
153Design Optimization
such as searching only in the vicinity of the original solution in addition to
or in place of random search over the whole design space. Understandably,
many implementations are possible:
1. Randomly searching the whole feasible set. This becomes problem-
atic in higher-dimensional problems.
2. Weighting the probability density function of the search locations
based on where solutions are likely to be found.
3. Starting with searching the whole design space and then narrowing
down to “good areas.”
4. Perturbing good solutions to improve upon them.
5. Working with a set of solutions instead of one and performing ran ...