Scaling up Mean Field Games with Online Mirror Descent
This addresses scaling challenges in large-scale multi-agent and multi-population games, establishing a state-of-the-art method.
The paper tackles scaling equilibrium computation in Mean Field Games by using Online Mirror Descent, showing it provably converges to a Nash equilibrium and empirically outperforms traditional algorithms like Fictitious Play, solving examples with hundreds of billions of states for the first time.
We address scaling up equilibrium computation in Mean Field Games (MFGs) using Online Mirror Descent (OMD). We show that continuous-time OMD provably converges to a Nash equilibrium under a natural and well-motivated set of monotonicity assumptions. This theoretical result nicely extends to multi-population games and to settings involving common noise. A thorough experimental investigation on various single and multi-population MFGs shows that OMD outperforms traditional algorithms such as Fictitious Play (FP). We empirically show that OMD scales up and converges significantly faster than FP by solving, for the first time to our knowledge, examples of MFGs with hundreds of billions states. This study establishes the state-of-the-art for learning in large-scale multi-agent and multi-population games.