Thousands of microscopic 'agents' crawl over the surface, leaving behind a chemical trail and returning to where the trail is strongest. From this simple rule, without any central control, an effective transport network grows.
The same real Physarum polycephalum builds networks that astonishingly accurately replicate the map of Tokyo subway lines or roads between cities. In 2010, Japanese scientists recreated the exact Tokyo rail network by laying plastic strands at station locations.
Each agent is 'blind' — it perceives only a trail just ahead of itself through three sensors (left/central/right) and goes back where the signal is stronger. No brain, no plan — yet an almost optimal network that is robust to disruptions.
⏸ Pause
↺ Reset
✕ Clear
🍎 Food
🧱 Wall
Spider web
Channels
Chaos
Random
Balanced parameters — a classic branched spider web with stable nodes.
Simulation step
0
Active agents
0
Cover with network
0%
Sources of food
0 / 8
In 'Feast' mode, a click adds a food source — agents will eventually establish a stable path to it. In 'Wall' mode, drag the mouse to draw an obstacle — the network will avoid it, searching for an alternative route, like true Physarum in a maze.
Click in the field — add food · Drag — rotation · Scroll — zoom