AGCT Evolutionary Nexus

Left Click (Drag): Drop Food

Spacebar: Spawn Random Agent

Powered by Google Cloud Run & NVIDIA GPUs

Evolutionary Dynamics

1. Foraging vs. Speed

Test: Stop feeding cells.
Observe: Fast cells starve. Population evolves into slow, efficient scavengers conserving energy.

2. The Apex Predator

Test: Create a massive food pile.
Observe: Cells gorge, physically expand, and cannibalize. Eventually, a colossal "Apex Predator" emerges.

3. The Baldwin Effect

Test: Spawn many random cells (Space).
Observe: Early generations learn via trial-and-error (RL). Later mutations lock in behaviors, spawning cells with an instinct to find food.

4. Reinforcement Learning (RL)

Concept: Individual Lifetime Adaptation.
Observe: As cells deplete energy, they randomly perturb neural weights (synapses). If this leads to food, the cell survives longer, effectively "learning" to steer.

5. Multi-Agent RL (MARL)

Concept: Competitive/Cooperative Arms Race.
Observe: Cells learn in a shared environment. Predators evolve to chase, forcing prey to evolve evasion or swarming tactics. Their neural networks co-adapt dynamically.