Conda

From conda document:

Package, dependency and environment management for any language—Python, R, Ruby, Lua, Scala, Java, JavaScript, C/ C++, Fortran, and more.

Conda is an open source package management system and environment management system that runs on Windows, macOS, and Linux. Conda quickly installs, runs and updates packages and their dependencies. Conda easily creates, saves, loads and switches between environments on your local computer. It was created for Python programs, but it can package and distribute software for any language.

Installation

There are multiple methods to install conda.

AUR

Install the python-conda^AUR package, or the miniconda3^AUR package for miniconda.

miniforge

Download miniforge from the GitHub repository.

Then run the installer with

$ bash /path/to/Miniforge3-Linux-x86_64.sh

Miniforge uses conda-forge as the default channel.

Note: The Anaconda Inc. also provides a conda package with their default channel. To add conda-forge as the default channel run

$ conda config --add channels conda-forge
$ conda config --set channel_priority strict

Here list advantages of conda-forge over anaconda default channel.

Usage

It is recommended to install new packages in a new environment instead of the base environment. To avoid conda automatically activating the base environment, edit:

~/.condarc

auto_activate_base: false

Warning: Automatically activating the base environment will break some applications that depend on system python, since the python in your conda environment will miss some dependencies.

Set default packages

Add create_default_packages section in ~/.condarc. For example:

~/.condarc

...
create_default_packages:
  - pip
  - ipython
  - numpy
  - scipy
  - libgcc-ng
  - mpich
  - rust

Create an environment

To crate a new environment myenv with default packages specified in previous section, run:

$ conda create --name myenv

To crate a new environment without default packages, run:

$ conda create --no-default-packages --name myenv

To create a environment with specified python version and packages, run:

$ conda create --name myenv python=3.9 numpy=1.23.5 astropy

To activate the environment:

$ conda activate myenv

List environments

To list all environment:

$ conda env list

Clone an environment

$ conda create --name myenvclone --clone myenv

Remove an environment

$ conda remove --name myenv --all

Export and import an environment

To export all packages in myenv environment:

$ conda activate myenv
$ conda env export > myenv.yml

If you only want to include packages you explicitly installed, add --from-history flag when exporting.

$ conda env export --from-history > myenv.yml

To create a new environment from a myenv.yml, run:

$ conda env create -f myenv.yml

Troubleshooting

qt.qpa.plugin: Could not find the Qt platform plugin "wayland" in ""

Install qt6-wayland to conda environment.

$ conda install qt6-wayland