回答編集履歴
2
追記
test
CHANGED
|
@@ -110,8 +110,16 @@
|
|
|
110
110
|
|
|
111
111
|
|
|
112
112
|
|
|
113
|
-
https://
|
|
113
|
+
https://stackoverflow.com/questions/4389517/in-place-type-conversion-of-a-numpy-array
|
|
114
114
|
|
|
115
115
|
|
|
116
116
|
|
|
117
|
-
|
|
117
|
+
リファレンスにはけっこう怖い記述があるので、必ずしもいい方法ではないかもしれません。
|
|
118
|
+
|
|
119
|
+
|
|
120
|
+
|
|
121
|
+
> For a.view(some_dtype), if some_dtype has a different number of bytes per entry than the previous dtype (for example, converting a regular array to a structured array), then the behavior of the view cannot be predicted just from the superficial appearance of a (shown by print(a)). It also depends on exactly how a is stored in memory. Therefore if a is C-ordered versus fortran-ordered, versus defined as a slice or transpose, etc., the view may give different results.
|
|
122
|
+
|
|
123
|
+
|
|
124
|
+
|
|
125
|
+
> https://docs.scipy.org/doc/numpy/reference/generated/numpy.ndarray.view.html
|
1
追記
test
CHANGED
|
@@ -15,3 +15,103 @@
|
|
|
15
15
|
|
|
16
16
|
|
|
17
17
|
質問文のやり方はnumpyを使うのであれば最良に近いと思います。同じメモリ領域を読み替えたりするのは難しいので、そこは諦めるのが前提になります。
|
|
18
|
+
|
|
19
|
+
|
|
20
|
+
|
|
21
|
+
|
|
22
|
+
|
|
23
|
+
### 追記
|
|
24
|
+
|
|
25
|
+
すみません、viewで型指定すればできそうです。ただしread-onlyになりそう。
|
|
26
|
+
|
|
27
|
+
|
|
28
|
+
|
|
29
|
+
```python
|
|
30
|
+
|
|
31
|
+
>>> import numpy as np
|
|
32
|
+
|
|
33
|
+
>>> a = np.random.randint(100, 1000, (4, 3), dtype=np.uint32)
|
|
34
|
+
|
|
35
|
+
>>> b = np.frombuffer(a.tobytes(), dtype=np.uint8).reshape([4, 3, 4])
|
|
36
|
+
|
|
37
|
+
>>> b
|
|
38
|
+
|
|
39
|
+
array([[[ 67, 1, 0, 0],
|
|
40
|
+
|
|
41
|
+
[148, 1, 0, 0],
|
|
42
|
+
|
|
43
|
+
[242, 1, 0, 0]],
|
|
44
|
+
|
|
45
|
+
|
|
46
|
+
|
|
47
|
+
[[ 55, 1, 0, 0],
|
|
48
|
+
|
|
49
|
+
[138, 0, 0, 0],
|
|
50
|
+
|
|
51
|
+
[ 80, 3, 0, 0]],
|
|
52
|
+
|
|
53
|
+
|
|
54
|
+
|
|
55
|
+
[[ 66, 1, 0, 0],
|
|
56
|
+
|
|
57
|
+
[ 71, 3, 0, 0],
|
|
58
|
+
|
|
59
|
+
[100, 0, 0, 0]],
|
|
60
|
+
|
|
61
|
+
|
|
62
|
+
|
|
63
|
+
[[ 3, 3, 0, 0],
|
|
64
|
+
|
|
65
|
+
[ 84, 3, 0, 0],
|
|
66
|
+
|
|
67
|
+
[134, 2, 0, 0]]], dtype=uint8)
|
|
68
|
+
|
|
69
|
+
>>> b2 = a.view(np.uint8).reshape(4, 3, 4)
|
|
70
|
+
|
|
71
|
+
>>> b2
|
|
72
|
+
|
|
73
|
+
array([[[ 67, 1, 0, 0],
|
|
74
|
+
|
|
75
|
+
[148, 1, 0, 0],
|
|
76
|
+
|
|
77
|
+
[242, 1, 0, 0]],
|
|
78
|
+
|
|
79
|
+
|
|
80
|
+
|
|
81
|
+
[[ 55, 1, 0, 0],
|
|
82
|
+
|
|
83
|
+
[138, 0, 0, 0],
|
|
84
|
+
|
|
85
|
+
[ 80, 3, 0, 0]],
|
|
86
|
+
|
|
87
|
+
|
|
88
|
+
|
|
89
|
+
[[ 66, 1, 0, 0],
|
|
90
|
+
|
|
91
|
+
[ 71, 3, 0, 0],
|
|
92
|
+
|
|
93
|
+
[100, 0, 0, 0]],
|
|
94
|
+
|
|
95
|
+
|
|
96
|
+
|
|
97
|
+
[[ 3, 3, 0, 0],
|
|
98
|
+
|
|
99
|
+
[ 84, 3, 0, 0],
|
|
100
|
+
|
|
101
|
+
[134, 2, 0, 0]]], dtype=uint8)
|
|
102
|
+
|
|
103
|
+
>>> (b == b2).all()
|
|
104
|
+
|
|
105
|
+
True
|
|
106
|
+
|
|
107
|
+
|
|
108
|
+
|
|
109
|
+
```
|
|
110
|
+
|
|
111
|
+
|
|
112
|
+
|
|
113
|
+
https://docs.scipy.org/doc/numpy/reference/generated/numpy.ndarray.view.html
|
|
114
|
+
|
|
115
|
+
|
|
116
|
+
|
|
117
|
+
https://stackoverflow.com/questions/4389517/in-place-type-conversion-of-a-numpy-array
|