MEPHISTO

Project Overview

MEPHISTO is a research project funded by the Agence Nationale de la Recherche (ANR), dedicated to the development of new mathematical and algorithmic methods for large-scale optimization problems with hierarchical and parsimonious structures.

The project aims to advance both the theoretical foundations and practical implementations of optimization techniques tailored to modern high-dimensional applications.

Scientific Objectives

• Design of parsimonious and hierarchically structured optimization models
• Development of scalable algorithms with provable guarantees
• Analysis of convergence and complexity in high-dimensional settings
• Efficient numerical implementations for large-scale problems
• Applications to inverse problems and scientific machine learning

Research Themes

• Structured sparsity and hierarchical regularization
• Multilevel optimization
• Large-scale convex and nonconvex optimization
• Scientific machine learning
• Matrix and tensor decompositions
• Reduced and mixed precision algorithms

Coordination

The project is coordinated by Elisa Riccietti.

Funding

This project is funded by the ANR – Agence Nationale de la Recherche, ANR-24-CE23-7039.

Contact

For more information about the project, collaborations, or research opportunities:

Elisa Riccietti
elisa.riccietti@ens-lyon.fr