cma.sigma_adaptation

module documentation

step-size adaptation classes, currently tightly linked to CMA, because hsig is computed in the base class

Class CMAAdaptSigmaBase step-size adaptation base class, implement hsig (for stalling distribution update) functionality via an isotropic evolution path.
Class CMAAdaptSigmaCSA CSA cumulative step-size adaptation AKA path length control.
Class CMAAdaptSigmaDistanceProportional artificial setting of sigma proportional to ||m||,
Class CMAAdaptSigmaMedianImprovement Compares median fitness to the 27%tile fitness of the previous iteration, see Ait ElHara et al, GECCO 2013.
Class CMAAdaptSigmaNone constant step-size sigma
Class CMAAdaptSigmaTPA two point adaptation for step-size sigma.
Variable csa_dampdown_fac Undocumented
Variable CSA_dampfac_mueff Damping for large mueff, the default was 2, however 10 would solve issue #231?
Variable CSA_dampfac_mueff_attenuation_dimension Damping attenuation for small dimension such that CSA_dampfac_mueff_inner is multiplied by [1/2, 3/4, 7/8,...] when dimension == [1, 2, 3,...] * attentuation_dimension and with a smaller factor in smaller dimension bounded to not get below 1.
Variable CSA_dampfac_mueff_inner Damping for large mueff, the default was 1
Variable tpa_dampdown_fac is 4 in the restricted CMA of 2024, was tentatively 2, 4 mostly doesn't change standard performance much
Function _norm Undocumented
Variable _inner_threshold Undocumented
Variable _true_inner Undocumented
csa_dampdown_fac: int =

Undocumented

CSA_dampfac_mueff: int =

Damping for large mueff, the default was 2, however 10 would solve issue #231?

CSA_dampfac_mueff_attenuation_dimension: int =

Damping attenuation for small dimension such that CSA_dampfac_mueff_inner is multiplied by [1/2, 3/4, 7/8,...] when dimension == [1, 2, 3,...] * attentuation_dimension and with a smaller factor in smaller dimension bounded to not get below 1.

CSA_dampfac_mueff_inner: int =

Damping for large mueff, the default was 1

tpa_dampdown_fac: int =

is 4 in the restricted CMA of 2024, was tentatively 2, 4 mostly doesn't change standard performance much

def _norm(x):

Undocumented

_inner_threshold: int =

Undocumented

_true_inner: bool =

Undocumented