src.baseline.bbo.cmaes

Module Contents

Classes

CMAES

Introduction A novel evolutionary optimization strategy based on the derandomized evolution strategy with covariance matrix adaptation. This is accomplished by efficientlyincorporating the available information from a large population, thus significantly re-ducing the number of generations needed to adapt the covariance matrix.

API

class src.baseline.bbo.cmaes.CMAES(config)

Bases: src.environment.optimizer.basic_optimizer.Basic_Optimizer

Introduction

A novel evolutionary optimization strategy based on the derandomized evolution strategy with covariance matrix adaptation. This is accomplished by efficientlyincorporating the available information from a large population, thus significantly re-ducing the number of generations needed to adapt the covariance matrix.

Original paper

“Reducing the time complexity of the derandomized evolution strategy with covariance matrix adaptation (CMA-ES).” Evolutionary Computation 11.1 (2003): 1-18.

Initialization

Introduction

Initializes the CMA-ES optimizer with the provided configuration settings.

Args:

• config (object): config object from src/config.py.

  • The Attributes needed for the CMAES are the following:

    • log_interval (int): Interval at which logs are recorded.Default is config.maxFEs/config.n_logpoint.

    • n_logpoint (int): Number of log points to record.Default is 50.

    • full_meta_data (bool): Flag indicating whether to use full meta data.Default is False.

    • __FEs (int): Counter for the number of function evaluations.Default is 0.

    • __config (object): Stores the config object from src/config.py.

    • NP(int): Set the population size in config to 50.

Attributes:

• __config (object): Configuration object containing algorithm parameters.

__str__()

Returns a string representation of the CMAES class.

Returns:

• str: The string “CMAES”.

run_episode(problem)

Introduction

Executes a single optimization episode using the CMA-ES (Covariance Matrix Adaptation Evolution Strategy) algorithm on a given problem instance. Tracks the optimization process, logs costs at specified intervals, and optionally collects meta-data about the search trajectory. Use the cma package to implement the CMA-ES algorithm.

Args:

• problem (object): An optimization problem instance that must provide the following attributes and methods:

  • dim (int): Dimensionality of the problem.

  • lb (np.ndarray): Lower bounds for the variables.

  • ub (np.ndarray): Upper bounds for the variables.

  • eval(x) (callable): Function to evaluate the objective at point x.

  • optimum (float or None): Known optimum value of the problem (if available).

Returns:

• dict: A dictionary containing:

  • 'cost' (list of float): The best cost found at each logging interval.

  • 'fes' (int): The total number of function evaluations performed.

  • 'metadata' (dict, optional): If self.full_meta_data is True, includes:

    • 'X' (list of np.ndarray): The population positions at each iteration (in original problem scale).

    • 'Cost' (list of np.ndarray): The corresponding costs for each population.