Courses
Essential Computing Tools for Scientists
Environment Management
Environment Management with Conda

Environment Management with Conda

Authors
Dr. Atle Rimehaug | Dr. Ole Bialas | Dr. Nicholas Del Grosso | Dr. Sangeetha Nandakumar
Open in JupyterLite Download Exercises

Setup

Installation

Visual Studio Code (VSCode)

Download Link

Miniforge

Miniforge provides a streamlined approach to Conda installation with a minimal environment and access to the conda-forge community repository, which is popular for its extensive collection of data science and scientific computing packages. Miniforge is available for all major operating systems, making it a flexible choice for cross-platform development.

Installation Instructions

If you do not already have Conda installed please follow these instructions to install Miniforge.

macOS & Linux
  1. Download the Miniforge installer for your platform from Miniforge releases .
  2. Open a terminal and navigate to the download location.
  3. Run the installer: 
    bash Miniforge3-Linux-x86_64.sh  # or the macOS equivalent
  4. Follow the prompts to complete installation.
Windows
  1. Download the Miniforge installer for Windows from Miniforge releases .
  2. Open the installer and follow the prompts.
  3. Once installed, open a new terminal or Command Prompt.

After installation, verify that Conda is installed by running:

conda --version

In this unit, we will learn how to use Conda to manage packages and environments. Conda allows you to create isolated environments, manage dependencies, and install packages seamlessly across platforms. This unit includes instructions for installing Conda via Miniforge and covers core Conda commands for creating, managing, and sharing environments.

Section 1: Creating and Managing Environments

One of the key advantages of Conda is its ability to create isolated environments, which are essentially self-contained workspaces with their own packages and dependencies. This allows you to experiment, develop, and analyze data without worrying about conflicts between different package versions, which can often disrupt workflows. This is especially helpful for reproducible research: by keeping dependencies organized, you ensure that your projects are replicable across systems.

This section covers the basics of creating, inspecting, activating, and deactivating Conda environments.

Code Description
conda create -n env_name Create a new environment named env_name
conda create -p dir_name Create a new environment in the specified directory dir_name
conda env list List all existing Conda environments
conda activate env_name Activate the environment env_name
conda deactivate Deactivate the current Conda environment
conda info --envs Show paths and basic info about all environments

Exercises

  1. Creating an environment by name

    1. Create a new empty Conda environment named test_env. 
      conda create -n test_env
    2. List all environments to verify that test_env was created. 
      conda info --envs
    3. Activate test_env and confirm activation. 
      conda activate test_env
    4. Deactivate the environment. 
      conda deactivate

  2. Creating an environment at a specific path

    1. Create a new empty Conda environment in the directory test_env2. 
      conda create -p test_env2
    2. List all environments to verify that test_env2 was created. 
      conda info --envs
    3. In the File Explorer, verify that test_env2 folder exists.
    4. Activate test_env2 and confirm activation. 
      conda activate test_env2
    5. Deactivate the environment. 
      conda deactivate

Section 2: Installing Packages in Environments

Managing packages within Conda environments allows you to install libraries and dependencies specific to each project. This helps avoid the challenges of “dependency hell” where incompatible package versions interfere with each other. For researchers, Conda’s ability to install non-Python packages (like r-base or system libraries) provides flexibility and power, especially for complex data workflows involving multiple languages.

Other tools, such as pip, are primarily designed for Python packages, while Conda can handle cross-language dependencies. This makes it especially useful for mixed-language research workflows involving Python and R.

conda-forge contains a wealth of scientific packages for download. Here is their list: https://conda-forge.org/packages/

Code Description
conda install package_name Install package_name into the active environment
conda install -c conda-forge package_name Install package_name from the conda-forge channel into the active environment.
conda install -n env_name package_name Install package_name into env_name
conda list List all installed packages in the active environment

Exercises

  1. Creating an environment and installing Python.

    1. Create and activate new environment named py_env. 
      conda create -n py_env
conda activate py_env
    2. Install python=3.9 in py_env. 
      conda install python=3.9
    3. List all installed packages and verify that you see Python with version 3.9.xx installed. 
      conda list
    4. (Alternative) Type python --version to verify Python was installed. 
      python --version
    5. Deactivate the environment 
      conda deactivate

  2. Install r-base in r_env environment and verify the installation.

    conda create -n r_env
conda activate r_env
conda install r-base
conda list
conda deactivate

  3. Install python=2.7 in py_env2 environment and verify the installation.

    conda create -n py_env2
conda activate py_env2
conda install python=2.7
conda deactivate

Section 3: Exporting and Importing Environments

A powerful feature of Conda is the ability to share environment configurations with collaborators. By exporting an environment to a file, you can create a “snapshot” of all installed packages and dependencies. This is essential for reproducibility in research, as others can recreate the exact environment to ensure consistency in analyses and results. Using Conda environment files can save time and reduce errors, especially when working on collaborative projects.

Removing unused environments and packages helps keep your Conda installation clean and organized, conserving storage space and reducing potential conflicts. Conda makes it easy to remove entire environments or individual packages, allowing you to focus only on the active dependencies you need for your current work.

This section shows how to save an environment configuration to a file and recreate it later.

Code Description
conda env export > environment.yml Save active environment configuration to environment.yml
conda remove -n env_name --all Remove the environment env_name completely
conda remove package_name Remove package_name from the active environment
conda env create -f environment.yml Create an environment from the file environment.yml
conda env update --file environment.yml Update the environment from environment.yml

Exercises

  1. Export the test_env environment to a file named test_env.yml.

    conda activate test_env
conda env export > test_env.yml
conda deactivate

  2. Deactivate and remove test_env using conda remove -n test_env --all. List all the environments to verify that it has been removed.

    conda remove -n test_env --all
conda info --envs

  3. Recreate test_env from the test_env.yml file.

    conda env create -f test_env.yml
conda info --envs

  4. Verify the recreation by activating test_env and listing the installed packages.

    conda activate test_env
conda list

  5. Remove all the environments you don’t want to keep after this session.