phamasaur / stochsearch

Stochastic search algorithms using python multiprocessing

Stochastic Search

This Python stochastic search package, stochsearch, includes an implementation of algorithms such as evolutionary algorithm, microbial genetic algorithm, and lamarckian evolutionary algorithm, using the Python pathos multiprocessing framework. Fitness evaluation of individuals in a population is carried out in parallel across CPUs in a multiprocessing pool with the number of processes defined by the user or by os.cpu_count() of the system. Read below for installation and usage instructions.

Installation

$ pip install stochsearch

Requirements: numpy, pathos

Usage

This section illustrates how to use this package for evolutionary search. It is similar for other search methods. The only items that may change are the parameters of the search. See this for a description and list of parameters for each search method.

Importing evolutionary search

from stochsearch import EvolSearch

Setup parameters for evolutionary search using a dictionary as follows

evol_params = {
    'num_processes' : 4, # (optional) number of proccesses for multiprocessing.Pool
    'pop_size' : 100,    # population size
    'genotype_size': 10, # dimensionality of solution
    'fitness_function': fitness_function, # custom function defined to evaluate fitness of a solution
    'elitist_fraction': 0.04, # fraction of population retained as is between generations
    'mutation_variance': 0.05, # mutation noise added to offspring.
    'fitness_args': np.arange(100), # (optional) fitness_function *argv, len(list) should be 1 or pop_size
}

Define a function that takes a genotype as argument and returns the fitness value for that genotype - passed as the ‘fitness_function’ key in the evol_params dictionary.

Create an evolutionary search object

es = EvolSearch(evol_params)

Executing the search

Option 1: Run the search for a certain number of generations

num_gens = 100
es.execute_search(num_gens)

Option 2: Step through the generations based on a condition

max_num_gens = 100
gen = 0
desired_fitness = 0.9
while es.get_best_individual_fitness() < desired_fitness and gen < max_num_gens:
        print("Gen #{} Best Fitness = {}".format(gen, es.get_best_individual_fitness()))
        es.step_generation()
        gen += 1

Accessing results

print('Max fitness of population = ',es.get_best_individual_fitness())
print('Best individual in population = ',es.get_best_individual())

See demos folder for a sample script.