Space, while inherent to the natural world, often finds itself omitted in bio-inspired computational system designs. Spatial Genetic Programming (SGP) is a Genetic Programming (GP) paradigm that incorporates space as a fundamental dimension, evolving alongside Linear Genetic Programming (LGP) programs. In SGP, each individual model is represented by a 2D space consisting of one or many LGP programs. These programs execute in an order controlled by their spatial position. The contributions of this work are: Introducing SGP as a tool for studying evolution of space in GP. Application of the proposed system to a range of problems including symbolic regression, classic control and decision-making problems and a comparison to other common GP paradigms. A study on how spatial dimension influences generational diversity, on emergence of spatially-induced localization within the system, and on the emergence of iterative structures within the system. The findings of this research open new avenues towards a better understanding of natural evolution and how the dimension of space could be useful as a handle for controlling important aspects of evolution.
Read
