Mean-field Limit and Computation in Kinetic Theory (Morlet Chair – Shi Jin) – POSTPONED

We will study machine-learning type of numerical methods for N-body problems at both classical and quantum levels. Recently the random batch method, motivated by the stochastic Gradient Descent method popular in machine learning, was introduced which reduce the computational cost of Nbody simulation from O(N^2) to O(N). We will study analytical properties of the method, its connection with mean-field limit for general interacting potentials, and its quantum analogy.

• Mihai Bostan (Aix-Marseille Université)
• François Golse (Ecole polytechnique)
• Maxime Hauray (Aix-Marseille Université)
• Shi Jin (Shanghai Jiao Tong University & Aix-Marseille Université)
• Jian-Guo Liu (Duke University)