can you explain the gep_simple method outputs-nevals?

[nickhuangxinyu]

can you explain the gep_simple method outputs-nevals?

[ShuhuaGao]

Hi, you can find details in this file. Nonetheless, I will make some brief explanation below.

• The first step is selection. N individuals are chosen as the offspring. Then, we apply mutation and crossover to the offspring.
• You can see from the _apply_crossover here that the recombination is not done for each possible pair of individuals (i.e., no permutation). Instead, crossover only happens between individuals 0 and 1, 2 and 3, 4 and 5, ..., in a sequential way (subject to the crossover probability). If, say, individuals A and B are recombined into two children A' and B', then A and B are replaced by A' and B' immediately in the offspring population.
• In the final offspring, if an individual is produced by mutation or crossover, then its fitness is labeled invalid and will be evaluated in the next generation (see this line). By contrast, if an individual is simply a copy of an original one, then there is no need to evaluate it again to reduce computational cost. Thus, the neval is just the number of invalid children that need reevaluation, which keeps changing among generations. If you are writing a paper and want to analyze the time complexity, simply view neval as N in the big-O notation.