The concept of messy chromosomes, first introduced by Goldberg for messy genetic algorithms, is applied in the method of simulated annealing. The messy chromosome is decoded into the input for an optimized function. Unlike simple chromosomes (vectors), messy chromosomes can have more than one possible value for a variable in a vector, or no value at all, in which case the value for input vector is copied from the best vector found so far. The resulting robust and efficient stochastic optimization method is suitable for finding correct minima of multimodal functions with global minimum separated by a barrier from other minima.
展开▼
