Data-driven model reduction of agent-based systems using the Koopman generator.

摘要: The dynamical behavior of social systems can be described by agent-based models. Although single agents follow easily explainable rules, complex time-evolving patterns emerge due to their interaction. simulation and analysis such models, however, is often prohibitively time-consuming if the number large. In this paper, we show how Koopman operator theory used derive reduced models using only data. Our goal learn coarse-grained represent dynamics ordinary or stochastic differential equations. new variables are, for instance, aggregated state model, modeling collective larger groups entire population. Using benchmark problems with known demonstrate that obtained are in good agreement analytical results, provided numbers sufficiently