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]]])