annotate pymedcouple @ 31:1a2bb52722c2

pymedcouple: replace slow Python partsort with faster built-in full sort
author Jordi Gutiérrez Hermoso <jordigh@octave.org>
date Sun, 18 Jan 2015 20:45:35 -0500
parents cd3094a08e03
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
5
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
1 #!/usr/bin/python
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
2
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
3 import random
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
4
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
5 from itertools import tee, izip
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
6
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
7 def wmedian(A, W):
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
8 """This computes the weighted median of array A with corresponding
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
9 weights W.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
10 """
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
11
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
12 AW = zip(A, W)
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
13 n = len(AW)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
14
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
15 wtot = sum(W)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
16
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
17 beg = 0
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
18 end = n - 1
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
19
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
20 while True:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
21 mid = (beg + end)//2
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
22
31
1a2bb52722c2 pymedcouple: replace slow Python partsort with faster built-in full sort
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 24
diff changeset
23 AW = sorted(AW, key = lambda x: x[0]) # A partial sort would suffice here
1a2bb52722c2 pymedcouple: replace slow Python partsort with faster built-in full sort
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 24
diff changeset
24
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
25 trial = AW[mid][0]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
26
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
27 wleft = wright = 0
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
28 for (a, w) in AW:
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
29 if a < trial:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
30 wleft += w
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
31 else:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
32 # This also includes a == trial, i.e. the "middle"
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
33 # weight.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
34 wright += w
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
35
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
36 if 2*wleft > wtot:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
37 end = mid
21
d63e763f8ac1 wmedian: don't infinite loop if wleft == wright
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 20
diff changeset
38 elif 2*wright < wtot:
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
39 beg = mid
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
40 else:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
41 return trial
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
42
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
43
10
3d145bcc8694 Remove more debug code
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 9
diff changeset
44 def medcouple_1d(X, eps1 = 2**-52, eps2 = 2**-1022):
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
45 """Calculates the medcouple robust measure of skewness.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
46
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
47 Parameters
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
48 ----------
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
49 y : array-like, 1-d
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
50
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
51 Returns
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
52 -------
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
53 mc : float
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
54 The medcouple statistic
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
55
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
56 .. [1] G. Brys, M. Hubert, and A. Struyf "A Robust Measure of
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
57 Skewness." Journal of Computational and Graphical Statistics, Vol.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
58 13, No. 4 (Dec., 2004), pp. 996- 1017
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
59
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
60 .. [2] D. B. Johnson and T. Mizoguchi "Selecting the Kth Element
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
61 in $X + Y$ and $X_1 + X_2 + \cdots X_m$". SIAM Journal of
9
e6bcaf38edaa Fix citation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 8
diff changeset
62 Computing, Vol. 7, No. 2 (May 1978), pp. 147-53.
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
63
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
64 """
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
65 # FIXME: Figure out what to do about NaNs.
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
66
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
67 n = len(X)
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
68 n2 = (n - 1)//2
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
69
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
70 if n < 3:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
71 return 0
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
72
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
73 Z = sorted(X, reverse=True)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
74
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
75 if n % 2 == 1:
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
76 Zmed = Z[n2]
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
77 else:
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
78 Zmed = (Z[n2] + Z[n2 + 1])/2
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
79
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
80 #Check if the median is at the edges up to relative epsilon
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
81 if abs(Z[0] - Zmed) < eps1*(eps1 + abs(Zmed)):
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
82 return -1.0
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
83 if abs(Z[-1] - Zmed) < eps1*(eps1 + abs(Zmed)):
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
84 return 1.0
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
85
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
86 # Centre Z wrt median, so that median(Z) = 0.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
87 Z = [z - Zmed for z in Z]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
88
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
89 # Scale inside [-0.5, 0.5], for greater numerical stability.
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
90 Zden = 2*max(Z[0], -Z[-1])
5
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
91 Z = [z/Zden for z in Z]
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
92 Zmed /= Zden
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
93
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
94 Zeps = eps1*(eps1 + abs(Zmed))
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
95
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
96 # These overlap on the entries that are tied with the median
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
97 Zplus = [z for z in Z if z >= -Zeps]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
98 Zminus = [z for z in Z if Zeps >= z]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
99
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
100 n_plus = len(Zplus)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
101 n_minus = len(Zminus)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
102
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
103
10
3d145bcc8694 Remove more debug code
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 9
diff changeset
104 def h_kern(i, j):
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
105 """Kernel function h for the medcouple, closing over the values of
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
106 Zplus and Zminus just defined above.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
107
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
108 In case a and be are within epsilon of the median, the kernel
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
109 is the signum of their position.
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
110
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
111 """
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
112 a = Zplus[i]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
113 b = Zminus[j]
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
114
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
115 if abs(a - b) <= 2*eps2:
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
116 h = signum(n_plus - 1 - i - j)
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
117 else:
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
118 h = (a + b)/(a - b)
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
119
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
120 return h
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
121
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
122 # Init left and right borders
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
123 L = [0]*n_plus
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
124 R = [n_minus - 1]*n_plus
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
125
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
126 Ltot = 0
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
127 Rtot = n_minus*n_plus
22
f5b4a2ab6204 medcouple_1d: modify the definition of medc_idx to match R's definition
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 21
diff changeset
128 medc_idx = Rtot//2
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
129
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
130 # kth pair algorithm (Johnson & Mizoguchi)
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
131 while Rtot - Ltot > n_plus:
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
132
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
133 # First, compute the median inside the given bounds
6
e3b1dcc51e6a Fix tee usage
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 5
diff changeset
134 # (Be stingy, reuse same generator)
e3b1dcc51e6a Fix tee usage
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 5
diff changeset
135 [I1, I2] = tee(i for i in xrange(0, n_plus) if L[i] <= R[i])
e3b1dcc51e6a Fix tee usage
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 5
diff changeset
136
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
137 A = [h_kern(i, (L[i] + R[i])//2) for i in I1]
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
138 W = [R[i] - L[i] + 1 for i in I2]
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
139 Am = wmedian(A, W)
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
140
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
141 Am_eps = eps1*(eps1 + abs(Am))
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
142
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
143 # Compute new left and right boundaries, based on the weighted
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
144 # median
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
145 P = []
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
146 Q = []
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
147
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
148 j = 0
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
149 for i in xrange(n_plus - 1, -1, -1):
20
8e02346789b2 Fix off by one error in P computation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 13
diff changeset
150 while j < n_minus and h_kern(i, j) - Am > Am_eps:
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
151 j += 1
12
c81f6d263897 Spaces around plus and minus
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 11
diff changeset
152 P.append(j - 1)
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
153
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
154 P.reverse()
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
155
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
156 j = n_minus - 1
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
157 for i in xrange(0, n_plus):
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
158 while j >= 0 and h_kern(i, j) - Am < -Am_eps:
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
159 j -= 1
12
c81f6d263897 Spaces around plus and minus
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 11
diff changeset
160 Q.append(j + 1)
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
161
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
162 # Check on which side of those bounds the desired median of
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
163 # the whole matrix may be.
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
164 sumP = sum(P) + len(P)
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
165 sumQ = sum(Q)
11
4c7d7ff28a0e Rename mid_idx to medc_idx
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 10
diff changeset
166
4c7d7ff28a0e Rename mid_idx to medc_idx
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 10
diff changeset
167 if medc_idx <= sumP - 1:
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
168 R = P
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
169 Rtot = sumP
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
170 else:
11
4c7d7ff28a0e Rename mid_idx to medc_idx
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 10
diff changeset
171 if medc_idx > sumQ - 1:
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
172 L = Q
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
173 Ltot = sumQ
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
174 else:
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
175 return Am
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
176
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
177 # Didn't find the median, but now we have a very small search
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
178 # space to find it in, just between the left and right boundaries.
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
179 # This space is of size Rtot - Ltot which is <= n_plus
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
180 A = []
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
181 for (i, (l, r)) in enumerate(izip(L, R)):
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
182 for j in xrange(l, r + 1):
10
3d145bcc8694 Remove more debug code
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 9
diff changeset
183 A.append(h_kern(i, j))
31
1a2bb52722c2 pymedcouple: replace slow Python partsort with faster built-in full sort
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 24
diff changeset
184
1a2bb52722c2 pymedcouple: replace slow Python partsort with faster built-in full sort
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 24
diff changeset
185 A.sort() # A partial sort would suffice here
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
186 A.reverse()
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
187
23
29a178b23219 Whitespace style fixes
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 22
diff changeset
188 Am = A[medc_idx - Ltot]
11
4c7d7ff28a0e Rename mid_idx to medc_idx
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 10
diff changeset
189
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
190 return Am
13
077261db7a58 Whitespace cleanup
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 12
diff changeset
191
4
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
192
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
193 def signum(x):
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
194 if x > 0:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
195 return 1
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
196 if x < 0:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
197 return -1
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
198 else:
74d0d08dbc95 Add Python implementation
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents:
diff changeset
199 return 0
5
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
200
8
b3e878bb793d Remove debug statements
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 7
diff changeset
201
5
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
202 def main():
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
203 import sys
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
204 fname = sys.argv[1]
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
205 with open(fname) as f:
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
206 data = [float(x) for x in f.readlines() if x.strip() != ""]
8
b3e878bb793d Remove debug statements
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 7
diff changeset
207
7
6c7c7dc9d8ef hairball
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 6
diff changeset
208 print "%.16g" % medcouple_1d(data)
5
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
209
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
210 if __name__ == "__main__":
cbe17f888c79 Make it executable
Jordi Gutiérrez Hermoso <jordigh@octave.org>
parents: 4
diff changeset
211 main()