reverdy, an interdisciplinary CNRS researcher in computer science and computational cosmology at LAPP, and by Walter Boscheri (HDR) of the Mathematics Laboratory at the University of Savoie Mont Blanc (LAMA) for aspects related to numerical methods. Near-daily interactions at LAPP will ensure scientific oversight of the dissertation, facilitate discussion of research avenues, and help identify the most relevant directions.
Regular meetings with LAMA will complement this support, particularly regarding mathematical issues, the analysis of the developed numerical methods, and the study of their properties.
The doctoral student will be invited to regularly present their research work at biweekly group meetings, laboratory meetings, as well as at workshops and conferences covering the various aspects of the thesis: high-performance computing, numerical methods, and computational cosmology. An annual meeting of the thesis advisory committee will provide an opportunity to conduct a thorough review of the project's progress, assess its development, and, if necessary, address any difficulties encountered. Finally, the doctoral student will complete the training courses recommended by the doctoral school.

Constraints and risks

Short-term travel is expected both in France and abroad.

Compensation and benefits

Compensation

2300 € gross monthly

Annual leave and RTT

44 jours

Remote Working practice and compensation

Pratique et indemnisation du TT

Transport

Prise en charge à 75% du coût et forfait mobilité durable jusqu’à 300€

About the offer

About the CNRS

The CNRS is a major player in fundamental research on a global scale. The CNRS is the only French organization active in all scientific fields. Its unique position as a multi-specialist allows it to bring together different disciplines to address the most important challenges of the contemporary world, in connection with the actors of change.

CNRS

The research professions