Robust stochastic search techniques that do not require knowledge of the objective function to be optimized and which can search through large spaces quickly. They are derived from processes in evolutionary and molecular biology such as natural selection. Their operators (viz., crossover, mutation and reproduction) are isomorphic with the synonymous biological processes. Instead of DNA or RNA strands, genetic algorithms usually process strings of symbols of finite length, with these symbols encoding the parameters to be optimized. As with neural networks, they are based on an analogy with nature, in this case that the best algorithms ‘breed’ with each other to provide new variants in a ‘survival of the fittest’. They have a wide range of application beyond biological processes such as the traveling salesman problem, gas pipeline control, the design of aircraft, neural net architecture, models of international security, and strategy formulation.