Constraint Handling Heuristics for Evolution Strategies - Self-Adaptive Heuristics for Evolutionary Computation - page 130

whether it is worth to invest the higher number of fitness and constraint
function calls.
5. The TSES is inspired by the concepts of sex and pairing. Each individual is
assigned with a feature called sex which determines its selection objective.
Modifications are necessary to prevent a step size explosion. The experiments
revealed that the TSES is able to approximate the optimum in most of the
cases. Sometimes a performance win in comparison to DP or the DSES was
observed. An advantage of the TSES is that no infeasible starting points have
to be available at the beginning of the search. A modification, the TSES+
was introduced, which applies the two step selection operator to the sex o .
The TSES+ shows considerable performance improvements.
6. The proposed methods depend on new parameters, which have to be adapted
carefully. But this drawback can be weakened. Population sizes have to be
defined for almost every EA. The success of the TSES depends on the sex
ratios, but here we offer examples for successful population ratios. The DSES
only depends on two new parameters, and ϑ , which represent the speed
of the -reduction.
7. The NAES evolves the mutation ellipsoid rotation with a nested EA and
consequently shows a poor performance. Hence, we cannot recommend the
NAES for practical constraint handling. But the approach shows that the
rotation of the mutation ellipsoid improves the success rate and may be a
source for prospective constraint handling heuristics.