-
Notifications
You must be signed in to change notification settings - Fork 2
/
sobol.py
706 lines (620 loc) · 18.2 KB
/
sobol.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
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
"""
Authors:
Original FORTRAN77 version of i4_sobol by Bennett Fox.
MATLAB version by John Burkardt.
PYTHON version by Corrado Chisari
Original Python version of is_prime by Corrado Chisari
Original MATLAB versions of other functions by John Burkardt.
PYTHON versions by Corrado Chisari
Modified Python version by Holger Nahrstaedt
Original code is available from
http://people.sc.fsu.edu/~jburkardt/py_src/sobol/sobol.html
"""
import warnings
import numpy as np
from sklearn.utils import check_random_state
from ..space import Space
from .base import InitialPointGenerator
class Sobol(InitialPointGenerator):
"""Generates a new quasirandom Sobol' vector with each call.
Parameters
----------
skip : int
Skipped seed number.
randomize : bool, default=False
When set to True, random shift is applied.
Notes
-----
Sobol' sequences [1]_ provide :math:`n=2^m` low discrepancy points in
:math:`[0,1)^{dim}`. Scrambling them makes them suitable for singular
integrands, provides a means of error estimation, and can improve their
rate of convergence.
There are many versions of Sobol' sequences depending on their
'direction numbers'. Here, the maximum number of dimension is 40.
The routine adapts the ideas of Antonov and Saleev [2]_.
.. warning::
Sobol' sequences are a quadrature rule and they lose their balance
properties if one uses a sample size that is not a power of 2, or skips
the first point, or thins the sequence [5]_.
If :math:`n=2^m` points are not enough then one should take :math:`2^M`
points for :math:`M>m`. When scrambling, the number R of independent
replicates does not have to be a power of 2.
Sobol' sequences are generated to some number :math:`B` of bits. Then
after :math:`2^B` points have been generated, the sequence will repeat.
Currently :math:`B=30`.
References
----------
.. [1] I. M. Sobol. The distribution of points in a cube and the accurate
evaluation of integrals. Zh. Vychisl. Mat. i Mat. Phys., 7:784-802,
1967.
.. [2] Antonov, Saleev,
USSR Computational Mathematics and Mathematical Physics,
Volume 19, 1980, pages 252 - 256.
.. [3] Paul Bratley, Bennett Fox,
Algorithm 659:
Implementing Sobol's Quasirandom Sequence Generator,
ACM Transactions on Mathematical Software,
Volume 14, Number 1, pages 88-100, 1988.
.. [4] Bennett Fox,
Algorithm 647:
Implementation and Relative Efficiency of Quasirandom
Sequence Generators,
.. [5] Art B. Owen. On dropping the first Sobol' point. arXiv 2008.08051,
2020.
"""
def __init__(self, skip=0, randomize=True):
if not (skip & (skip - 1) == 0):
raise ValueError(
"The balance properties of Sobol' points require"
" skipping a power of 2."
)
if skip != 0:
warnings.warn(
f"{skip} points have been skipped: "
f"{skip} points can be generated before the "
f"sequence repeats."
)
self.skip = skip
self.num_generated = 0
self.randomize = randomize
self.dim_max = 40
self.log_max = 30
self.atmost = 2**self.log_max - 1
self.lastq = None
self.maxcol = None
self.poly = None
self.recipd = None
self.seed_save = -1
self.v = np.zeros((self.dim_max, self.log_max))
self.dim_num_save = -1
def init(self, dim_num):
self.dim_num_save = dim_num
self.v = np.zeros((self.dim_max, self.log_max))
self.v[0:40, 0] = np.transpose(
[
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
1,
]
)
self.v[2:40, 1] = np.transpose(
[
1,
3,
1,
3,
1,
3,
3,
1,
3,
1,
3,
1,
3,
1,
1,
3,
1,
3,
1,
3,
1,
3,
3,
1,
3,
1,
3,
1,
3,
1,
1,
3,
1,
3,
1,
3,
1,
3,
]
)
self.v[3:40, 2] = np.transpose(
[
7,
5,
1,
3,
3,
7,
5,
5,
7,
7,
1,
3,
3,
7,
5,
1,
1,
5,
3,
3,
1,
7,
5,
1,
3,
3,
7,
5,
1,
1,
5,
7,
7,
5,
1,
3,
3,
]
)
self.v[5:40, 3] = np.transpose(
[
1,
7,
9,
13,
11,
1,
3,
7,
9,
5,
13,
13,
11,
3,
15,
5,
3,
15,
7,
9,
13,
9,
1,
11,
7,
5,
15,
1,
15,
11,
5,
3,
1,
7,
9,
]
)
self.v[7:40, 4] = np.transpose(
[
9,
3,
27,
15,
29,
21,
23,
19,
11,
25,
7,
13,
17,
1,
25,
29,
3,
31,
11,
5,
23,
27,
19,
21,
5,
1,
17,
13,
7,
15,
9,
31,
9,
]
)
self.v[13:40, 5] = np.transpose(
[
37,
33,
7,
5,
11,
39,
63,
27,
17,
15,
23,
29,
3,
21,
13,
31,
25,
9,
49,
33,
19,
29,
11,
19,
27,
15,
25,
]
)
self.v[19:40, 6] = np.transpose(
[
13,
33,
115,
41,
79,
17,
29,
119,
75,
73,
105,
7,
59,
65,
21,
3,
113,
61,
89,
45,
107,
]
)
self.v[37:40, 7] = np.transpose([7, 23, 39])
# Set POLY.
self.poly = [
1,
3,
7,
11,
13,
19,
25,
37,
59,
47,
61,
55,
41,
67,
97,
91,
109,
103,
115,
131,
193,
137,
145,
143,
241,
157,
185,
167,
229,
171,
213,
191,
253,
203,
211,
239,
247,
285,
369,
299,
]
# Find the number of bits in ATMOST.
self.maxcol = _bit_hi1(self.atmost)
# Initialize row 1 of V.
self.v[0, 0 : self.maxcol] = 1
# Check parameters.
if dim_num < 1 or self.dim_max < dim_num:
raise ValueError(
f'I4_SOBOL - Fatal error!\n'
f' The spatial dimension DIM_NUM should '
f'satisfy:\n'
f' 1 <= DIM_NUM <= {self.dim_max}\n'
f' But this input value is DIM_NUM = {dim_num}'
)
# Initialize the remaining rows of V.
for i in range(2, dim_num + 1):
# The bits of the integer POLY(I) gives the form of polynomial I.
# Find the degree of polynomial I from binary encoding.
j = self.poly[i - 1]
m = 0
j //= 2
while j > 0:
j //= 2
m += 1
# Expand this bit pattern to separate components
# of the logical array INCLUD.
j = self.poly[i - 1]
includ = np.zeros(m)
for k in range(m, 0, -1):
j2 = j // 2
includ[k - 1] = j != 2 * j2
j = j2
# Calculate the remaining elements of row I as explained
# in Bratley and Fox, section 2.
for j in range(m + 1, self.maxcol + 1):
newv = self.v[i - 1, j - m - 1]
p2 = 1
for k in range(1, m + 1):
p2 *= 2
if includ[k - 1]:
newv = np.bitwise_xor(
int(newv), int(p2 * self.v[i - 1, j - k - 1])
)
self.v[i - 1, j - 1] = newv
# Multiply columns of V by appropriate power of 2.
p2 = 1
for j in range(self.maxcol - 1, 0, -1):
p2 *= 2
self.v[0:dim_num, j - 1] = self.v[0:dim_num, j - 1] * p2
# RECIPD is 1/(common denominator of the elements in V).
self.recipd = 1.0 / (2 * p2)
self.lastq = np.zeros(dim_num)
def generate(self, dimensions, n_samples, random_state=None):
"""Creates samples from Sobol' set.
Parameters
----------
dimensions : list, shape (n_dims,)
List of search space dimensions.
Each search dimension can be defined either as
- a `(lower_bound, upper_bound)` tuple (for `Real` or `Integer`
dimensions),
- a `(lower_bound, upper_bound, "prior")` tuple (for `Real`
dimensions),
- as a list of categories (for `Categorical` dimensions), or
- an instance of a `Dimension` object (`Real`, `Integer` or
`Categorical`).
n_samples : int
The order of the Sobol' sequence. Defines the number of samples.
random_state : int, RandomState instance, or None (default)
Set random state to something other than None for reproducible
results.
Returns
-------
sample : array_like (n_samples, dim)
Sobol' set.
"""
total_n_samples = self.num_generated + n_samples
if not (total_n_samples & (total_n_samples - 1) == 0):
warnings.warn(
"The balance properties of Sobol' points require "
"n to be a power of 2. {0} points have been "
"previously generated, then: n={0}+{1}={2}. ".format(
self.num_generated, n_samples, total_n_samples
)
)
if self.skip != 0 and total_n_samples > self.skip:
raise ValueError(
f"{self.skip} points have been skipped: "
f"generating "
f"{n_samples} more points would cause the "
f"sequence to repeat."
)
rng = check_random_state(random_state)
space = Space(dimensions)
n_dim = space.n_dims
transformer = space.get_transformer()
space.set_transformer("normalize")
r = np.full((n_samples, n_dim), np.nan)
seed = self.skip
for j in range(n_samples):
r[j, 0:n_dim], seed = self._sobol(n_dim, seed)
if self.randomize:
r = _random_shift(r, rng)
r = space.inverse_transform(r)
space.set_transformer(transformer)
self.num_generated += n_samples
return r
def _sobol(self, dim_num, seed):
"""Generates a new quasirandom Sobol' vector with each call.
Parameters
----------
dim_num : int
Number of spatial dimensions.
`dim_num` must satisfy 1 <= DIM_NUM <= 40.
seed : int
the `seed` for the sequence.
This is essentially the index in the sequence of the quasirandom
value to be generated. On output, `seed` has been set to the
appropriate next value, usually simply `seed`+1.
If `seed` is less than 0 on input, it is treated as though it were 0.
An input value of 0 requests the first (0-th) element of
the sequence.
Returns
-------
vector, seed : np.array (n_dim,), int
The next quasirandom vector and the seed of its next vector.
"""
# Things to do only if the dimension changed.
if dim_num != self.dim_num_save:
self.init(dim_num)
seed = int(np.floor(seed))
if seed < 0:
seed = 0
pos_lo0 = 1
if seed == 0:
self.lastq = np.zeros(dim_num)
elif seed == self.seed_save + 1:
# Find the position of the right-hand zero in SEED.
pos_lo0 = _bit_lo0(seed)
elif seed <= self.seed_save:
self.seed_save = 0
self.lastq = np.zeros(dim_num)
for seed_temp in range(int(self.seed_save), int(seed)):
pos_lo0 = _bit_lo0(seed_temp)
for i in range(1, dim_num + 1):
self.lastq[i - 1] = np.bitwise_xor(
int(self.lastq[i - 1]), int(self.v[i - 1, pos_lo0 - 1])
)
pos_lo0 = _bit_lo0(seed)
elif self.seed_save + 1 < seed:
for seed_temp in range(int(self.seed_save + 1), int(seed)):
pos_lo0 = _bit_lo0(seed_temp)
for i in range(1, dim_num + 1):
self.lastq[i - 1] = np.bitwise_xor(
int(self.lastq[i - 1]), int(self.v[i - 1, pos_lo0 - 1])
)
pos_lo0 = _bit_lo0(seed)
# Check that the user is not calling too many times!
if self.maxcol < pos_lo0:
raise ValueError(
f'I4_SOBOL - Fatal error!\n'
f' Too many calls!\n'
f' MAXCOL = {self.maxcol}\n'
f' L = {pos_lo0}\n'
)
# Calculate the new components of QUASI.
quasi = np.zeros(dim_num)
for i in range(1, dim_num + 1):
quasi[i - 1] = self.lastq[i - 1] * self.recipd
self.lastq[i - 1] = np.bitwise_xor(
int(self.lastq[i - 1]), int(self.v[i - 1, pos_lo0 - 1])
)
self.seed_save = seed
seed += 1
return [quasi, seed]
def _bit_hi1(n):
"""Returns the position of the high 1 bit base 2 in an integer.
Parameters
----------
n : int
Input, should be positive.
"""
bin_repr = np.binary_repr(n)
most_left_one = bin_repr.find('1')
if most_left_one == -1:
return 0
else:
return len(bin_repr) - most_left_one
def _bit_lo0(n):
"""Returns the position of the low 0 bit base 2 in an integer.
Parameters
----------
n : int
Input, should be positive.
"""
bin_repr = np.binary_repr(n)
most_right_zero = bin_repr[::-1].find('0')
if most_right_zero == -1:
most_right_zero = len(bin_repr)
return most_right_zero + 1
def _random_shift(dm, random_state=None):
"""Random shifting of a vector.
Randomization of the quasi-MC samples can be achieved in the easiest manner
by random shift (or the Cranley-Patterson rotation).
References
-----------
.. [1] C. Lemieux, "Monte Carlo and Quasi-Monte Carlo Sampling," Springer
Series in Statistics 692, Springer Science+Business Media, New York,
2009
Parameters
----------
dm : array, shape(n, d)
Input matrix.
random_state : int, RandomState instance, or None (default)
Set random state to something other than None for reproducible
results.
Returns
-------
dm : array, shape(n, d)
Randomized Sobol' design matrix.
"""
rng = check_random_state(random_state)
# Generate random shift matrix from uniform distribution
shift = np.repeat(rng.rand(1, dm.shape[1]), dm.shape[0], axis=0)
# Return the shifted Sobol' design
return (dm + shift) % 1