What is the significance of the mutation rate in genetic algorithms?

Question

Hello. Let’s say I set a mutation rate of 5% in an AG. What does that mean?

• What is it about the entire population, and for every individual, I’m going to go through all of his genes, and I make a change with a 5% probability? Here, the entire population will have about 5% of their genes altered.
• What about the entire population, and for each individual, I change a random gene of his with a 5% probability? Here, at the end of the mutation, about 5% of the population will have an altered gene.

Another thing... conventionally, is the mutation applied to the entire population, including the selected parents, or just the offspring? The second option is biologically more plausible.

Answer 1

The mutation rate refers to the number of individuals in the population who will be mutated.

It is usually very small and serves to maintain the genetic variability of chromosomes over generations.

You can of course referee another parameter in your implementation that controls the percentage of genes that will be mutated during the process.

Regarding the other question, in general, the mutation comes right after the crossing. Thus, it acts only on the offspring.

I suggest the Manual of Evolutionary Computing and Metaheuristics as an introduction to the theme.