Experiment/Experiments/New

Create New Experiment

Understanding Experiment Groups

This experiment tests whether adaptive mutation (scaling mutation rate by fitness) accelerates convergence to Nash Equilibrium compared to static mutation.

Control Group
Static Mutation — Fixed rate ε = 0.05 applied uniformly to all offspring. Traditional genetic algorithm approach serving as the baseline.
Experimental Group
Adaptive Mutation — Dynamic rate calibrated to start at ~5% (same as Control), then scales with fitness: low-fitness agents mutate more (exploration), high-fitness agents mutate less (exploitation).

For statistical significance, run at least 5 experiments of each group with different random seeds.

Static mutation: Fixed rate ε = 0.05

750 ticks ≈ 12 seconds of simulated agent lifetime

Fixed mutation rate (e.g., 0.05 = 5%)

Top percentage to select (e.g., 0.2 = top 20%)

Cancel