next up previous print clean
Next: Parameter Encoding Up: Velocity Inversion Previous: Velocity Inversion

Description of the Micro Genetic Algorithm Program

The micro-genetic algorithm program used in this study is generation-based (that is, children replace the parents each generation except for those parents selected by elitism), uses binary encoding with a tournament selection for choosing the parents for the next generation and uniform crossover. Other operators such as jump and creep mutation and niching were not used, following the results in Alvarez (2002). Table 1 summarizes the parameters of the micro-genetic algorithm.
 
Table 1: Summary of the micro-genetic algorithm parameters
Population size 5 Crossover rate 0.95
Mutation rate 0.0 Creep mutation rate 0.0
Minimum model parameter value 2200 Maximum model parameter value 3200
Elitism Yes Niching No
Selection strategy Tournament Number of children 1


next up previous print clean
Next: Parameter Encoding Up: Velocity Inversion Previous: Velocity Inversion
Stanford Exploration Project
11/11/2002