Coordination
through emergence
A simulation-first platform where distributed agents coordinate through local rules -- inspired by neural networks, fungal mycelium, and swarm biology -- producing emergent global intelligence.
How the brain works
Like biological neural systems, Kernel has no central controller. Intelligence emerges from thousands of simple agents following local rules -- each one a neuron in a distributed brain.
The simulation core is Rust compiled to WebAssembly. Three.js renders the emergent patterns in real-time. TOML configs define experiments without code changes.
Signal Flow
Every frame: spatial hash rebuild, behavior computation (boids + chemotaxis + pursuit/flee), movement integration, pheromone diffusion, energy lifecycle. All in under 16ms.
Biological simulations
Flocking
Reynolds boids. 1000+ agents self-organize into coherent flocks from three local rules. No leader. No plan. Pure emergence.
Pheromone Paths
Ant-colony optimization. Chemical trails diffuse and evaporate, forming efficient transport networks through stigmergy.
Ecosystem
Predator-prey with energy, reproduction, and bidirectional mutation. Populations oscillate via Lotka-Volterra dynamics.
Mycelium Network
Adaptive graph where signals strengthen active edges and prune idle ones. Hebbian learning in a fungal topology.
Repository structure
Mathematical foundations
Flock Equilibrium
Signal Half-life
Trail Persistence
Full proofs and stability analysis in research/theory.md