Oscillating solutions to the mean-field Langevin descent-ascent flow
Jean-Christophe Mourrat, Loucas Pillaud-Vivien

TL;DR
This paper provides a counterexample demonstrating that the mean-field Langevin descent-ascent flow can exhibit non-convergent cyclic behavior under certain conditions involving double well payoff functions and strong coupling.
Contribution
It shows that the mean-field Langevin descent-ascent flow does not always converge, especially with specific payoff functions and parameter regimes, challenging previous assumptions.
Findings
The mean-field dynamics can admit a limit cycle in certain settings.
Strong coupling and small entropic regularization lead to cyclic, non-convergent behavior.
The paper constructs a specific counterexample with double well functions.
Abstract
We present a counterexample to the statement of convergence of the mean-field Langevin descent-ascent flow on . We consider payoff functions that are shaped as a double well in each coordinate, and for which the deterministic dynamics admits a limit cycle. When the coupling between the two coordinates is sufficiently strong and the entropic regularization sufficiently small, we show that the mean-field dynamics remains close to this cyclic behavior, and in particular, does not converge.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
