Priority Research Programs and Equipment - PEPR
Digital for Exascale
NumPEx is a French program dedicated to Exascale
High-performance computing (HPC), high-performance data analytics (HPDA), and Artificial Intelligence (AI) pose significant challenges across scientific, societal, economic, and ethical realms. These technologies, including modeling and data analysis, are crucial decision support tools addressing societal issues and competitiveness in French research and development. Digital resources, essential across science and industry, demand high-performance hardware. HPC enables advanced modeling, while HPDA handles heterogeneous and massive data. The solution to exploding demand is the upcoming “exascale” computers, a new generation with extraordinary capabilities.
In this context, the French program Digital for Exascale (NumPEx), funded under the France 2030 Investment Plan and overseen by the CEA, CNRS and Inria, aims at designing and developing the software components that will equip future exascale machines. NumPEx will deliver Exascale-grade numerical methods, softwares, and training, allowing France to remain one of the leaders in the field. It will contribute to take bridging the gap between cutting-edge software development and application domains to prepare the major scientific and industrial application codes to fully exploit the capabilities of these machines. Application domains of the NumPEx program include, but are not limited to, weather forecasting and climate, aeronautics, automotive, astrophysics, high energy physics, material science, energy production and management, biology and health.
Targeted Projects
NumPEx brings together five projects focused on Exascale
These five projects aim to enable France and Europe to meet the challenges of Exascale by:
- Coupling high performance computing, big data analysis and artificial intelligence
- Accelerating knowledge production, integrating research, applications, hardware technologies and software technologies
Production of the Program
NumPEx HAL collection
Here you will find our scientific publications, hosted on HAL, with the latest results and breakthroughs from the NumPEx community. Whether you are a researcher, an industry professional or simply a curious person interested in exascale computing, these publications will give you a glimpse of the future of HPC.
118 publications
Efficient Fine-Scale Simulation of Nonlinear Hyperelastic Lattice Structures
Clément Guillet, Thibaut Hirschler, Pierre Jolivet, Pablo Antolin, Robin Bouclier
Computer Methods in Applied Mechanics and Engineering · vol. 461 · pp. 119202
Sharp Probabilistic Normwise Backward Error Bounds for LU Factorization
Liam Burke, El-Mehdi El Arar, Stef Graillat, Theo Mary
Preprint
QABBA: Symbolic Time Series Compression via Integer-Quantized Aggregation
Erin Carson, Xinye Chen, Fei He, Cheng Kang
Preprint
Mixed Precision Augmented GMRES
Yongseok Jang, Pierre Jolivet, Theo Mary
Numerical Linear Algebra with Applications · vol. 33 · no. 4 · pp. e70117
Automated Numerical Stability Analysis of Deep Learning Operators
Xinye Chen
Preprint · arxiv
A fixed-accuracy randomized interpolative decomposition based on pivoted QR
Alfredo Buttari, Karmijn Hoogveld, Theo Mary
Preprint
TOTO: Transparent I/O Tuning for HPC Applications
Francieli Boito, Luan Teylo, Mihail Popov, Laora Aimi, Alexis Bandet, Laércio Lima Pilla + 1 more
ICS 2026 - International Conference on Supercomputing · Belfast, GB
Mixed precision Newton's method for optimization
Nicolas Brisebarre, Giuseppe Carrino, Theo Mary, Elisa Riccietti
Preprint
Online learning of subgrid-scale models for quasi-geostrophic turbulence in planetary interiors
Hugo Frezat, Thomas Gastine, Alexandre Fournier
Preprint
Mixed-precision classic algebraic multigrid preconditioning for large-scale problems
Emmanuel Agullo, Ani Anciaux-Sedrakian, Alfredo Buttari, Petr Vacek
Preprint
Sharper Bounds on the Required Fixed-Point Precision for Emulating Floating-Point Matrix Multiplication
Massimiliano Fasi, Theo Mary, Mantas Mikaitis
Preprint
Learning Where to Simulate: Generative Active Sampling for Online PDE Surrogate Training
Pierre Cesar, Sofya Dymchenko, Abhishek Purandare, Bruno Raffin
Preprint · arxiv
Systematic Error-Aware Mixed-Precision Approach for Exploiting Microscaling Formats in Cholesky Factorization
Abdurauf Abdurakhmanov, Theo Mary, Hatem Ltaief, David E Keyes, Suhaib A. Fahmy
Preprint
Doubly adaptive precision restarted GMRES
Emmanuel Agullo, Luc Giraud, Pierre Jolivet, Theo Mary, Alexandre Tabouret
Preprint
Mixed Precision Compression of Tucker Decomposition
Grey Ballard, Theo Mary, Bhisham Dev Verma
Preprint
Enabling non-Flexible Mixed-Precision Preconditioned GMRES with Memory Accessors
Patrick Amestoy, Ouassel El Habti, Antoine Jego, Jean-Yves L'Excellent, Theo Mary
Preprint
Greedy Adaptive Task Splitting with Prediction and Placement for Heterogeneous Systems
Thomas Morin, Nathalie Furmento, Abdou Guermouche, Samuel Thibault, Pierre-André Wacrenier
HCW - 35th Heterogeneity in Computing Workshop · Nouvelle-Orleans, US
Mixed Precision Perturbation Analysis of Two-Level Domain Decomposition Preconditioners
Tom Caruso, Pierre Jolivet, Theo Mary, Frédéric Nataf, Pierre-Henri Tournier
Preprint
Méthode d'éléments finis discontinus hybridisée pour la résolution itérative accélérée de problèmes d'ondes en fréquence
Axel Modave, Roland Greffe, Christophe Geuzaine, Ari Rappaport, Ahmed Chabib
CSMA 2026 - 17ème Colloque National en Calcul des Structures · Giens, FR
Automated Precision Tuning for Numerical Algorithms
Xinye Chen, Thibault Hilaire, Fabienne Jézéquel
27th Conference of the International Linear Algebra Society · Blacksburg, US
Save the Date
NumPEx events
Discover the next NumPEx events: our seminars and conferences, as well as partner events
september, 2026







