Easy gene is a simple way to introduce genetic algorythms into your project. With a total line count of 37 lines. Insperation for this project was https://www.youtube.com/watch?v=RxTfc4JLYKs.
Lets say we are trying to create a ideal animal. In the real world the score_animal(animal) function is unknown. We have two parameters, arm_length and torso_length. In this example we start with a population of 8 animals with distributed values. We use mutation rates of 0.2 for both parameters, save two parents every generation, and evolve 100 times.
from easy_gene import evolve
def score_animal(animal): # max score is when the value is 3, 6
return 1 / (abs(animal['arm_length'] - 3) + abs(animal['torso_length'] - 6) + 1)
population = [
{'arm_length': 2.0, 'torso_length': 2.0},
{'arm_length': 6.0, 'torso_length': 7.0},
{'arm_length': 4.0, 'torso_length': 2.0},
{'arm_length': 9.0, 'torso_length': 5.0},
{'arm_length': 4.0, 'torso_length': 2.0},
{'arm_length': 9.0, 'torso_length': 5.0},
{'arm_length': 4.0, 'torso_length': 2.0},
{'arm_length': 9.0, 'torso_length': 5.0},
]
mutation_rates = {'arm_length': 0.2, 'torso_length': 0.2}
num_parents = 2
iterations = 100
max_scores = []
for _ in range(iterations):
scores = [score_animal(animal) for animal in population]
population = evolve(population, scores, num_parents, mutation_rates)
max_scores.append(max(scores))
print(population[0])result:
{'arm_length': 3.1127586555075375, 'torso_length': 6.533470856616043}
max scores:
- x axis = iteration
- y axis = score
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent