Adjust the
parameters.
Watch it emerge.
A laboratory of live, playable models — flocks, epidemics, cellular automata and colonies. Every system here is computed in real time from simple local rules. Pull a slider; the global pattern reorganises before your eyes.
One family of simulations across 5 domains — cellular automata, agent-based worlds, nonlinear dynamics, learning systems and biology. Each card is a live in-browser model (or a recorded run) distilled from a deeper standalone research project.
FIG. 02OPEN ▸FIG. 03OPEN ▸FIG. 21OPEN ▸FIG. 22OPEN ▸FIG. 23OPEN ▸FIG. 26OPEN ▸
Cellular automata
Game of Life
Conway, 1970
A two-state cellular automaton. Four rules give rise to gliders, guns and universal computation.
3 parametersReal-time
Turing patterns
Reaction – Diffusion
Gray–Scott
Two chemicals, diffusing and reacting. Turing's morphogenesis grows spots, mazes and coral from noise.
3 parametersReal-time
1-D automata
Elementary CA
Wolfram · 1-D
One integer 0–255 sets eight rules for a row of cells — out come fractals (90), chaos (30) and Rule 110, a Turing-complete computer.
3 parametersReal-time
Logic circuits
Wireworld
Silverman, 1987
A four-state automaton where electrons race along wires — enough to build logic gates, memory and, in principle, a whole computer.
2 parametersReal-time
Emergent order
Langton's Ant
Langton, 1986
Two rules — turn right on white, left on black — and ~10,000 steps of chaos suddenly self-organise into an endless diagonal highway.
2 parametersReal-time
Continuous automata
Lenia
Chan, 2018 · continuous CA
Life made continuous: real-valued cells, a smooth ring kernel and a bell-shaped growth rule conjure the Orbium — a creature that genuinely swims.
3 parametersReal-time
FIG. 01OPEN ▸FIG. 07OPEN ▸FIG. 08OPEN ▸FIG. 13OPEN ▸FIG. 14OPEN ▸FIG. 15OPEN ▸FIG. 16OPEN ▸FIG. 17OPEN ▸FIG. 18OPEN ▸FIG. 19OPEN ▸
Emergence
Flocking
Reynolds boids
Local rules, global order. Hundreds of agents self-organise into a murmuration.
7 parametersReal-time
Contagion
Epidemic Spread
SIR · agent-based
Susceptible, infected, recovered. A contact network in motion — flatten the curve in real time.
6 parametersReal-time
Swarm intelligence
Ant Colony
Stigmergy
Pheromone foraging. Trails self-organise between nest and food with no central control.
5 parametersReal-time
Game theory
Cooperation
Spatial Prisoner's Dilemma
Defection wins in a well-mixed crowd — but on a grid, cooperators huddle into clusters that defend their borders, and cooperation survives.
3 parametersReal-time
Social dynamics
Segregation
Schelling, 1971
Each agent only wants a few like-minded neighbours — yet that mild preference tips a mixed city into stark, self-organised segregation.
3 parametersReal-time
Social dynamics
Echo Chambers
Bounded confidence
People only listen to those who already think like them. Tune the open-mindedness and watch a society reach consensus — or fracture into camps.
3 parametersReal-time
Decision
Voting Methods
Social choice
The same voters, four fair-looking rules, four different winners — a live tour of the spoiler effect and Arrow's impossibility theorem.
3 parametersReal-time
Bounded rationality
Minority Game
El Farol
Selfish agents with no communication and short memories self-organise to share a scarce resource, hovering right at its capacity.
3 parametersReal-time
Markets
Market
Fundamentalists vs chartists
Price emerges from value-investors versus trend-chasers. Trend-following breeds bubbles, crashes and the clustered volatility of real markets.
4 parametersReal-time
Econophysics
Wealth & Inequality
Kinetic exchange
Start everyone equal and let them trade by fair coin flips; a Pareto elite emerges from pure chance — and tax policy fights back.
4 parametersReal-time
FIG. 04OPEN ▸FIG. 05OPEN ▸FIG. 06OPEN ▸
Chaos
Route to Chaos
Logistic map
Turn one knob and a stable population splits into 2-, 4-, 8-cycles and then deterministic chaos — at Feigenbaum's universal rate.
2 parametersReal-time
Dynamical system
Predator – Prey
Lotka–Volterra
Coupled ODEs for two species. Watch the eternal oscillation and its phase-space orbit.
5 parametersReal-time
Synchronization
Synchrony
Kuramoto
Hundreds of oscillators with their own rhythms; turn up the coupling and they snap into lockstep — the math behind brain waves and firefly flashing.
3 parametersReal-time
FIG. 10OPEN ▸FIG. 11OPEN ▸FIG. 12OPEN ▸FIG. 20OPEN ▸FIG. 25OPEN ▸
Neural memory
Associative Memory
Hopfield, 1982
Store patterns as energy minima. Feed in a corrupted cue and watch the network fall back into the memory it most resembles.
4 parametersReal-time
Reinforcement
Reward Prediction
TD-learning
A value signal learns to predict reward — and its prediction error behaves like dopamine, marching from the reward back to the cue.
4 parametersReal-time
Reinforcement
Q-Learning
Watkins, 1989
An agent learns a gridworld by trial and error; value floods outward from the goal and the greedy policy snaps into place.
4 parametersReal-time
Emergence
Future Freedom
Causal entropic forces
A particle with no goal, drifting to wherever its futures stay most open — Wissner-Gross's thesis that intelligence is a force toward freedom of action.
3 parametersReal-time
▶ recorded
Deep RL
Deep RL — Atari Pong
IQN agent
A distributional deep-RL agent (Implicit Quantile Networks) playing Pong from raw pixels — recorded during a greedy evaluation.
Recorded runVideo
FIG. 09OPEN ▸
▶ recordedFIG. 24OPEN ▸
Neuroscience
Action Potential
Hodgkin–Huxley
The all-or-nothing nerve impulse from real ion-channel kinetics — inject current and watch a single neuron fire the spike behind every thought.
4 parametersReal-time
▶ recordedWhole-organism
C. elegans
OpenWorm · c302 + Sibernetic
A 1-mm worm simulated whole: ~300 neurons (c302 / NEURON) driving a soft-body physics model — and a forward crawl emerges. A recorded run.
Recorded runVideo
DrawingSimulation Lab — Sheet 01 / 27
DisciplineComplex Systems
MethodNumerical / Agent
Date2026.05.30