Resources: python & jupyter

Posted on 2024-01-01 by George McNinch

Overview

Our course will use the python programming language to perform some machine computations needed to carry out mathematical modeling tasks.

And parts of the notes of the course lectures will be made available as jupyter notebooks.

There are a few ways that you should learn to interact with python and jupyter for our class:

  1. via Google’s colab
  2. by executing python code in a python interpreter on your computer
  3. (optionally) using jupyter notebooks on your computer.

Here is a bit of what you need to know to enable these interactions:

Before diving into the details, let me try to answer a few question that you may have:

Q - do I need to install juptyer on my computer, or I just use colab?

A - It is probably possible to just use colab. colab provides a reasonable environment for reading the course material, though for code execution it may (?) be a tad slow.

But: using python exclusively through colab is a somewhat imperfect environment for producing code, and at least in the long run you’ll be better served by learning to work with these tools on your own computer.

An overview of jupyter notebooks

Jupyter notebooks provide a way to package written material – with possible mathematical markup – together with an interactive environment for executing computer code.

The material for this course will mostly be presented in the form of these notebooks. You can recognize notebook files by their filename extension – their file name has the form *.ipynb.

On the course site, notebooks will be posted with two links – one to colab, and one to an *.ipynb file.

  • The simplest way to interact with jupyter notebooks is to use Google’s colab. Clicking one of the links will open the notebook in the cloud via the site colab.research.google.com.

  • The other one link permits you to download the notebook file to your computer. In order to use this file, you will need to have a working installation of python and jupyter. See below for some discussion of how to set up the required tools and how to use jupyter.

Interacting with jupyter notebooks via Google’s CoLab

As mentioned above, the course material will be presented in the form of jupyter notebooks, and for each notebook there will be a link to a copy of this notebook available via colab.

Colab is a web-based free-to-use Jupyter notebook environment that runs in the cloud and stores its notebooks on Google Drive. When you view a notebook in colab containing code, that code can be executed. The code execution occurs in the cloud, which basically means on some google computer. Thus, google’s computer has python and the relevant libraries installed and available.

You can choose to interact with the course notebooks on colab, abd create your own notebooks there.

A good starting point to learn how to use colab (and jupyter notebooks in general) is the document Overview of Colaboratory Features.

Installing the tools on your computer

Here I hope to answer the questions: “What do I need to do to be able to view and execute jupyter notebooks on my computer?”

You first need to have python installed. In many cases, this is already true.

We are going to install jupyter and some python libraries in a virtual environment; you can read about virtual environments here.

To begin, create a directory named jupyter somewhere on your computer (you can actually call it whatever you like…)

Open a terminal on your computer, and change to the directory you just created. To do so, use a command something like:

george@valhalla:~$ cd jupyter
george@valhalla:~/jupyter$

(You should only type the cd jupyter bit. And you may need to type a more complicated string if the directory you created is nested below your home directory).

Now create a virtual environment as follows:

$ python -m venv .venv

You may need to instead type

$ python3 -m venv .venv

Among other things, this should create a directory named .venv beneath your jupyter directory.

You now need to activate the virtual environment in your terminal. In linux or on a mac, you do this as follows:

$ source .venv/bin/activate

Looking at the docs https://docs.python.org/3/library/venv.html is appears that in windows you proceed as follows

# in cmd.exe
C:\jupyter\> .venv/Scripts\activate.bat

#or
# in powershell
C:\jupyter\> .venv\Scripts\Activate.ps1

After activation, the terminal “prompt” should appear differently. For example, you might see something like this:

(.venv) george@valhalla:~/jupyter$

You can no exit the virtual environment by typing deactivate, like this:

(.venv) george@valhalla:~/jupyter$ deactivate
george@valhalla:~/jupyter$

Now, with the virtual environment active, you can install the following packages using a command called pip:

(.venv) george@valhalla~/jupyter$ pip install jupyter numpy sympy scipy pandas pydot matplotlib

pip will install all these packages in your jupyter/.venv directory. In this case, packages occupying ~750M of disk space are installed and can be removed simply by deleting your jupyter directory.

Interacting with jupyter on your computer

You can always view and interact with jupyter notebooks on colab. But if you carried out the above instructions for installing conda, you can now use jupyter notebooks on your own computer.

Start jupyter from the command-line. I actually find that I prefer the jupyter lab interface to the the plain jupyter notebook interface (and honestly I’m not sure I understand why there are two different interfaces) but you should experiment for yourself.

So, with you virtual envirnoment activated type:

(.venv) george@valhalla:~$ jupyter-lab

There will be some output to the terminal that you can mostly ignore. The significant effect of running this program is that a page should open in your web browser.

The page that opens will have a URL that looks something like:

http://localhost:8888/lab

This URL tells you that there is now a server running on your local machine, and that your web browser is communicating with that server.

In the web page at this location, you should see in particular an interface where you can interact with jupyter notebooks. You can create a new notebook from the File menu: New -> Notebook. If you are asked to Select Kernel, you should select Python 3.

You are now editing a file name Untitled.ipynb.

If you save a jupyter notebook (i.e. a ipynb file) in your directory tree, you can open it using the File menu: Open from path

You can end the jupyter notebook session by typing C-c (twice) in the terminal where you started the command, or by menu-driven commands in the web-interface.

Some learning resources for python and scientific computing

(Honestly, I just borrowed this list from a previous instructor, but the links look useful).