Genetic Algorithm in Matlab with Non-linear Constraints: -

according documentation on ga, program not demand trial points satisfy non-linear constraints satisfied in generation g. question this: how ga select elite children -- possible select points violate non-linear constraints? specifically, how non-linear constraints imbedded in fitness function? fitness funciton of form: obejctive function + penalty, where

penalty = sum_k max(g_k(x),0)^2 + sum_k h_k(x)^2

where g_k(x) non-linear inequality constraints, , h_k(x) non-linear equality constraints. if fitness function of form, can user change penalty function to, say,

sum_k max(g_k(x),0) + sum_k |h_k(x)|?