Autonomous camera for agent-based in situ visualization

Abstract

For effective in situ visualization in high performance computing, a viewpoint should be placed close to a critical spot or volume-of-interest (VOI). In order to track unpredictable motions of VOI in simulations, we propose introducing agent-based modeling to the in-situ visualization. In this method, agents are autonomous cameras, and their environment is the simulation. As a demonstration experiment of the agent-based in situ visualization, we put a camera agent to 3D cellular automata. The camera agent successfully tracks a VOI of cells in highly complex time development.