Genetic algorithm for Python 3.
This algorithm is simply a class which can be directly implemented in your project.
You will need the following packages to run this algorithm (also listed in src/gapy/requirements.txt)
numpy
pandas
scipy
Pandas is only used for aesthetic reasons on the write method for the printing and is entirely optional. To remove this feature, simply comment out the method and set logging to False.
A testing version is available in PyPI:
pip3 install genapy
The file test.py presents an example on how to use this algorithm. It requires a fitness function, which must receive a numpy array containing the population on each line, each composed of real numbers, and output a numpy vector with the fitness of each individual as a real number.
Probabilities for crossover or mutation are numbers in the interval [0,1] and set by the parameters crossover and mutation respectively. They both default to 0.5.
The output range of each parameter used to make up the individual can be scaled separately by passing a numpy array containing the upper and lower bounds in the shape (k,2), where k is the number of parameters (chromosome_size), and also setting has_mask to True.
The number of bits for each parameter can also be set as an integer in resolution.
A printing delay for the write method is set by passing the desired value in seconds on time_print.