ref:cma_optimizer

CmaOptimizer

Optimizer based on the CMA-ES algorithm by [Hansen].

Inherits from EsOptimizer.

Public Attributes

Parameter Type Description
max_errors int Maximum number of errors allowed during evaluation, use a negative value equates to lambda - max_errors; default = 0.
lambda int Lambda parameter (population size); default = automatic.
mu int Mu parameter (offspring size); default = automatic.
sigma double Sigma parameter (step size); default = automatic.
random_seed long Random seed used by the optimizer; default = 123.
flat_fitness_epsilon double Epsilon value to detect flat fitness; default = 1e-6.
init_file path Parameter file (.par) to be used for initial parameter values.
use_init_file bool Use init_file (if exists); default = true.
use_init_file_std bool Use the standard deviations from the init_file; when set to false, the initial standard deviation is computed as follows: STDEV = parameter value * init_file_std_factor + init_file_std_offset; default = true.
init_file_std_factor double Factor by which to multiply the standard deviations from the init_file; default = 1.0.
init_file_std_offset double Offset added to the standard deviations from init_file; default = 0.
init_file_include String Pattern matching the init_file parameters to include (semicolon seperated); default = “” (all).
init_file_exclude String Pattern matching the init_file parameters to exclude (semicolon seperated); default = “” (none).
init ParInitSettings Section with parameter initialization settings (.par), multiple allowed; see ParInitSettings for details.
max_threads size_t Maximum number of threads to use for this optimization (deprecated, use global settings instead); default = 32.
thread_priority int Thread priority of the optimization; 0 = lowest, 7 = highest, default = 1.
max_generations size_t Number of iterations after which to stop the optimization; default = 100000.
min_progress double Minimum progress after which to stop the optimization; default = 1e-5.
min_progress_samples size_t Minimum number of samples after which progress is measured; default = 200.
window_size size_t Window size used for measureing progress; default = 500.
min_improvement_for_file_output Real The minimum relative improvement needed for file output; default = 0.05.
max_generations_without_file_output size_t The maximum number of iterations without file output; default = 1000.
target_fitness double Target fitness value, stop optimization if better; default = not set.
log_level xo::log::level Set the log level [1-7] for the optimization.log file (higher is less logging); default = 3.
signature_prefix String Prefix signature with custom string. Special tags: DATE_TIME, DATE_TIME_EXACT, SCONE_VERSION.
signature_postfix String Append custom string to signature. Special tags: DATE_TIME, DATE_TIME_EXACT, SCONE_VERSION.
signature String Set custom signature and omit the auto-generated signature.

Converted from doxygen using dokugen