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
|
# mode: run
# tag: numpy
cimport cython
import numpy as np
# I'm making these arrays have slightly irregular strides deliberately
int_arr_1d = np.arange(20, dtype=int)[::4]
int_arr_2d = np.arange(500, dtype=int).reshape((50, -1))[5:8, 6:8]
double_arr_1d = int_arr_1d.astype(np.double)
double_arr_2d = int_arr_2d.astype(np.double)
# Numpy has a cutoff at about 500 where it releases the GIL, so test some large arrays
large_int_arr_1d = np.arange(1500, dtype=int)
large_int_arr_2d = np.arange(1500*600, dtype=int).reshape((1500, -1))
large_double_arr_1d = large_int_arr_1d.astype(np.double)
large_double_arr_2d = large_int_arr_2d.astype(np.double)
# it's fairly hard to test that nogil results in the GIL actually
# being released unfortunately
@cython.ufunc
cdef double triple_it(long x) nogil:
"""triple_it doc"""
return x*3.
def test_triple_it():
"""
Ufunc also generates a signature so just look at the end
>>> triple_it.__doc__.endswith('triple_it doc')
True
>>> triple_it(int_arr_1d)
array([ 0., 12., 24., 36., 48.])
>>> triple_it(int_arr_2d)
array([[168., 171.],
[198., 201.],
[228., 231.]])
Treat the large arrays just as a "don't crash" test
>>> _ = triple_it(large_int_arr_1d)
>>> _ = triple_it(large_int_arr_2d)
"""
@cython.ufunc
cdef double to_the_power(double x, double y):
return x**y
def test_to_the_power():
"""
>>> np.allclose(to_the_power(double_arr_1d, 1.), double_arr_1d)
True
>>> np.allclose(to_the_power(1., double_arr_2d), np.ones_like(double_arr_2d))
True
>>> _ = to_the_power(large_double_arr_1d, -large_double_arr_1d)
>>> _ = to_the_power(large_double_arr_2d, -large_double_arr_2d)
"""
@cython.ufunc
cdef object py_return_value(double x):
if x >= 0:
return x
# default returns None
def test_py_return_value():
"""
>>> py_return_value(5.)
5.0
>>> py_return_value(double_arr_1d).dtype
dtype('O')
>>> py_return_value(-1.) # returns None
>>> _ = py_return_value(large_double_arr_1d)
"""
@cython.ufunc
cdef double py_arg(object x):
return float(x)
def test_py_arg():
"""
>>> py_arg(np.array([1, "2.0", 3.0], dtype=object))
array([1., 2., 3.])
>>> _ = py_arg(np.array([1]*1200, dtype=object))
"""
@cython.ufunc
cdef (double, long) multiple_return_values(long x):
return x*1.5, x*2
@cython.ufunc
cdef (double, long) multiple_return_values2(long x):
inefficient_tuple_intermediate = (x*1.5, x*2)
return inefficient_tuple_intermediate
def test_multiple_return_values():
"""
>>> multiple_return_values(int_arr_1d)
(array([ 0., 6., 12., 18., 24.]), array([ 0, 8, 16, 24, 32]))
>>> multiple_return_values2(int_arr_1d)
(array([ 0., 6., 12., 18., 24.]), array([ 0, 8, 16, 24, 32]))
"""
@cython.ufunc
cdef cython.numeric plus_one(cython.numeric x):
return x+1
def test_plus_one():
"""
This generates all the fused combinations
>>> plus_one(int_arr_1d) # doctest: +ELLIPSIS
array([ 1, 5, 9, 13, 17]...)
>>> plus_one(double_arr_2d)
array([[57., 58.],
[67., 68.],
[77., 78.]])
>>> plus_one(1.j)
(1+1j)
"""
###### Test flow-control ######
# An initial implementation of ufunc did some odd restructuring of the code to
# bring the functions completely inline at the Cython level. These tests were to
# test that "return" statements work. They're less needed now, but don't do any
# harm
@cython.ufunc
cdef double return_stops_execution(double x):
return x
print "This should not happen"
@cython.ufunc
cdef double return_in_if(double x):
if x<0:
return -x
return x
@cython.ufunc
cdef double nested_loops(double x):
cdef double counter=0
while x>counter:
counter+=10.
for i in range(100):
if i>x:
return i
return x-counter
def test_flow_control():
"""
>>> np.allclose(return_stops_execution(double_arr_1d), double_arr_1d)
True
>>> return_in_if(-1.)
1.0
>>> return_in_if(2.0)
2.0
>>> nested_loops(5.5)
6.0
>>> nested_loops(105.)
-5.0
"""
@cython.ufunc
cdef double nested_function(double x):
def f(x):
return x*2
return f(x)
def test_nested_function():
"""
>>> np.allclose(nested_function(double_arr_1d), 2*double_arr_1d)
True
>>> nested_function(-1.)
-2.0
"""
@cython.ufunc
cdef double can_throw(double x):
if x<0:
raise RuntimeError
return x
def test_can_throw():
"""
>>> arr = double_arr_1d.copy()
>>> arr[1] = -1.
>>> can_throw(arr)
Traceback (most recent call last):
...
RuntimeError
>>> large_arr = large_double_arr_1d.copy()
>>> large_arr[-4] = -2.
>>> can_throw(large_arr)
Traceback (most recent call last):
...
RuntimeError
>>> large_arr2d = large_double_arr_2d.copy()
>>> large_arr2d[100, 200] = -1.
>>> can_throw(large_arr2d)
Traceback (most recent call last):
...
RuntimeError
"""
|
