GitLab Repo

amachine.am_random 

 1from scipy.special import softmax
 2from scipy.stats import beta
 3import numpy as np
 4import random
 5
 6_rng = np.random.default_rng()
 7
 8def resolve_rng(np_rng):
 9	global _rng
10	return _rng if np_rng is None else np_rng
11
12def exp_uniform_blend(n: int, alpha: float, np_rng=None ) -> list[float]:
13	global _rng
14	rng = resolve_rng(np_rng)
15	exp_probs = np.exp(-np.arange(n))
16	exp_probs /= exp_probs.sum()
17	uniform = np.ones(n) / n
18	probs = ( (1 - alpha) * exp_probs + alpha * uniform)
19	return rng.permutation( probs ).tolist()
20
21def beta_dist( n, a, b, np_rng=None ):
22	global _rng
23	rng = resolve_rng(np_rng)
24	points = np.linspace(0, 1, n + 2 )[ 1:-1] 
25	values = beta.pdf(points, a, b)
26	probs  = values / values.sum()
27	return rng.permutation( probs ).to_list()
28
29def dirichlet_dist(n, alpha, np_rng=None ):
30	global _rng
31	rng = resolve_rng(np_rng)
32	return rng.dirichlet(np.ones(n) * alpha)
33
34def uniform_dist( n ):
35	return np.ones(n) / n
36
37def srand_global(seed: int):
38	global _rng
39	# _rng = np.random.default_rng(11)
40	np.random.seed(seed)
41	random.seed(seed)
def resolve_rng(np_rng): 
 9def resolve_rng(np_rng):
10	global _rng
11	return _rng if np_rng is None else np_rng
def exp_uniform_blend(n: int, alpha: float, np_rng=None) -> list[float]: 
13def exp_uniform_blend(n: int, alpha: float, np_rng=None ) -> list[float]:
14	global _rng
15	rng = resolve_rng(np_rng)
16	exp_probs = np.exp(-np.arange(n))
17	exp_probs /= exp_probs.sum()
18	uniform = np.ones(n) / n
19	probs = ( (1 - alpha) * exp_probs + alpha * uniform)
20	return rng.permutation( probs ).tolist()
def beta_dist(n, a, b, np_rng=None): 
22def beta_dist( n, a, b, np_rng=None ):
23	global _rng
24	rng = resolve_rng(np_rng)
25	points = np.linspace(0, 1, n + 2 )[ 1:-1] 
26	values = beta.pdf(points, a, b)
27	probs  = values / values.sum()
28	return rng.permutation( probs ).to_list()
def dirichlet_dist(n, alpha, np_rng=None): 
30def dirichlet_dist(n, alpha, np_rng=None ):
31	global _rng
32	rng = resolve_rng(np_rng)
33	return rng.dirichlet(np.ones(n) * alpha)
def uniform_dist(n): 
35def uniform_dist( n ):
36	return np.ones(n) / n
def srand_global(seed: int): 
38def srand_global(seed: int):
39	global _rng
40	# _rng = np.random.default_rng(11)
41	np.random.seed(seed)
42	random.seed(seed)