Review

Core idea

A NumPy array is a shaped numerical container.

Before computing, inspect:

Arrays versus lists

List addition concatenates lists. Array addition adds element by element.

[1, 2] + [10, 20]

produces:

[1, 2, 10, 20]

np.array([1, 2]) + np.array([10, 20])

produces:

[11 22]

Creating arrays

Useful constructors:

Shape and axis

For a matrix with shape (2, 3):

• axis 0 indexes the two rows;
• axis 1 indexes the three columns;
• A.sum(axis=0) produces one value per column;
• A.sum(axis=1) produces one value per row.

Always attach meaning to the axes.

Indexing

: means all values along that axis.

Views and copies

Basic slices share data with the original array.

Use:

part = array_slice.copy()

when you want a separate array before mutation.

Reductions

Common reductions:

Use axis=... to summarize along a specific axis.

Array errors

When an array operation fails, inspect:

1. shape;
2. dtype;
3. the smallest example that still fails.