Designing a competent simple genetic algorithm for search and optimization

Abstract

Simple genetic algorithms have been used to solve many water resources problems, but specifying the parameters that control how adaptive search is performed can be a difficult and time‐consuming trial‐and‐error process. However, theoretical relationships for population sizing and timescale analysis have been developed that can provide pragmatic tools for vastly limiting the number of parameter combinations that must be considered. The purpose of this technical note is to summarize these relationships for the water resources community and to illustrate their practical utility in a long‐term groundwater monitoring design application. These relationships, which model the effects of the primary operators of a simple genetic algorithm (selection, recombination, and mutation), provide a highly efficient method for ensuring convergence to near‐optimal or optimal solutions. Application of the method to a monitoring design test case identified robust parameter values using only three trial runs.