Class OpenEvolve

openevolve/controller.py:56–583

Main controller for OpenEvolve Orchestrates the evolution process, coordinating between the prompt sampler, LLM ensemble, evaluator, and program database. Features: - Tracks the absolute best program across evolution steps - Ensures the best solution is not lost during the

class OpenEvolve:
    """
    Main controller for OpenEvolve

    Orchestrates the evolution process, coordinating between the prompt sampler,
    LLM ensemble, evaluator, and program database.

    Features:
    - Tracks the absolute best program across evolution steps
    - Ensures the best solution is not lost during the MAP-Elites process
    - Always includes the best program in the selection process for inspiration
    - Maintains detailed logs and metadata about improvements
    """

    def __init__(
        self,
        initial_program_path: str,
        evaluation_file: str,
        config: Config,
        output_dir: Optional[str] = None,
    ):
        # Load configuration (loaded in main_async)
        self.config = config

        # Set up output directory
        self.output_dir = output_dir or os.path.join(
            os.path.dirname(initial_program_path), "openevolve_output"
        )
        os.makedirs(self.output_dir, exist_ok=True)

        # Set up logging
        self._setup_logging()

        # Manual mode queue lives in <openevolve_output>/manual_tasks_queue
        self._setup_manual_mode_queue()

        # Set random seed for reproducibility if specified
        if self.config.random_seed is not None:
            import hashlib
            import random

            import numpy as np

            # Set global random seeds
            random.seed(self.config.random_seed)
            np.random.seed(self.config.random_seed)

            # Create hash-based seeds for different components
            base_seed = str(self.config.random_seed).encode("utf-8")
            llm_seed = int(hashlib.md5(base_seed + b"llm").hexdigest()[:8], 16) % (2**31)

            # Propagate seed to LLM configurations
            self.config.llm.random_seed = llm_seed
            for model_cfg in self.config.llm.models:
                if not hasattr(model_cfg, "random_seed") or model_cfg.random_seed is None:
                    model_cfg.random_seed = llm_seed
            for model_cfg in self.config.llm.evaluator_models:
                if not hasattr(model_cfg, "random_seed") or model_cfg.random_seed is None: