genetic-algorithm-matlab
A very simple Genetic Algorithm implementation for matlab, easy to use, easy to modify and runs fast. Even has some visualization too.
To Run
Run the FunctionOptimization script.
To Modify Optimization Function
Replace your own function into EvaluateIndividual.m script. Note that this genetic algorithm tries to maximise the output so invert your function according to your needs. Right now it tries to locate the peak of a double variable function. It can be adjusted to optimize for more than two variable functions.
To Modify Genetic Algorithm Parameters
- All the parameters are located in the FunctionOptimization.m script.
- populationSize -> number of individuals in a population
- numberOfGenes -> number of bits per chromosome
- crossoverProbability -> probability that a crossover will happen between two individuals
- mutationProbability -> probability that a mutation will occur in an individual
- tournamentSelectionParameter -> parameter that's used to calculate the probabilities for individuals to be chosen in a tournament -> 'p*(1-p)^k' where k denotes the k'th worst individual in the tournament pool
- variableRange -> the range in which the genes will be decoded into. basically minimum and maximum values of the parameters
- numberOfGenerations -> number of iterations to run genetic algorithm
- numberOfVariables -> number of variables stored in one chromosome
- tournamentSize -> this value determines the number of individuals to be taken into a tournament. an individual of this pool is then chosen for mating with a probability calculated from tournamentSelectionParameter
- numberOfReplications -> after a generation is run, this number of best individuals are copied back into the population to ensure the solution quality does not degrade
- verbose -> if true; progress is printed
- draw_plots -> if true; progress is plotted
Unit Tests
They are simply there to test the individual methods/steps of the genetic algorithm. Can be used for debugging.
Licensing Stuff
Please dont remove my name from the codes.