Simulating Agent Territorial Battle (6)

Описание: In this animation, I simulate agent territorial competition, heavily inspired by the game paper.io. Each agent will perform random walks, with heuristcs to bias the direction of movement, and trace a path. Once the agent's path connects back to its territory, the enclosed area will be filled and turn into additional territory. Agents are able to steal territory from each other, and if Agent A every hits another Agent B's active path (not yet enclosed), Agent B will be eliminated. One of three movement heuristics are given to each agent: low-risk (tend to stay near territory), regular, and high-risk (tend to go far out from territory). Animations 1, 2, 3 show an agent for each heuristic in an isolated environment. Animation 4 shows 10 agents of each movement heuristic in a competitive environment. Animations 5, 6 show data plots for the competitive simulation displayed in Animation 4.

Author: Joshua Lu

This animation was created for the EPS 109 course "Computer simulations with Jupyter notebooks" at the University of California, Berkeley: https://militzer.berkeley.edu/EPS109. It offers a general introduction to computer simulations for all science majors. Participants will use the Jupyter notebooks to perform calculations, design 3D graphics, and build on existing computer programs. No prior programming experience is required. The course teaches fundamental skills for scientists and helps students with future homework. It covers data processing techniques as well standard simulations methods including Monte Carlo and molecular dynamics. The algorithms are illustrated with a variety of applications in earth and planetary science as well as astronomy.