Latent Action Monte Carlo Tree Search (LAMCTS)

class pypop7.optimizers.bo.lamcts.LAMCTS(problem, options)[source]

Latent Action Monte Carlo Tree Search (LAMCTS).

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):
    • ’n_individuals’ - number of individuals/samples (int, default: 100),

    • ’c_e’ - factor to control exploration (float, default: 0.01),

    • ’leaf_size’ - leaf size (int, default: 40).

Examples

Use the black-box optimizer LAMCTS to minimize the well-known test function Rosenbrock:

 1>>> import numpy
 2>>> from pypop7.benchmarks.base_functions import rosenbrock  # function to be minimized
 3>>> from pypop7.optimizers.bo.lamcts import LAMCTS
 4>>> problem = {'fitness_function': rosenbrock,  # to define problem arguments
 5...            'ndim_problem': 2,
 6...            'lower_boundary': -5.0*numpy.ones((2,)),
 7...            'upper_boundary': 5.0*numpy.ones((2,))}
 8>>> options = {'max_function_evaluations': 5000,  # to set optimizer options
 9...            'seed_rng': 1}
10>>> lamcts = LAMCTS(problem, options)  # to initialize the optimizer class
11>>> results = lamcts.optimize()  # to run the optimization process
12>>> print(f"LAMCTS: {results['n_function_evaluations']}, {results['best_so_far_y']}")
13LAMCTS: 5000, 0.0001

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

c_e

factor to control exploration.

Type:

float

init_individuals

number of initial individuals.

Type:

int

leaf_size

leaf size.

Type:

int

n_individuals

number of individuals/samples.

Type:

int

References

Wang, L., Fonseca, R. and Tian, Y., 2020. Learning search space partition for black-box optimization using monte carlo tree search. Advances in Neural Information Processing Systems, 33, pp.19511-19522. https://arxiv.org/abs/2007.00708 (an updated version) https://proceedings.neurips.cc/paper/2020/hash/e2ce14e81dba66dbff9cbc35ecfdb704-Abstract.html (the original version)

https://github.com/facebookresearch/LA-MCTS (an updated version) https://github.com/facebookresearch/LaMCTS (the original version)