Convergence

Advanced programming

Convergence in terms of genetic algorithms is a special case when a genetic algorithm needs to stop due to the fact that every identity in the population is identical. There is full convergence and premature convergence. Full convergence can be seen in algorithms only using cross-over. Premature convergence occurs when a population has converged to a single solution, but that solution is not as high of quality as expected, for example the population has gotten stuck. To avoid convergence, a variety of diversity generating techniques can be used. Convergence does not always indicate a negative sign.