Fast Evolutionary Programming (FEP)

class pypop7.optimizers.ep.fep.FEP(problem, options)

Fast Evolutionary Programming with self-adaptive mutation of individual step-sizes (FEP).

Note

FEP was proposed mainly by Yao et al. in 1999 (the recipient of IEEE Evolutionary Computation Pioneer Award 2013 and IEEE Frank Rosenblatt Award 2020 ), where the classical Gaussian sampling distribution is replaced by the heavy-tailed Cachy distribution for better exploration on multi-modal black-box optimization problems.

Parameters:

• problem (dict) –

  problem arguments with the following common settings (keys):

  • ’fitness_function’ - objective function to be minimized (func),

  • ’ndim_problem’ - number of dimensionality (int),

  • ’upper_boundary’ - upper boundary of search range (array_like),

  • ’lower_boundary’ - lower boundary of search range (array_like).

• options (dict) –

  optimizer options with the following common settings (keys):

  • ’max_function_evaluations’ - maximum of function evaluations (int, default: np.Inf),

  • ’max_runtime’ - maximal runtime to be allowed (float, default: np.Inf),

  • ’seed_rng’ - seed for random number generation needed to be explicitly set (int);

  and with the following particular settings (keys):

  • ’sigma’ - initial global step-size, aka mutation strength (float),

  • ’n_individuals’ - number of offspring, aka offspring population size (int, default: 100),

  • ’q’ - number of opponents for pairwise comparisons (int, default: 10),

  • ’tau’ - learning rate of individual step-sizes self-adaptation (float, default: 1.0/np.sqrt(2.0*np.sqrt(problem[‘ndim_problem’]))),

  • ’tau_apostrophe’ - learning rate of individual step-sizes self-adaptation (float, default: 1.0/np.sqrt(2.0*problem[‘ndim_problem’]).

Examples

Use the optimizer FEP to minimize the well-known test function Rosenbrock:

>>> import numpy  # engine for numerical computing
>>> from pypop7.benchmarks.base_functions import rosenbrock  # function to be minimized
>>> from pypop7.optimizers.ep.fep import FEP
>>> problem = {'fitness_function': rosenbrock,  # to define problem arguments
...            'ndim_problem': 2,
...            'lower_boundary': -5.0*numpy.ones((2,)),
...            'upper_boundary': 5.0*numpy.ones((2,))}
>>> options = {'max_function_evaluations': 5000,  # to set optimizer options
...            'seed_rng': 2022,
...            'sigma': 3.0}  # global step-size may need to be tuned
>>> fep = FEP(problem, options)  # to initialize the optimizer class
>>> results = fep.optimize()  # to run its optimization/evolution process
>>> # to return the number of function evaluations and the best-so-far fitness
>>> print(f"FEP: {results['n_function_evaluations']}, {results['best_so_far_y']}")
FEP: 5000, 0.005781004466936902

For its correctness checking, refer to this code-based repeatability report for more details.

best_so_far_x

final best-so-far solution found during entire optimization.

Type:

array_like

best_so_far_y

final best-so-far fitness found during entire optimization.

Type:

array_like

n_individuals

number of offspring, aka offspring population size.

Type:

int

q

number of opponents for pairwise comparisons.

Type:

int

sigma

initial global step-size, aka mutation strength.

Type:

float

tau

self-adaptation learning rate of individual step-sizes.

Type:

float

tau_apostrophe

self-adaptation learning rate of individual step-sizes.

Type:

float

References

Yao, X., Liu, Y. and Lin, G., 1999. Evolutionary programming made faster. IEEE Transactions on Evolutionary Computation, 3(2), pp.82-102.

Chellapilla, K. and Fogel, D.B., 1999. Evolution, neural networks, games, and intelligence. Proceedings of the IEEE, 87(9), pp.1471-1496.

Bäck, T. and Schwefel, H.P., 1993. An overview of evolutionary algorithms for parameter optimization. Evolutionary Computation, 1(1), pp.1-23.