Welcome to delo’s documentation!

Contents:

delo

Differential Evolution (DE) optimization algorithms perform satisfactorily even on complex problems in higher dimensionality. However, it is difficult to a priori choose optimal parameters. In this package, we propose DElo (DE with adaptation based on Elo rating system). Elo rating, originally used in chess, is a way to measure dynamic fitness.

Installation

Open terminal and run command:

pip install delo

Example 1

 1from delo import DElo, DescribedFunction
 2import numpy as np
 3
 4def square(x):
 5    return np.sum(x ** 2, axis=1)
 6described_function = DescribedFunction(square, dimension=2,
 7                                       domain_lower_limit=-10,
 8                                       domain_upper_limit=10)
 9algorithm = DElo(10)
10algorithm.optimize(described_function)

Example 2

 1from delo import DElo, DescribedFunction
 2import numpy as np
 3
 4def my_single_argument_function(x):
 5    return np.sum(x ** 2)
 6
 7def my_multi_argument_wrapping(x):
 8    return np.array([my_single_argument_function(xi) for xi in x])
 9
10described_my_function = delo.DescribedFunction(my_multi_argument_wrapping,
11                                               dimension=5,
12                                               domain_lower_limit=-5,
13                                               domain_upper_limit=5)
14algorithm = delo.DElo(100)
15algorithm.optimize(described_my_function, max_f_evals=10000)

Contribute

  • Issue Tracker: github.com/MrDomani/delo/issues

  • Source Code: github.com/MrDomani/delo

License

The project is licensed under the MIT license.