GA Exercise (more)

If you have done the GA theory, you can now understand the code of this GA program.

Click to run World: GA (Finnegans Wake) at Ancient Brain.

• Clone and Edit this World.
• Turn the debug info on and off to see how it works.

Population size

• Change population size to:
  • 1
  • 2
  • 10
  • 10000.
  What happens? What is the best value?

Mutation rate

• Change mutation rate to:
  • 0
  • 0.0001
  • 1
  What happens? What is the best value?

• Q. If mutation rate = 0 but you have crossover, why does it still not work?

High mutation rate

• Try a very high mutation rate, like 0.5
• This seems to work fine on short strings, but fails totally on long strings.
• Why? I am not sure myself.

POWERBASE

• POWERBASE is explained in the code. Change POWERBASE to:
  • 1
  • 1.1
  • 1.5
  • 10
  What happens? What is the best value?

---

Exercises

• Instead of always doing crossover, have a probability of crossover. Make 2 children at a time. If no crossover, we get clones of the 2 parents. Edit the code and test it.

• Make probability of mutation start high and decline. Same with crossover. Does this help?