Post-Doctoral Research Visit F / M Acceleration of Full Waveform Inversion for Seismic Inversio[...]

INRIA; Pau; Sur place; EUR 60 000 - 80 000; Plein temps

Résumé du poste

Un institut de recherche en informatique recherche un chercheur postdoctoral pour développer des méthodes numériques avancées combinées à l'apprentissage machine afin d'améliorer l'inversion de formes d'onde. Le candidat retenu travaillera dans un environnement dynamique et contribuera à des projets de recherche de pointe. Un bon niveau en programmation et la maîtrise de l'anglais sont requis. Ce poste offre une rémunération de 2788€ par mois avant impôts.

Prestations

Repas subventionnés

Remboursement partiel des frais de transport public

Télétravail possible

Équipement professionnel disponible

Accès à une formation professionnelle

Couverture de sécurité sociale

Qualifications

• Compétences en programmation solides, en particulier dans les environnements HPC.
• Capacité à travailler en équipe et à présenter des résultats à des publics variés.
• Maîtrise de l'anglais à l'écrit et à l'oral.

Responsabilités

• Contribuer au développement de stratégies d'accélération pour la FWI.
• Engagement dans les activités scientifiques et collaboratives de l'équipe.
• Développement et documentation de logiciels.

Connaissances

Développement de logiciels

Collaboration et communication

Compétences en HPC

Description du poste

Contexte et atouts du poste

The Makutu project team specializes in large-scale simulations applied to the reconstruction of complex media, used to gain a better understanding of the internal dynamics of environments that are difficult or even impossible to probe. To this end, it develops advanced numerical methods that are integrated into open-source platforms deployed on state-of-the-art HPC environments.

Full-Waveform Inversion (FWI) is a powerful imaging technique that aims to recover physical parameters of a medium by minimizing the misfit between observed and simulated wavefields. It plays a central role inparticular in subsurface exploration (e.g., geophysics).

In the frequency domain , FWI involves solving large-scale wave equations at multiple frequencies. This leads to the repeated inversion of very large, often ill-conditioned, linear systems — beyond the capabilities of standard direct solvers. While iterative solvers offer an alternative, they raise significant algorithmic challenges, particularly in the multiple right-hand sides setting, which is crucial for inversion involving multiple sources. Additional strategies also have to be investigated to reduce the computational cost of wave modeling, such as reduced-order models or learning-based techniques. In both cases, the compromise between accuracy of the solutions and the computational cost has to be carefully handled.

This postdoctoral position offers a unique opportunity to combine advanced numerical methods with machine learning to accelerate and enhance FWI. The successful candidate will gain expertise in high-performance computing while working within a dynamic team involved in flagship projects such as INCORWAVE (ERC) and Exa-MA (PEPR Numpex). Strong international collaborations will provide an excellent environment for developing cutting-edge research.

Mission confiée

The successful candidate will contribute to the development and evaluation of acceleration strategies for frequency-domain FWI , by combining numerical methods, model reduction techniques, and learning-based approaches. Potential research directions include (but are not restricted to) :

• Leveraging reduced-order models (e.g., via Proper Orthogonal Decomposition, SVD, or projection-based techniques) to accelerate the direct solution phase.
• Applying machine learning techniques (e.g., neural networks, statistical models) to interpolate or predict frequency responses , enabling significant savings in frequency sampling.
• Exploiting matrix structures (e.g., block patterns, low-rank properties) in discretized wave equations.
• Designing efficient preconditioners for iterative solvers (e.g., Krylov, block Krylov methods) or any DDM inspired method
• Developing scalable multi-right-hand-side solvers adapted to inversion workflows.
• Exploring hybrid approaches that combine direct solvers (possibly partially factorized) with iterative schemes.
• These developments will be validated on representative test cases in seismic imaging and helioseismic inverse problems , in collaboration with domain specialists. The computational developments and new methodologies will be implemented in the open-source software hawen (

Principales activités

The recruited postdoctoral researcher will be expected to take an active role in the team’s scientific and collaborative activities, with missions including :

• Scientific monitoring & knowledge sharing : conducting regular technology watch and maintaining an updated bibliography, shared with the team.
• Collaboration & teamwork : engaging with team members through dedicated working groups closely related to the postdoctoral topic.
• Software development & documentation : systematically documenting all software contributions to ensure reusability and accessibility for both the team and external users.
• Research dissemination : writing research reports, presenting results at international conferences, and publishing in peer-reviewed journals.
• User engagement & training : providing training for key users of the service and contributing to the animation of an active user community.
• Partnership & outreach : presenting project progress to partners, including stakeholders and funding bodies.
• Additional contributions : supporting other scientific and technical activities relevant to the project.

Compétences

Software development : solid programming skills, with good practice in HPC environments particularly valued.

Collaboration & communication : strong interpersonal skills with a focus on teamwork; ability to present results to both scientific and non-scientific audiences (including funding bodies).

Languages : fluency in English (written and spoken).

Other qualities : autonomy, scientific curiosity, initiative, and a strong commitment to collaborative research.

Avantages

• Subsidized meals
• Partial reimbursement of public transport costs
• Possibility of teleworking and flexible organization of working hours
• Professional equipment available (videoconferencing, loan of computer equipment, etc.)
• Social, cultural and sports events and activities
• Access to vocational training
• Social security coverage

Rémunération

2788€ per month before taxs