Abstract

Especially in the last years the idea of finding a simpler meta-model for a developed simulation model has become more and more popular as not only the research questions, but also the resulting models have become more and more complex. The meta-model hereby helps understanding the behaviour of the original model and supports its validation and verification process. Moreover, it also gives a second perspective on the modelled system. Mean-field theory is a very formal link between microscopic and macroscopic models and can be used to find meta-models for either of the two types. Usually so-called mean-field analysis is used to find a summarising simpler macroscopic model for a given complex microscopic approach, but we will emphasise the inverse process in this work: Applying inverse mean-field analysis on an ordinary differential equation model we systematically derive a microscopic representation of the model. On purpose we chose a very unusual model to apply the method to: the undamped linearised pendulum equation. Hereby we want to emphasise the method’s flexibility and generality, investigate possible benefits of the gained microscopic meta-model for this strange field of application, and discuss about interpretation of the elements of the resulting agent-based pendulum model.