A python metaheuristic optimization library. Built for easy extension and usage.
Warning: Optimal is in beta. API may change. I will do my best to note any breaking changes in this readme, but no guarantee is given.
Supported metaheuristics:
- Genetic algorithms (GA)
- Gravitational search algorithm (GSA)
- Cross entropy (CE)
- Population-based incremental learning (PBIL)
pip install optimal
import math
from optimal import GenAlg
from optimal import Problem
from optimal import helpers
# The genetic algorithm uses binary solutions.
# A decode function is useful for converting the binary solution to real numbers
def decode_ackley(binary):
# Helpful functions from helpers are used to convert binary to float
# x1 and x2 range from -5.0 to 5.0
x1 = helpers.binary_to_float(binary[0:16], -5.0, 5.0)
x2 = helpers.binary_to_float(binary[16:32], -5.0, 5.0)
return x1, x2
# ackley is our fitness function
# This is how a user defines the goal of their problem
def ackley_fitness(solution):
x1, x2 = solution
# Ackley's function
# A common mathematical optimization problem
output = -20 * math.exp(-0.2 * math.sqrt(0.5 * (x1**2 + x2**2))) - math.exp(
0.5 * (math.cos(2 * math.pi * x1) + math.cos(2 * math.pi * x2))) + 20 + math.e
# You can prematurely stop the metaheuristic by returning True
# as the second return value
# Here, we consider the problem solved if the output is <= 0.01
finished = output <= 0.01
# Because this function is trying to minimize the output,
# a smaller output has a greater fitness
fitness = 1 / output
# First return argument must be a real number
# The higher the number, the better the solution
# Second return argument is a boolean, and optional
return fitness, finished
# Define a problem instance to optimize
# We can optionally include a decode function
# The optimizer will pass the decoded solution into your fitness function
# Additional fitness function and decode function parameters can also be added
ackley = Problem(ackley_fitness, decode_function=decode_ackley)
# Create a genetic algorithm with a chromosome size of 32,
# and use it to solve our problem
my_genalg = GenAlg(32)
best_solution = my_genalg.optimize(ackley)
print best_solution
Important notes:
- Fitness function must take solution as its first argument
- Fitness function must return a real number as its first return value
For further usage details, see comprehensive doc strings.
Renamed helpers.binary_to_int offset option to lower_bound, and renamed helpers.binary_to_float minimum and maximum options to lower_bound and upper_bound respectively.
Moved a number of options from Optimizer to Optimizer.optimize
Renamed common.random_solution_binary to common.random_binary_solution, and common.random_solution_real to common.random_real_solution
problem now an argument of Optimizer.optimize, instead of Optimizer.__init__.
max_iterations now an argument of Optimizer.optimize, instead of Optimizer.__init__.
Optimizer now takes a problem instance, instead of a fitness function and kwargs.
Library reorganized with greater reliance on __init__.py.
Optimizers can now be imported with:
from optimal import GenAlg, GSA, CrossEntropy
Etc.