-
Related: Lists in Python
-
https://colab.research.google.com/drive/1VdLyfnop7Tt4wx4BmS6WHSSQg3n5wXrX?pli=1&authuser=1
-
If you apply a function to a numpy array, it very often operates element wise.
-
A numpy array exists specifically to do math!
- Also they are more memory efficient, faster, and more convenient!
import numpy as np
L = [1,2,3]
A = np.array([1,2,3])
# the for loops does the same thing for both
for i in L:
print(i)
for i in A:
print(i)
L.append(4) # L = [1,2,3,4]
A.append(4) # ERROR! (arrays can't be changed)
"""
L = [1,2,3,4]
A = [1,2,3]
"""
L + [5] # L = [1,2,3,4,5]
A + [5] # A = [6,7,8] (BROADCASTING)
A + np.array([5]) # A = [6,7,8]
L * 2 # L = [1,2,3,4,1,2,3,4], same as L + L
A * 2 # A = [2,4,6], same as A + A
# examples of functions
np.sqrt(A)
np.log(A)Broadcasting
- A numpy array broadcasts!
- NumPy performs element-wise operations on arrays of different shapes
- The 2nd item gets “broadcasted” to the original item
"""Case 1: Scalar and array"""
A = np.array([1,2,3])
# Broadcasting scalar 1 to each element
A + 1 # [2,3,4]
"""Case 2: Arrays with different shapes"""
# shape (2,3)
A = np.array(
[
[1,2,3],
[4,5,6]
]
)
# shape (3,)
B = np.array(
[10,20,30]
)
# Broadcasting B to match the shape of A
result = A + B
# [[11 22 33]
# [14 25 36]]
"""Case 3: Mismatched Shapes that Work with Broadcasting"""- Related: Shapes & dimensions