Numpy Note
March 22, 2021
In [1]: import numpy as np
1 Creating Numpy Array
In [4]: A = np.array([[[ 0, 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]]])
In [22]: B = np.arange(48)
B
Out[22]: array([ 0, 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])
In [23]: C = np.zeros((4,4))
C
Out[23]: array([[0., 0., 0., 0.],
[0., 0., 0., 0.],
[0., 0., 0., 0.],
[0., 0., 0., 0.]])
In [25]: D = np.ones((5, 5))
D
Out[25]: array([[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1.]])
In [27]: E = np.full((6, 6), 10)
E
1
Out[27]: array([[10, 10, 10, 10, 10, 10],
[10, 10, 10, 10, 10, 10],
[10, 10, 10, 10, 10, 10],
[10, 10, 10, 10, 10, 10],
[10, 10, 10, 10, 10, 10],
[10, 10, 10, 10, 10, 10]])
In [29]: F = np.eye(10)
F
Out[29]: array([[1., 0., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 1., 0., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 1., 0., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 1., 0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 1., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 1., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 1., 0., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 1., 0., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 1., 0.],
[0., 0., 0., 0., 0., 0., 0., 0., 0., 1.]])
2 Array Shape
In [9]: A.shape
Out[9]: (4, 1, 8)
In [20]: B = B.reshape(1, 6, 8)
3 Array Axis
Imagine collapsing the chosen axis
In [12]: A.max()
Out[12]: 31
In [13]: A.max(axis = 0)
Out[13]: array([[24, 25, 26, 27, 28, 29, 30, 31]])
In [14]: A.max(axis = 1)
Out[14]: array([[ 0, 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]])
In [15]: A.max(axis = 2)
2
Out[15]: array([[ 7],
[15],
[23],
[31]])
4 Indexing and Slicing
4.0.1 Single Element
In [161]: a = np.arange(12).reshape(3, 4)
a
Out[161]: array([[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]])
In [162]: a[1, 2]
Out[162]: 6
In [163]: a[-1, -1]
Out[163]: 11
4.0.2 Subarray
Index Array Form: array[array, array]
Output: 1D array of values with indices (x, y)
Note: return a copy of the subarray
In [176]: a = np.arange(12).reshape(3, 4)
a
Out[176]: array([[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]])
In [177]: # (0,0), (0,3), (2,0), (2,3)
b = a[[0, 0, 2, 2], [0, 3, 0, 3]]
b
Out[177]: array([ 0, 3, 8, 11])
In [178]: # when shapes doesn't match attempt broadcast
# essentially b = a[[1, 2], [2, 2]]
b = a[[1,2], [2]]
b
Out[178]: array([ 6, 10])
3
Slices Form: array[slice, slice]
Output: Array slice with same dimension
Note: return the view of the subarray. use copy() if want a copy
In [192]: a = np.arange(12).reshape(3, 4)
a
Out[192]: array([[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]])
In [193]: b = a[:2, 1:3]
b
Out[193]: array([[1, 2],
[5, 6]])
Index Array with Slices Form: array[slice/array, slice/array]
Output: Array with same dimension Note: return a copy
In [197]: a = np.arange(12).reshape(3, 4)
a
Out[197]: array([[ 0, 1, 2, 3],
[ 4, 5, 6, 7],
[ 8, 9, 10, 11]])
In [198]: b = a[[0, 2], 1:3]
b
Out[198]: array([[ 1, 2],
[ 9, 10]])
5 Broadcasting and Vectorization
Broadcasting Rule: for each dimension, either equal or one is 1
In [18]: A + B
Out[18]: array([[[ 0, 2, 4, 6, 8, 10, 12, 14],
[ 8, 10, 12, 14, 16, 18, 20, 22],
[16, 18, 20, 22, 24, 26, 28, 30],
[24, 26, 28, 30, 32, 34, 36, 38],
[32, 34, 36, 38, 40, 42, 44, 46],
[40, 42, 44, 46, 48, 50, 52, 54]],
[[ 8, 10, 12, 14, 16, 18, 20, 22],
[16, 18, 20, 22, 24, 26, 28, 30],
[24, 26, 28, 30, 32, 34, 36, 38],
4
[32, 34, 36, 38, 40, 42, 44, 46],
[40, 42, 44, 46, 48, 50, 52, 54],
[48, 50, 52, 54, 56, 58, 60, 62]],
[[16, 18, 20, 22, 24, 26, 28, 30],
[24, 26, 28, 30, 32, 34, 36, 38],
[32, 34, 36, 38, 40, 42, 44, 46],
[40, 42, 44, 46, 48, 50, 52, 54],
[48, 50, 52, 54, 56, 58, 60, 62],
[56, 58, 60, 62, 64, 66, 68, 70]],
[[24, 26, 28, 30, 32, 34, 36, 38],
[32, 34, 36, 38, 40, 42, 44, 46],
[40, 42, 44, 46, 48, 50, 52, 54],
[48, 50, 52, 54, 56, 58, 60, 62],
[56, 58, 60, 62, 64, 66, 68, 70],
[64, 66, 68, 70, 72, 74, 76, 78]]])
