SimonBlanke /
Gradient-Free-Optimizers
| Total Complexity | 12 |
| Total Lines | 98 |
| Duplicated Lines | 11.22 % |
| Changes | 0 | ||
Duplicate code is one of the most pungent code smells. A rule that is often used is to re-structure code once it is duplicated in three or more places.
Common duplication problems, and corresponding solutions are:
| 1 | # Author: Simon Blanke |
||
| 2 | # Email: [email protected] |
||
| 3 | # License: MIT License |
||
| 4 | |||
| 5 | import numpy as np |
||
| 6 | |||
| 7 | from .base_population_optimizer import BasePopulationOptimizer |
||
| 8 | from ._spiral import Spiral |
||
| 9 | |||
| 10 | |||
| 11 | View Code Duplication | def centeroid(array_list): |
|
|
Duplication
introduced
by
|
|||
| 12 | centeroid = [] |
||
| 13 | for idx in range(array_list[0].shape[0]): |
||
| 14 | center_dim_pos = [] |
||
| 15 | for array in array_list: |
||
| 16 | center_dim_pos.append(array[idx]) |
||
| 17 | |||
| 18 | center_dim_mean = np.array(center_dim_pos).mean() |
||
| 19 | centeroid.append(center_dim_mean) |
||
| 20 | |||
| 21 | return centeroid |
||
| 22 | |||
| 23 | |||
| 24 | class SpiralOptimization(BasePopulationOptimizer): |
||
| 25 | name = "Spiral Optimization" |
||
| 26 | _name_ = "spiral_optimization" |
||
| 27 | __name__ = "SpiralOptimization" |
||
| 28 | |||
| 29 | optimizer_type = "population" |
||
| 30 | computationally_expensive = False |
||
| 31 | |||
| 32 | def __init__( |
||
| 33 | self, |
||
| 34 | search_space, |
||
| 35 | initialize={"grid": 4, "random": 2, "vertices": 4}, |
||
| 36 | constraints=[], |
||
| 37 | random_state=None, |
||
| 38 | rand_rest_p=0, |
||
| 39 | nth_process=None, |
||
| 40 | population=10, |
||
| 41 | decay_rate=0.99, |
||
| 42 | ): |
||
| 43 | super().__init__( |
||
| 44 | search_space=search_space, |
||
| 45 | initialize=initialize, |
||
| 46 | constraints=constraints, |
||
| 47 | random_state=random_state, |
||
| 48 | rand_rest_p=rand_rest_p, |
||
| 49 | nth_process=nth_process, |
||
| 50 | ) |
||
| 51 | |||
| 52 | self.population = population |
||
| 53 | self.decay_rate = decay_rate |
||
| 54 | |||
| 55 | self.particles = self._create_population(Spiral) |
||
| 56 | self.optimizers = self.particles |
||
| 57 | |||
| 58 | @BasePopulationOptimizer.track_new_pos |
||
| 59 | def init_pos(self): |
||
| 60 | nth_pop = self.nth_trial % len(self.particles) |
||
| 61 | |||
| 62 | self.p_current = self.particles[nth_pop] |
||
| 63 | self.p_current.decay_rate = self.decay_rate |
||
| 64 | |||
| 65 | return self.p_current.init_pos() |
||
| 66 | |||
| 67 | def finish_initialization(self): |
||
| 68 | self.sort_pop_best_score() |
||
| 69 | self.center_pos = self.pop_sorted[0].pos_current |
||
| 70 | self.center_score = self.pop_sorted[0].score_current |
||
| 71 | |||
| 72 | self.search_state = "iter" |
||
| 73 | |||
| 74 | @BasePopulationOptimizer.track_new_pos |
||
| 75 | def iterate(self): |
||
| 76 | while True: |
||
| 77 | self.p_current = self.particles[ |
||
| 78 | self.nth_trial % len(self.particles) |
||
| 79 | ] |
||
| 80 | |||
| 81 | self.sort_pop_best_score() |
||
| 82 | self.p_current.global_pos_best = self.pop_sorted[0].pos_current |
||
| 83 | |||
| 84 | pos_new = self.p_current.move_spiral(self.center_pos) |
||
| 85 | |||
| 86 | if self.conv.not_in_constraint(pos_new): |
||
| 87 | return pos_new |
||
| 88 | return self.p_current.iterate() |
||
| 89 | |||
| 90 | @BasePopulationOptimizer.track_new_score |
||
| 91 | def evaluate(self, score_new): |
||
| 92 | if self.search_state == "iter": |
||
| 93 | if self.pop_sorted[0].score_current > self.center_score: |
||
| 94 | self.center_pos = self.pop_sorted[0].pos_current |
||
| 95 | self.center_score = self.pop_sorted[0].score_current |
||
| 96 | |||
| 97 | self.p_current.evaluate(score_new) |
||
| 98 |
Loading history...