Before installing Scientific Python libraries, you need to have Python itself installed. There are two, largely equivalent, ways of doing that, and we describe both below.
If you have a working version of Python on your system already (check
by running python3
), you can skip to setting up a virtual environment.
This is the official Python distribution, which uses the pip
package manager. pip
installs packages from Python Package Index, or PyPI for short.
Download the installer from https://www.python.org/downloads/.
A virtual environment is a workspace into which you can install Python libraries, separate from what is being used by your operating system.
Create a new virtual environment in a directory called
py3
:
python -m venv py3
Start using it as follows:
source py3/bin/activate
Also, make sure you have pip
installed—that is Python’s default package manager:
python -m ensurepip
You are now ready to install Scientific Python packages using pip
! For example:
pip install ipython numpy scipy
You should now be able to run IPython (the interactive Python shell) to try out NumPy:
$ ipython
In [1]: import numpy as np
In [2]: np.linspace(0, 10, 5)
Out[2]: array([ 0. , 2.5, 5. , 7.5, 10. ])
Miniforge is a small Python distribution based around the conda package manager, and installs packages from the community repository conda-forge.
Conda is a bit different from Python’s pip
package manager in that
it can, in addition to Python libraries, also install compilers,
libraries, and so forth.
Download the latest version from GitHub.
Back up your shell init (~/.zshrc
, ~/.bashrc
, etc.), since the installer will modify these.
Run the installer (typically, sh Miniforge3-Linux-x86_64.sh
at the terminal), and when it asks you “Do you wish the installer to
initialize Miniforge3 by running conda init?” enter “yes”.
If you don’t like the changes made to your shell init, restore it from backup,
and enable miniforge
with source ~/miniforge3/bin/activate
.
A virtual environment is a workspace into which you can install Python libraries, separate from what is being used by your operating system.
Create a new virtual environment called science
(or name it whatever
you like!):
conda create -n science
Switch to the new environment (you need to do this every time you want to use it):
conda activate science
You are now ready to install Scientific Python packages using conda
!
For example:
conda install ipython numpy scipy
You should now be able to run IPython (the interactive Python shell) to try out NumPy:
$ ipython
In [1]: import numpy as np
In [2]: np.linspace(0, 10, 5)
Out[2]: array([ 0. , 2.5, 5. , 7.5, 10. ])