gradient_free_optimizers.optimizers.exp_opt.random_annealing

Complexity

Total Complexity

2

Size/Duplication

Total Lines	38
Duplicated Lines	0 %

Importance

Changes

0

2 Methods

Rating	Name	Duplication	Size	Complexity
A	RandomAnnealingOptimizer.iterate()	0	7	1
A	RandomAnnealingOptimizer.__init__()	0	17	1

# Author: Simon Blanke
# Email: simon.blanke@yahoo.com
# License: MIT License


from ..local_opt import HillClimbingOptimizer
from ...search import Search


class RandomAnnealingOptimizer(HillClimbingOptimizer, Search):
    name = "Random Annealing"

    def __init__(
        self,
        *args,
        epsilon=0.03,
        distribution="normal",
        n_neighbours=3,
        annealing_rate=0.98,
        start_temp=10,
        **kwargs,
    ):
        super().__init__(*args, **kwargs)
        self.epsilon = epsilon
        self.distribution = distribution
        self.n_neighbours = n_neighbours
        self.annealing_rate = annealing_rate
        self.start_temp = start_temp
        self.temp = start_temp

    @HillClimbingOptimizer.track_nth_iter
    @HillClimbingOptimizer.random_restart
    def iterate(self):
        pos = self._move_climb(self.pos_current, epsilon_mod=self.temp)
        self.temp = self.temp * self.annealing_rate

        return pos