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
|
#cython: embedsignature=True
#cython: language_level=3
import numpy
cimport numpy as np
import cython
from libc.math cimport log
from numpy cimport NPY_FLOAT64 as NPY_float64
@cython.boundscheck(False)
@cython.wraparound(False)
def split_eq_freq(np.ndarray[np.float64_t, ndim=2] dist not None, int n):
cdef int llen = dist.shape[1]
if n >= llen: #n is greater than distributions
return [(v1+v2)/2 for v1,v2 in zip(dist[0], dist[0][1:])]
cdef np.float64_t N = dist[1].sum()
cdef int toGo = n
cdef np.float64_t inthis = 0
cdef np.float64_t prevel = -1
cdef np.float64_t inone = N/toGo
points = []
cdef Py_ssize_t i
cdef np.float64_t v
cdef np.float64_t k
cdef np.float64_t vn
for i in range(llen):
v = dist[0,i]
k = dist[1,i]
if toGo <= 1:
break
inthis += k
if inthis < inone or i == 0:
prevel = v
else: #current count exceeded
if i < llen - 1 and inthis - inone < k / 2:
#exceeded for less than half the current count:
#split after current
vn = dist[0,i+1]
points.append((vn + v)/2)
N -= inthis
inthis = 0
prevel = vn
else:
#split before the current value
points.append((prevel + v)/2)
N -= inthis - k
inthis = k
prevel = v
toGo -= 1
if toGo:
inone = N/toGo
return points
@cython.wraparound(False)
@cython.boundscheck(False)
def entropy_normalized1(np.ndarray[np.float64_t, ndim=1] D):
"""
Compute entropy of distribution in `D` (must be normalized).
"""
cdef np.float64_t R = 0.
cdef Py_ssize_t j
cdef np.float64_t t
cdef np.float64_t log2 = 1./log(2.)
for j in range(D.shape[0]):
t = D[j]
if t > 0.:
if t > 1.0: t = 1.0
R -= t*log(t)*log2
return R
@cython.wraparound(False)
@cython.boundscheck(False)
def entropy_normalized2(np.ndarray[np.float64_t, ndim=2] D):
"""
Compute entropy of distributions in `D`.
Rows in `D` must be a distribution (i.e. sum to 1.0 over `axis`).
"""
cdef np.ndarray[np.float64_t, ndim=1] R = numpy.zeros(D.shape[0])
cdef Py_ssize_t i,j
cdef np.float64_t t
cdef np.float64_t log2 = 1./log(2.)
for i in range(D.shape[0]):
for j in range(D.shape[1]):
t = D[i,j]
if t > 0.:
if t > 1.0: t = 1.0
R[i] -= t*log(t)*log2
return R
|
