Arrays versus lists

Why another kind of list?

A Python list can hold anything, which is exactly why it is slow for numbers: every element is a separate object, and every arithmetic operation is a separate interpreted step.

A NumPy array holds one type in one contiguous block of memory, and its operations run as compiled loops:

import numpy as np

a = np.array([1, 2, 3, 4])
a.dtype     # dtype('int64')  — one type for the whole array
a.shape     # (4,)            — its dimensions
a.size      # 4

The first thing that surprises people coming from lists:

[1, 2, 3] * 2          # [1, 2, 3, 1, 2, 3]  — repeats the list
np.array([1, 2, 3]) * 2   # array([2, 4, 6]) — multiplies every element

That difference is the whole point. On an array, an operation means "do this to every element", and you almost never write a loop.

Your turn: build an array from readings, and record its dtype name, its shape, and the result of doubling every value.

Creating arrays without a list

NumPy has its own constructors, and they are the usual way to start:

np.zeros(3)            # array([0., 0., 0.])
np.ones(3)             # array([1., 1., 1.])
np.arange(0, 10, 2)    # array([0, 2, 4, 6, 8]) — like range()
np.linspace(0, 1, 5)   # array([0., 0.25, 0.5, 0.75, 1.]) — n evenly spaced points

arange takes a step and stops before the end, exactly like range. linspace takes a count and includes both ends — that difference catches people out.

Your turn: build evens (0 to 8 in twos), ramp (five evenly spaced points from 0 to 100 inclusive), and grid (a 2 by 3 array of zeros).