I understand that ECJ's postEvalationStatistics() provides best individual's fitness, its genotype, evaluated tag by each subpopulation for every generations and eventually reports the best individual in the run. So, there seem to be a difference in generation 0 and other onward generations in terms of fitness. Because generation 0 provides those information without doing any genetic operations (like selection, mutation, cross-over etc.) while the other generation does. So,
-If population size, genotype encoding etc. parameters are same for two algorithms : A, B, then will we not get around same average objective fitness for both algorithms in generation 0; over a number say 50 independent trials? Objective fitness is the sum of genes in individual's genotype.
Thanks a lot.