/
res.py
185 lines (159 loc) · 9.58 KB
/
res.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
import numpy as np # engine for numerical computing
from pypop7.optimizers.es.es import ES
class RES(ES):
"""Rechenberg's (1+1)-Evolution Strategy with 1/5th success rule (RES).
.. note:: `RES` is the first evolution strategy with self-adaptation of the *global* step-size (designed
by Rechenberg, one recipient of `IEEE Evolutionary Computation Pioneer Award 2002
<https://tinyurl.com/456as566>`_). As theoretically investigated in his *seminal* Ph.D. dissertation
at Technical University of Berlin, the existence of narrow **evolution window** explains the necessarity
of *global* step-size adaptation to maximize the local convergence progress, if possible.
Since there is only one parent and only one offspring for each generation (iteration), `RES` generally
shows limited *exploration* ability for large-scale black-box optimization (LBO). Therefore, it is
recommended to first attempt more advanced ES variants (e.g., `LMCMA`, `LMMAES`) for LBO. Here we include
`RES` (AKA two-membered ES) mainly for *benchmarking* and *theoretical* purposes.
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`),
* 'mean' - initial (starting) point, aka mean of Gaussian search distribution (`array_like`),
* if not given, it will draw a random sample from the uniform distribution whose search range is
bounded by `problem['lower_boundary']` and `problem['upper_boundary']`.
* 'lr_sigma' - learning rate of global step-size self-adaptation (`float`, default:
`1.0/np.sqrt(problem['ndim_problem'] + 1.0)`).
Examples
--------
Use the black-box optimizer `RES` to minimize the well-known test function
`Rosenbrock <http://en.wikipedia.org/wiki/Rosenbrock_function>`_:
.. code-block:: python
:linenos:
>>> import numpy # engine for numerical computing
>>> from pypop7.benchmarks.base_functions import rosenbrock # function to be minimized
>>> from pypop7.optimizers.es.res import RES
>>> 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,
... 'mean': 3.0*numpy.ones((2,)),
... 'sigma': 3.0} # global step-size may need to be tuned
>>> res = RES(problem, options) # to initialize the black-box optimizer class
>>> results = res.optimize() # to run its optimization/evolution process
>>> # to return the number of function evaluations and the best-so-far fitness
>>> print(f"RES: {results['n_function_evaluations']}, {results['best_so_far_y']}")
RES: 5000, 0.00011689296624022443
For its correctness checking of coding, refer to `this code-based repeatability report
<https://tinyurl.com/5n6ndrn7>`_ for more details.
Attributes
----------
best_so_far_x : `array_like`
final best-so-far solution found during entire optimization.
best_so_far_y : `array_like`
final best-so-far fitness found during entire optimization.
lr_sigma : `float`
learning rate of global step-size self-adaptation.
mean : `array_like`
initial (starting) point, aka mean of Gaussian search distribution.
sigma : `float`
final global step-size, aka mutation strength (updated during optimization).
References
----------
Auger, A., Hansen, N., López-Ibáñez, M. and Rudolph, G., 2022.
`Tributes to Ingo Rechenberg (1934--2021).
<https://dl.acm.org/doi/10.1145/3511282.3511283>`_
ACM SIGEVOlution, 14(4), pp.1-4.
Agapie, A., Solomon, O. and Giuclea, M., 2021.
`Theory of (1+1) ES on the RIDGE.
<https://ieeexplore.ieee.org/abstract/document/9531957>`_
IEEE Transactions on Evolutionary Computation, 26(3), pp.501-511.
Hansen, N., Arnold, D.V. and Auger, A., 2015.
`Evolution strategies.
<https://link.springer.com/chapter/10.1007%2F978-3-662-43505-2_44>`_
In Springer Handbook of Computational Intelligence (pp. 871-898). Springer, Berlin, Heidelberg.
Kern, S., Müller, S.D., Hansen, N., Büche, D., Ocenasek, J. and Koumoutsakos, P., 2004.
`Learning probability distributions in continuous evolutionary algorithms–a comparative review.
<https://link.springer.com/article/10.1023/B:NACO.0000023416.59689.4e>`_
Natural Computing, 3, pp.77-112.
Beyer, H.G. and Schwefel, H.P., 2002.
`Evolution strategies–A comprehensive introduction.
<https://link.springer.com/article/10.1023/A:1015059928466>`_
Natural Computing, 1(1), pp.3-52.
Rechenberg, I., 2000.
`Case studies in evolutionary experimentation and computation.
<https://www.sciencedirect.com/science/article/pii/S0045782599003813>`_
Computer Methods in Applied Mechanics and Engineering, 186(2-4), pp.125-140.
Rechenberg, I., 1989.
`Evolution strategy: Nature’s way of optimization.
<https://link.springer.com/chapter/10.1007/978-3-642-83814-9_6>`_
In Optimization: Methods and Applications, Possibilities and Limitations (pp. 106-126).
Springer, Berlin, Heidelberg.
Rechenberg, I., 1984.
`The evolution strategy. A mathematical model of darwinian evolution.
<https://link.springer.com/chapter/10.1007/978-3-642-69540-7_13>`_
In Synergetics—from Microscopic to Macroscopic Order (pp. 122-132). Springer, Berlin, Heidelberg.
Schumer, M.A. and `Steiglitz, K. <https://www.cs.princeton.edu/~ken/>`_, 1968.
`Adaptive step size random search.
<https://ieeexplore.ieee.org/abstract/document/1098903>`_
IEEE Transactions on Automatic Control, 13(3), pp.270-276.
"""
def __init__(self, problem, options):
options['n_parents'] = 1 # mandatory setting
options['n_individuals'] = 1 # mandatory setting
ES.__init__(self, problem, options)
if self.lr_sigma is None:
self.lr_sigma = 1.0/np.sqrt(self.ndim_problem + 1.0)
assert self.lr_sigma > 0, f'`self.lr_sigma` = {self.lr_sigma}, but should > 0.'
def initialize(self, args=None, is_restart=False):
mean = self._initialize_mean(is_restart) # mean of Gaussian search distribution
y = self._evaluate_fitness(mean, args) # fitness
best_so_far_y = np.copy(y)
self._list_initial_mean.append(np.copy(mean))
return mean, y, best_so_far_y
def iterate(self, args=None, mean=None): # to sample and evaluate only one offspring
x = mean + self.sigma*self.rng_optimization.standard_normal((self.ndim_problem,))
y = self._evaluate_fitness(x, args)
return x, y
def restart_reinitialize(self, args=None, mean=None, y=None, best_so_far_y=None, fitness=None):
if not self.is_restart:
return mean, y, best_so_far_y
self._list_fitness.append(best_so_far_y)
is_restart_1, is_restart_2 = self.sigma < self.sigma_threshold, False
if len(self._list_fitness) >= self.stagnation:
is_restart_2 = (self._list_fitness[-self.stagnation] - self._list_fitness[-1]) < self.fitness_diff
is_restart = bool(is_restart_1) or bool(is_restart_2)
if is_restart:
self._print_verbose_info(fitness, y, True)
if self.verbose:
print(' ....... *** restart *** .......')
self._n_restart += 1
self._list_generations.append(self._n_generations) # for each restart
self._n_generations = 0
self.sigma = np.copy(self._sigma_bak)
mean, y, best_so_far_y = self.initialize(args, True)
self._list_fitness = [best_so_far_y]
return mean, y, best_so_far_y
def optimize(self, fitness_function=None, args=None): # for all generations (iterations)
fitness = ES.optimize(self, fitness_function)
mean, y, best_so_far_y = self.initialize(args)
while not self.termination_signal:
self._print_verbose_info(fitness, y)
x, y = self.iterate(args, mean)
self._n_generations += 1
if self._check_terminations():
break
self.sigma *= np.power(np.exp(float(y < best_so_far_y) - 0.2), self.lr_sigma)
if y <= best_so_far_y:
mean, best_so_far_y = x, y
mean, y, best_so_far_y = self.restart_reinitialize(args, mean, y, best_so_far_y, fitness)
return self._collect(fitness, y, mean)