in_code_demo_1.py

"""
  In code demo for discrete euclidean domains
"""

import numpy as np
from dragonfly import load_config, maximise_function


def objective(x):
  """ Objective. """
  return np.linalg.norm(x[0])


def main():
  """ Main function. """
  size = 1000
  dim = 3
  disc_euc_items = list(np.random.random((size, dim)))
  domain_vars = [
    {'type': 'discrete_euclidean', 'items': disc_euc_items},
    ]
  config_params = {'domain': domain_vars}
  config = load_config(config_params)
  max_num_evals = 100

  # specify optimisation method
#   opt_method = 'bo' # Bayesian optimisation
  opt_method = 'ea' # evolutionary algorithm
#   opt_method = 'rand' # random search
  opt_val, opt_pt, history = maximise_function(objective, config.domain, max_num_evals,
                                               config=config, opt_method=opt_method)
  print(opt_pt, opt_val)


if __name__ == '__main__':
  main()