G e n e t i c   A l g o r i t h m s   O n

N e t a d e l i c a

Genetic algorithms (GAs) simulate the process of biological evolution on a computer. They are used to find highly fit solutions to problems where the possible solutions are too numerous to be evaluated individually.

You work with a population of individuals ("members"). Each member comprises a set of genes (numbers of arbitrary base, often 2). The process is roughly as follows:

  1. Create an initial population of randomly-generated individuals
  2. Evaluate each member's fitness ("fitness" depends entirely on the task in hand; a member's fitness might be how many 1s its genes contain, or how well it plays chess using its genes as "weights" for various evaluation parameters)
  3. Kill the bottom x% (least fit) of the population (often 50%)
  4. Let the fittest members breed ("reproduction"):
  5. Choose two members for breeding ("selection") - lots of variations here: e.g. either select two at random ("uniform"), or weight the selection according to a member's fitness ("fitness-proportionate")
  6. Breed them ("crossover" - more options here: e.g., a child can be formed by single-point crossover (the parents' genes are swapped over at some random point along their chromosome), two-point crossover (the parents' genes are swapped over at two random points), uniform crossover (the parents' genes are selected bit-by-bit randomly), and weighted crossover (the parents' genes are selected bit-by-bit randomly but weighted according to each parents' fitness)
  7. Apply mutation - each gene of the child is subject to a small chance of mutating to a different allele (an allele is a possible value for a gene, in binary, the alleles are "0" and "1") - a common mutation rate is 0.001
  8. The children form the new population of members
  9. Go to step 2 until you've evolved a suitably fit member or population

Back to the Netadelica GA home page

Back to the Netadelica home page