run_simulation.py

import os
os.environ["KMP_DUPLICATE_LIB_OK"] = "TRUE"
os.environ["OMP_NUM_THREADS"] = "1"

import argparse
import json
from evo_brain import AutonomousAgent, Visualizer

def main():
    parser = argparse.ArgumentParser(description="Run EvoBrain Autonomous Self-Evolving Agent Simulation")
    parser.add_argument("--stages", type=int, default=5, help="Number of life cycle stages")
    parser.add_argument("--threshold", type=float, default=0.2, help="Plateau MSE threshold for synaptic expansion")
    parser.add_argument("--initial-hidden", type=int, default=16, help="Initial hidden units in brain")
    parser.add_argument("--expansion-size", type=int, default=8, help="Hidden units to add upon synaptic expansion")
    parser.add_argument("--output-dir", type=str, default="results", help="Directory to save generated outputs")
    args = parser.parse_args()

    os.makedirs(args.output_dir, exist_ok=True)
    os.makedirs(os.path.join(args.output_dir, "figures"), exist_ok=True)

    print(f"=== Initializing EvoBrain (Initial Hidden: {args.initial_hidden}, Threshold: {args.threshold}) ===")
    agent = AutonomousAgent(
        input_dim=16,
        initial_hidden=args.initial_hidden,
        target_threshold=args.threshold,
        expansion_size=args.expansion_size
    )

    records = agent.run_life_cycle(stages=args.stages)
    losses = [r["final_loss"] for r in records]
    stages = [r["stage"] for r in records]

    print("\n=== Simulation Summary ===")
    for r in records:
        ev_msg = f"-> Synaptic growth: {r['old_hidden']} -> {r['new_hidden']} units" if r["evolved"] else "Converged (No expansion)"
        print(f"Stage {r['stage']}: Loss = {r['final_loss']:.4f} | {ev_msg}")

    loss_fig_path = os.path.join(args.output_dir, "figures", "cli_loss_trend.png")
    Visualizer.plot_loss_trend(stages, losses, target_threshold=args.threshold, save_path=loss_fig_path)
    print(f"\nLoss curve saved to: {loss_fig_path}")

    tsne_2d_path = os.path.join(args.output_dir, "figures", "cli_tsne_2d.png")
    Visualizer.plot_tsne_2d(agent.brain, steps=args.stages, save_path=tsne_2d_path)
    print(f"2D t-SNE latent map saved to: {tsne_2d_path}")

    tsne_3d_path = os.path.join(args.output_dir, "figures", "cli_tsne_3d.png")
    Visualizer.plot_tsne_3d(agent.brain, steps=args.stages, save_path=tsne_3d_path)
    print(f"3D t-SNE latent map saved to: {tsne_3d_path}")

if __name__ == "__main__":
    main()