next up previous
Next: Initial Experiments Up: Evolving Brick structures Previous: Mutation and Crossover

  
Evolutionary Algorithm

We use a plain steady-state genetic algorithm, initialized with a population of one single brick. Through mutation and crossover, a population of 1000 individuals is generated and then evolved:

1.
While maximum fitness < Target fitness
2.
Do

(a)
Randomly select mutation or crossover.
(b)
Select 1 (for mutation) or 2 (for crossover) random individual(s) with fitness proportional probability.
(c)
Apply mutation or crossover operator
(d)
Generate physical model and test for gravitational load
(e)
If the new model will support its own weight

i.
Then replace a random individual with it (chosen with inverse fitness proportional probability).



Pablo Funes
2001-05-08