Adaptive agent based model to simulate a necessity of many religions, economies or political systems

To understand principal tendencies and outcomes of changes in modern world we model the societies as the multi agent systems which are aimed to function in varying environments, adapt to the unexpected alterations, to comply the fitness function and survive. We model the agents as very simple systems, the single layer perceptrons. Failure to comply the survivability condition results in the agent being removed from the “society” and be replaced by a "newborn". The offspring inherits some "upbringing" information from its "parent agent". To increase ability of agents to adapt to environmental changes more rapidly we suggest storing their gains accumulated during a period of last environmental changes, adding a noise to training signals and level of "survival threshold". We analyzed a split of agents into many groups with moderate cooperation between the agents inside the group and very limited cooperation between the groups. A presence of optimal number of groups and necessity of small groups were demonstrated. The look at the society and a system of collaborating adaptive agents suggests a new way of thinking. Our study gives one more argument that presence of variety of religions, economies and political systems could help human population overcome everlastingly changing environments.