Skip main navigation

New offer! Get 30% off your first 2 months of Unlimited Monthly. Start your subscription for just £29.99 £19.99. New subscribers only. T&Cs apply

Find out more

Anatomy of NumPy arrays

In this article we look more deeply into the internals of NumPy arrays.
© CC-BY-NC-SA 4.0 by CSC - IT Center for Science Ltd.

The datatype of a NumPy array is called ndarray.

If one looks into what an ndarray is actually made of, one can see that it consists of the following:

  • one dimensional contiguous block of memory: raw data
  • indexing scheme: how to locate an element
  • data type descriptor: how to interpret an element

Array layout

NumPy indexing

There are many possible ways of arranging the elements of a N-dimensional array in a 1-dimensional block (i.e. memory). NumPy uses striding where a N-dimensional index (n[0], n[1], …, n[-1]) corresponds to the offset from the beginning of a 1-dimensional block.

If n[k] is the index in dimension k for an element and s[k] is the stride in that dimension, then the offset for an element is:

offset = sum(s[k] * n[k] for k in range(N))

Array indexing

When one assigns a variable to a slice of another array

b = a[1:8:2, 3:12:3]

the variable b has the same raw data as a, but only different
strides. Thus, changing the contents of b will also change the contents of
a, i.e. b is view to a as discussed earlier.

In addition to slicing NumPy allows indexing also with integer arrays or
Boolean masks:

a = np.arange(0.0, 1.0, 0.1)
ind = np.array([1, 1, 0, 4])
b = a[ind] # b = array([0.1, 0.1, 0. , 0.4])

m = a > 0.5
b = a[m] # b = array([0.6, 0.7, 0.8, 0.9])

In these cases b cannot be created just by modifying the strides, so b
will hold a copy of the data in a. Now, modifications of b are not
affecting a.

Attributes of an ndarray

a = np.array(...)

  • a.flags : various information about memory layout
  • a.strides: bytes to step in each dimension when traversing
  • a.itemsize: size of one array element in bytes
  • Python buffer object pointing to start of arrays data
  • a.__array_interface__: Python internal interface

Try to investigate attributes of different types of arrays!

© CC-BY-NC-SA 4.0 by CSC - IT Center for Science Ltd.
This article is from the free online

Python in High Performance Computing

Created by
FutureLearn - Learning For Life

Reach your personal and professional goals

Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.

Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.

Start Learning now