In recent years evolution of players strategy on static networks has been extensively studied.
On the other hand, many of studies focuses only the process of network formation, such as
preference attachment process. I considered the general case when both agents and networks
are dynamic. I am interested in the co-evolution of network topology and players strategy.
To model such a co-evolution dynamics we applied Q-learning in a repeated normal games.
I first provide a comprehensive analysis of a Boltzmann Q-learning in Two-player Two action
games. Then, I will provide my results on N-player game on network, and finally show emergence
of specific network as a result of the co-evolution learning.