This experiment implements a GPU-accelerated Gray-Scott reaction-diffusion system — a mathematical model that produces organic, self-organizing patterns from two interacting chemical species. The system was first described by Gray and Scott in 1984 and produces an extraordinary variety of pattern types depending on feed and kill rate parameters.
The simulation runs entirely on the GPU using WebGL 2.0 framebuffer ping-pong, computing thousands of Laplacian diffusion steps per second across a high-resolution grid. Two chemical concentrations (U and V) react and diffuse according to coupled partial differential equations, spontaneously forming spots, stripes, spirals, and labyrinthine structures.
Interactive controls expose the feed rate (F), kill rate (k), diffusion ratio, and simulation speed. Different parameter regions produce distinct morphologies — from mitotic cell-like spots to coral-like branching to maze-like Turing patterns.