Erik Reinertsen     Blog     Research     Tech     Teaching     Talks

Set up Miniconda on Aguirre Lab machine


Here are instructions for a user of an Aguirre Lab machine to set up Miniconda.

Miniconda is installed in the administrator account, aguirrelab. You should access and use this installation, and can modify it and install packages.

Activate:

$ source /home/aguirrelab/miniconda3/bin/activate

Your console prompt should now be prepended by (base).

Create an environment from your environment.yml. My usual environment is in my dotfiles repo. It is called py37 and it contains Python 3.7.7 (Tensorflow 2 does not yet support Python 3.8) and a variety of popular packages such as Tensorflow, Numpy, Pandas, h5py, etc. Your repo may have its own environment.yml, or you can create an environment with commands.

$ conda env create -f environment.yml
$ conda activate py37

Your prompt should now be prepended by (py37) (or whatever your environment name is).

If you use my .zshrc, you are done!

If not, update your .zshrc so when you launch a new shell it knows the paths to conda:

$ conda init zsh

Append to the end of your ~/.zshrc the following to always activate the py37environment when you log in:

CONDA_CUSTOM_ENV="py37"
conda deactivate; conda activate $CONDA_CUSTOM_ENV

The deactivation following by reactivation seems redundant but it is required to avoid problems with tmux.

Check to ensure your path has the miniconda environment version of Python, and that the version is correct:

$ which python
/home/aguirrelab/miniconda3/envs/py37/bin/python
  
$ python --version
Python 3.7.7