| Issue |
ESAIM: ProcS
Volume 77, 2024
CEMRACS 2022 - Transport in physics, biology and urban traffic
|
|
|---|---|---|
| Page(s) | 145 - 175 | |
| DOI | https://doi.org/10.1051/proc/202477145 | |
| Published online | 18 November 2024 | |
Deep learning for mean field optimal transport*
1
Université Côte d’Azur, 28 Avenue de Valrose, 06103 Nice, France
2
IRMA UMR 7501, Université de Strasbourg, 7 Rue René Descartes, 67000 Strasbourg, France
3
NYU Shanghai Frontiers Science Center of Artificial Intelligence and Deep Learning; NYU-ECNU Institute of Mathematical Sciences at NYU Shanghai, 3663 Zhongshan Road North, Shanghai, 200062, China
4
Mohammed VI Polytechnic University. Lot 660, Hay Moulay Rachid Ben Guerir, 43150, Morocco
5
Georgia Institute of Technology, 686 Cherry Street NW, Atlanta, 30332, USA
a e-mail: sebastian.baudelet@univ-cotedazur.fr
b e-mail: brieuc.frenais@math.unistra.fr
c e-mail: mathieu.lauriere@nyu.edu
d e-mail: amal.machtalay@um6p.ma
e e-mail: yzhu738@gatech.edu
Mean field control (MFC) problems have been introduced to study social optima in very large populations of strategic agents. The main idea is to consider an infinite population and to simplify the analysis by using a mean field approximation. These problems can also be viewed as optimal control problems for McKean-Vlasov dynamics. They have found applications in a wide range of fields, from economics and finance to social sciences and engineering. Usually, the goal for the agents is to minimize a total cost which consists in the integral of a running cost plus a terminal cost. In this work, we consider MFC problems in which there is no terminal cost but, instead, the terminal distribution is prescribed as in optimal transport problem. By analogy with MFC, we call such problems mean field optimal transport problems (or MFOT for short) since they can be viewed as a generalization of classical optimal transport problems when mean field interactions occur in the dynamics or the running cost function. We propose three numerical methods based on neural networks. The first one is based on directly learning an optimal control. The second one amounts to solve a forward-backward PDE system characterizing the solution. The third one relies on a primal-dual approach. We illustrate these methods with numerical experiments conducted on two families of examples.
The authors would like to thank the CIRM for welcoming the CEMRACS 2022, the organizers of the CEMRACS 2022 for the opportunity to work on this project as well as their respective institutions. They are also grateful to the NYU-ECNU Institute of Mathematical Sciences at NYU Shanghai, the CIMPA fellowships program, and the ENS Rennes for their support. This work was supported in part through the NYUSH IT High Performance Computing resources, services, and staff expertise, as well as the ASCC Toubkal cluster resources.
© EDP Sciences, SMAI 2024
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Current usage metrics show cumulative count of Article Views (full-text article views including HTML views, PDF and ePub downloads, according to the available data) and Abstracts Views on Vision4Press platform.
Data correspond to usage on the plateform after 2015. The current usage metrics is available 48-96 hours after online publication and is updated daily on week days.
Initial download of the metrics may take a while.