Regardless of whether you use Windows, Mac, or Linux, we recommend using the Anaconda python package manager and Python 3.6 (or higher). The instructions below are for Unix. Please adapt accordingly if you are on a non-Unix environment.
We will offer very limited support at office hours for installing or maintaining Python on your laptop/desktop/cluster. It is your responsibility to acquire and maintain the packages and environments you need.
We have verified that these instructions successfully install the required environment on common laptops, a standard issue Tufts EECS unix account home directory or in a Tufts High Performance Computing account.
One Time Setup: Install "miniconda"
Expected time for this step: 15 minutes
Install Miniconda (which is a bare essentials version of Anaconda) to your $HOME directory, or whereever you prefer to install it.
The installation could be done by following the instructions here: https://conda.io/projects/conda/en/latest/user-guide/install/index.html
Tip for Unix users: When it asks whether to put Miniconda on your PATH within your .bashrc, say yes.
One Time Setup: Install all necessary Python packages (including autograd)
Expected time for this step: 10 minutes
Expected disk space: About 1GB (sorry! modern Python is quite bloated)
After you have conda installed, you can install a set of standard packages that we have prepared as a convenient .yml specification file: https://www.cs.tufts.edu/comp/135/2020f/conda_envs/comp135_2020f_env.yml
After downloading this file, you can setup the environment and install all required packages in just one terminal command:
$ conda env create -f comp135_2020f_env.yml
Use the environment
At a terminal:
$ conda activate comp135_2020f_env