Using Conda and Conda Command Cheatsheet

Installing Conda package on macOS

Do:

brew install --cask conda

This will download miniconda, coda and anaconda3 and all other required affiliated packages. The install location may vary, but is most likely to be in /usr/local/anaconda3/

After the installation, add the path to the anaconda3 installation directory to the $PATH system variable, by:

echo 'export PATH="/usr/local/anaconda3/bin:$PATH"' >> ~/.zshrc
source ~/.zshrc

Installing Conda package on Linux server / remote cluster

It’s best to install the “miniconda” essential-only distribution, since the IDE, Jupyter-notebook, Spyder etc. will likely not be needed on a remote server / cluster! Do:

wget https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh

Test installation with conda list. This should show a list of installed packages.

See: https://docs.conda.io/projects/conda/en/latest/user-guide/install/linux.html

Some older systems like the CentOS-based IPMU-GPU server may give out the following error, upon running the bash script:

Installer requires GLIBC >=2.28, but system has 2.17.

The ideal solution in this case is to install an older compatible version of conda instead of the latest one! Do:

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_22.11.1-1-Linux-x86_64.sh
bash Miniconda3-py39_22.11.1-1-Linux-x86_64.sh

This will install an older compatible version of the “conda” package. Do source ~/.bashrc to have the conda command available to the terminal. conda list should then work seamlessly!

Initializing Conda Environment

Initialize conda for your shell permanently:

conda init

This will add command lines to your ~/.zshrc file to initialize conda path variables at each startup. The modified ~/.zshrc file looks like:

# export PATH="/usr/local/anaconda3/bin:$PATH"  # commented out by conda initialize

# >>> conda initialize >>>
# !! Contents within this block are managed by 'conda init' !!
__conda_setup="$('/usr/local/anaconda3/bin/conda' 'shell.zsh' 'hook' 2> /dev/null)"
if [ $? -eq 0 ]; then
    eval "$__conda_setup"
else
    if [ -f "/usr/local/anaconda3/etc/profile.d/conda.sh" ]; then
        . "/usr/local/anaconda3/etc/profile.d/conda.sh"
    else
        export PATH="/usr/local/anaconda3/bin:$PATH"
    fi
fi
unset __conda_setup
# <<< conda initialize <<<

Do a source ~/.zshrc run after conda init.

Creating Conda Environments

Now, a new conda environment can be easily created using:

conda create -n ENV_NAME

If there is a environment configuration file available, you can pass the file name too:

conda env create -n ENV_NAME --file=FILE_PATH

If just conda env create -n ENV_NAME is used, without passing any filename path for the environment configuration file.

Check the list of all available conda enviroments using :

conda info --envs
conda env list

See : https://docs.conda.io/projects/conda/en/stable/user-guide/getting-started.html

Installing New Packages

Individual packages can be installed into this environment by:

conda activate ENV_NAME
conda install PKG_NAME
pip3 install PKG_NAME

To install packages to a specific environment, do:

conda install --name ENV_NAME -c conda-forge PKG_NAME

where, -c conda-forge denotes the conda channel to use for the download.

Edit global list of default channels

Usually most of the packages that I am interested in are only available in conda-forge. This channel can be added to the list of default channels via the following. If this is done, then we don’t have to use: -c conda-forge:

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