Abstract:Mean field limits are an important tool in the context of large-scale dynamical systems, in particular, when studying multiagent and interacting particle systems. While the continuous-time theory is well-developed, few works have considered mean field limits for deterministic discrete-time systems, which are relevant for the analysis and control of large-scale discrete-time multiagent system. We prove existence results for the mean field limit of very general discrete-time control systems, for which we utilize kernel mean embeddings. These results are then applied in a typical optimal control setup, where we establish the mean field limit of the relaxed dynamic programming principle. Our results can serve as a rigorous foundation for many applications of mean field approaches for discrete-time dynamical systems.
What problem does this paper attempt to address?
This paper aims to solve the Mean Field Limit (MFL) problem in large - scale discrete - time dynamic systems, especially in multi - agent systems (MAS) and interacting particle systems (IPS). Specifically, the paper focuses on the Mean Field Limit of deterministic discrete - time systems. Such systems are particularly important in engineering applications, but currently there are relatively few studies in this regard.
### Main problems
1. **Existence problem**: The paper first addresses the existence problem of the Mean Field Limit. For continuous - time systems, there is already a well - established theory to deal with the Mean Field Limit, but for discrete - time systems, especially discrete - time systems with control inputs, the relevant theory is not yet perfect. By introducing the Kernel Mean Embeddings (KME) method, the paper proves the existence of the Mean Field Limit for very general discrete - time control systems.
2. **Optimal control problem**: The paper further applies these existence results to the optimal control problem of discrete - time control systems. Specifically, the paper establishes a Mean Field Limit version of the discrete - time dynamic programming principle, providing a solid theoretical basis for the control and simulation of large - scale discrete - time multi - agent systems.
### Main contributions
1. **New Mean Field Limit existence results**: The paper for the first time presents Mean Field Limit existence results applicable to general discrete - time control systems, and these results are highly general.
2. **Application of Kernel Mean Embeddings**: The paper introduces Kernel Mean Embeddings as a tool to prove the existence of the Mean Field Limit. This is one of the major contributions in this field.
3. **Optimal control application**: The paper applies these existence results to the optimal control problem of discrete - time control systems, especially the Mean Field Limit version of the relaxed dynamic programming principle, providing theoretical support for the control of large - scale discrete - time multi - agent systems.
### Conclusion
The paper provides a strict theoretical basis for the Mean Field control of large - scale discrete - time multi - agent systems. These results can be widely applied in fields such as robotics, power systems, and social dynamics. The current research work is further exploring the specific implementation of these theories in optimal control applications, especially the nonlinear model predictive control (NMPC) of large - scale multi - agent systems.