Gray-Scott
This simulation needs WebAssembly, and it could not start in this browser.
The image is a frame from a real run. Check that JavaScript is on, or try another browser.
The Gray-Scott model describes two chemicals spread over a surface: a substrate that is fed in from outside, and a catalyst that consumes it to make more of itself and is slowly removed. In 1993 John Pearson surveyed its two rate constants systematically, the feed rate and the kill rate , and found a wide range of patterns: dividing spots, stripes, labyrinths, spiral waves and chaos. Reaction-diffusion systems of this kind are studied as models for animal coat markings and for the banded vegetation of dry regions.
Governing equation
The reaction is : it takes two molecules of to convert one of , which makes autocatalytic. The substrate is replenished at rate , and is removed at rate . With the default coefficients, and (Pearson’s values in grid units), the substrate spreads twice as fast as the catalyst. A patch of therefore stays compact while it drains the substrate around it, and that is what lets patterns form. The grid is advanced with explicit Euler steps, , and its edges wrap around.
What you see
The color shows the concentration of . Every run starts from , with a noisy square of mixed concentrations in the center, and is run for 200 steps with the mitosis values before the chosen and take over.
Each preset is one point in the plane, named after the pattern it is set up to show:
- mitosis: ,
- coral: ,
- maze: ,
- spots: ,
- worms: ,
- u-skate: , , small U-shaped structures that glide across the grid
- fingerprints: ,
- spirals: ,
Small changes matter: all eight lie within 0.014 of each other in .
Parameters
feed rate f: how fast the substrate is replenished from outside, 0.01 to 0.09.
kill rate k: how fast the catalyst is removed, 0.04 to 0.075. Too high a value and dies out.
Dᵤ diffusion, D_v diffusion: the two diffusion coefficients. Patterns need clearly larger than .
resolution: the side length of the square grid, 50 to 400 cells.