Managing Environments in Anaconda

Tutorial for managing conda environments

https://conda.io/docs/user-guide/tasks/manage-environments.html

Part 0: Setup

1. Make sure Anaconda is installed: https://www.anaconda.com/download/

Part 1: Installing a new environment

1. Create new environment, called "env1" with python 2.7 the package, ipython:

   conda create --name env1 python=2.7 ipython
   

2. Activate your new environment

   source activate env1 # remove source for windows
   

3. Use ipython to see if it worked

   ipython
   
   import sys
   sys.executable
   sys.version
   quit
   

   You should see something similar to this:

   (env1) D-10-19-251-209:~ cswitzer$ ipython
   Python 2.7.15 |Anaconda, Inc.| (default, May  1 2018, 18:37:05) 
   Type "copyright", "credits" or "license" for more information.
   .
   .
   .
   
   In [1]: import sys
   
   In [2]: sys.executable
   Out[2]: '/Users/cswitzer/anaconda/envs/env1/bin/python'
   
   In [3]: sys.version
   Out[3]: '2.7.15 |Anaconda, Inc.| (default, May  1 2018, 18:37:05) 
   [GCC 4.2.1 Compatible Clang 4.0.1 (tags/RELEASE_401/final)]'
   
   

4. Deactivate the environment

   source deactivate
   

• Create new environment, called "env2" with python 3.7 and the package ipython
• Activate that environment.
• Run ipython, and check to see if you're using the python version from your new environment

Part 2: Installing a jupyter kernel, so you can select this environment in jupyter notebook

1. Install ipykernel in your new environment

   source activate env1
   conda install ipykernel
   

2. Use the package ipykernel to install kernel, so you can use that environment in jupyter notebook

   • install kernelspec for this user

     python -m ipykernel install --user --name env1
     

   • check to see if the kernel has been installed

     jupyter kernelspec list
     

   • deactivate environment

     source deactivate #remove "source" for windows
     

   • run jupyter lab

     jupyter lab
     

   • Start a new jupyter noetbook

   • check to see if you can select your new environment (Kernel -> change kernel -> env1)

   • use sys to check the environment and python version

• install a kernel for your new environment env2
• Run jupyter lab, and check to see if the new environment is available

Part 3: Using an environment.yml file

1. Download the file environment.yml https://github.com/callinSwitzer/conda_env_tutorial/blob/master/environment.yml
2. In terminal (or anaconda prompt), navigate to the folder where you saved this file.
3. Create a new environment, using this file (may take ~2 min to finish)

   conda env create -f environment.yml
   

4. List environments

   conda env list
   

5. Install kernel

   source activate machLearn_env
   python -m ipykernel install --user --name machLearn_env
   

• Check to see if this kernel is available in Jupyter lab

Part 4: Making an environment.yml file

1. Create an exact environment.yml file (note that this will overwrite any environment.yml file in your current directory)

source activate env1
conda env export > environment2.yml

2. Edit this exported file (by hand) to make it simpler

Part 5: Removing unwanted environments and kernels

1. List all conda environments

conda env list

2. To remove an environment, delete it's folder

3. List kernels

jupyter kernelspec list

4. To remove a kernel, delete it's folder

*** OPTIONAL: Remove unused packages and caches

conda clean --all --dry-run  # do a dry run first, without deleting anything

conda clean --all # actuall delete the things