Incollection,
The effect of learning parameters in complex fitness landscapes
Abstract Book of the XXIII IUPAP International Conference on Statistical Physics, Genova, Italy, (9-13 July 2007)
Abstract
Natural selection is grounded not only at the population level but also in processes at the individual level. In a large population of agents there are endogenous characteristics intrinsic to each agent that cannot be ignored, because they have enormous influence in the dynamics of the system. Two characteristics are relevant for develop this kind of models. First an individual bias that allows individuals to choose their best-suited kind of interaction in a non-symmetric fitness landscape. The second characteristic is a memory that allows agents to remember past interactions of their opponents in order to implement new optimal actions. This memory could be modeled using an Ising schema, making the system dependent on a learning parameter. In this work we analyze the effect of this learning parameter on the dynamics and distribution of the population of agents with large memory sizes and local bias in non-symmetric and complex fitness landscapes.
Tags
- bias
- complex
- dynamics
- fitness
- game
- individual
- landscapes
- learning
- population
- statphys23
- theory
- topic-5
