Free tools Windows power users keep installed
One-click scans. No signup required.
In Python, you can represent a 2D structure with a list of lists, or use a NumPy ndarray for multidimensional numerical work. A nested list is flexible and needs no extra package; a NumPy array adds explicit dimensions, a common element type, and concise array arithmetic.
Make a 2D structure with nested lists
Each inner list represents a row. For a rectangular grid, make every row the same length:
rows = [
[1, 2],
[3, 4],
[5, 6],
]
print(rows[0][1]) # 2Python uses zero-based indexing: rows[0] selects the first row, and the second index selects an item within that row. Built-in lists can also hold rows of different lengths, but such a value is not a regular rectangle. Check row lengths if your code relies on a grid with consistent columns. The Python tutorial’s list examples describe a matrix as a list of equal-length lists.
Convert the nested list to a NumPy array
Pass the nested sequence as one argument to np.array(). NumPy creates an array with dimensions inferred from the nesting and values:
Do these 3 things before closing this tab:
1Repair Windows errors before they cause bigger problems2Fix the driver behind crashes, sound loss and screen glitches3Clear out junk files and repair common Windows errors#1 Best Overall
import numpy as np
array = np.array(rows)
print(array)
print(array.shape) # (3, 2)
print(array.ndim) # 2
print(array.size) # 6
print(array.dtype) # inferred from the valuesThe properties answer different questions: shape gives the length of each axis, ndim gives the number of axes, size gives the total number of elements, and dtype gives the element type. If your code requires a particular numeric representation, specify it rather than relying on inference:
floats = np.array([[1, 2], [3, 4]], dtype=np.float64)NumPy’s array creation guide covers construction from nested sequences, data types, and shape-based constructors.
Create arrays by shape or reshape values
When you know the desired dimensions, constructors such as zeros and ones can be clearer than writing every element. You can also reshape a sequence when its element count matches the requested dimensions:
zeros = np.zeros((2, 3))
ones = np.ones((2, 3), dtype=int)
sequence = np.arange(6).reshape(2, 3)The shape tuple is (rows, columns) here. reshape(2, 3) requires six values, because the requested shape contains six positions.
Select an element, row, or column
For a nested list, use chained indexing. For a NumPy array, use comma-separated indexes for separate axes:
rows = [[10, 11, 12], [20, 21, 22]]
array = np.array(rows)
rows[0][1] # 11
array[0, 1] # 11
array[1] # second row
array[:, 0] # first column
array[0:2, 1:] # rows 0–1, columns 1 onwardThe colon selects all positions along an axis; a slice such as 1: selects from index 1 onward. In a built-in list, rows[0, 1] is not the usual row-and-column notation: a list expects one index, while NumPy arrays support a tuple of axis indexes. For more examples, see the NumPy beginner guide to indexing.
Use NumPy for elementwise arithmetic and broadcasting
Ordinary list operations are not elementwise numeric matrix operations. NumPy arithmetic applies operations to array elements:
array = np.array([[1, 2], [3, 4]])
array + 10
# array([[11, 12],
# [13, 14]])NumPy can also broadcast a compatible smaller shape across a larger one. Here the one-dimensional operand has two values, so it applies across the two columns of each row:
Outdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchPC Slower Than It Used to Be?
A free scan shows the junk files, broken settings and background clutter dragging Windows down - then fixes them in one click.Free scan · Windows 10 & 11array * np.array([10, 100])
# array([[ 10, 200],
# [ 30, 400]])Broadcasting is not arbitrary alignment: the dimensions must be compatible under NumPy’s rules. Its documentation defines broadcasting as how NumPy treats arrays with different shapes during arithmetic. Broadcasting can avoid materializing repeated copies, although some shape combinations can still have inefficient memory behavior. See the NumPy broadcasting guide for the compatibility rules.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Know when a NumPy slice changes the original
Basic NumPy slicing commonly returns a view that refers to the original array’s data. Editing that view can therefore change the original:
original = np.array([[1, 2], [3, 4]])
row_view = original[0]
row_view[0] = 99
print(original[0, 0]) # 99Call .copy() when you need independent array data:
independent = original[0].copy()
independent[0] = -1
print(original[0, 0]) # still 99Python list slicing creates a new outer list, but it does not recursively copy mutable objects inside that list. NumPy explains the view behavior and explicit-copy option in its copies and views guide.
Choose lists or NumPy based on the work
| Consideration | Nested Python lists | NumPy ndarray |
|---|---|---|
| Structure | Flexible sequences of Python objects; rows can be handled as lists. | Multidimensional structure with a shape and element dtype. |
| Indexing | Chained indexes, such as rows[1][2]. |
Comma-separated axis indexes, such as array[1, 2], plus multidimensional slicing. |
| Numeric operations | Use loops or other code to express elementwise calculations. | Elementwise arithmetic and broadcasting support concise numerical calculations. |
| Slicing | A slice makes a new list containing references to selected elements. | A basic slice commonly returns a view; use .copy() for independent data. |
| Good fit | Small, flexible, general-purpose nested data without a need for numerical array operations. | Regular numerical data where multidimensional operations, dtype control, or array indexing are useful. |
Use nested lists when their flexibility is enough. Choose NumPy when your data is a regular numerical grid and you want operations over whole arrays. The official documentation describes these behaviors and examples, but does not establish a universal speed or memory advantage for every workload; such comparisons depend on the data, operation, and environment.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
