Jake Vanderplas

University of Washington

Fall, 2013

This repository will contain all the curriculum and materials for the Astronomy 599 seminar, Fall 2013.

Jake Vanderplas is an NSF post-doctoral fellow at UW working jointly between the Computer Science department's Database Research Group and the Astronomy department's Survey Science Group. He received his PhD in Astronomy from the University of Washington in 2012. He is active in the open-source Scientific Python community, and actively contributes to and helps maintain several of the core packages in the SciPy ecosystem.

PAB 356 / 356A

• Two-day “Boot Camp”: September 19-20 / 23-24 9:00am-4:00pm (dates TBD)
• Weekly meetings: Mondays, 3:00-4:00pm throughout Fall quarter

This graduate seminar course will offer a comprehensive introduction to scientific computing in Python, geared toward students and researchers in Astronomy. The course is aimed at those who already have some familiarity with computational tools and programming languages -- i.e. who know what a terminal is and how to use it, and are familiar with basic programming concepts such as for loops and if-else statements -- though it will assume no familiarity with the Python programming language.

The two-day boot camp in the days before the start of the quarter will begin with a broad introduction to the basic syntax of Python: data types and operations, loops and conditionals, input and output of data, and basic object-oriented programming. From that base, we will move on to a broad overview of the main computing tools of scientific computing in Python: IPython for an interactive computing environment, NumPy for handling of array-based data, Matplotlib for visualization of this data, and Scipy for common tasks such as integration, interpolation, optimization, and linear algebra. In addition to this, we will introduce the use of Git, a distributed version control system, which we will be using throughout the quarter to keep track of code and assignments.

The hour-long seminar slots each week during the quarter will build upon the knowledge-base covered in the boot camp, and introduce more advanced Python tools such as Scikit-learn for machine learning and data mining, Scikit-image for pixel-level data processing, SymPy for symbolic mathematics, IPython for parallel processing, Cython for writing fast algorithms, and much more. Each week will include a short 1-2 hour take-home assignment aimed at reinforcing the concepts covered during class. The assignments will focus on tasks that will be useful in the day-to-day work of astronomical research.

While the course is designed for graduate students who are in their first year or two of research, it is also open to any faculty, researchers or others who would like to brush-up on their Python skills. Upper-level undergraduates and students from other departments will also be considered on a case-by-case basis. For more information, please email [email protected]

This will be a hands-on course, so it's vital that you come with a computer set up to use Python.

The instructions below assume you have access to a bash/csh shell. If you're on Mac or Linux, this is easy: just open a terminal and you have it. If you're on windows, this is much harder. Windows has a shell available, but it's not nearly as powerful as the bash/csh shell on Linux and Mac. Working with most scientific software is a pain on Windows: the best option is probably to install a dual-boot Linux system or a linux virtual machine. This is not reflective of any anti-windows bias on my own part: it's simply a reflection of the fact that in general, the scientific tools you'll use in your research career will be very difficult to use on Windows.

Another option is to sign-up for the free web-based wakari service. It's a browser-based scientific computing environment that lets you run all the things you'll need on a cloud-based computer. It's not quite as nice as having everything running locally, but it's a good option.

When you have your terminal up and running, make sure you have installed the git version management system. You can check if it's already installed on your system by typing at the command-line

$ git --version

Version 1.6 or better should be sufficient.

When you have git installed, please clone this repository. You can do this by navigating to the directory you'd like to use, and typing

$ git clone https://github.com/jakevdp/2013_fall_ASTR599.git

for HTTP, or for SSH

$ git clone [email protected]:jakevdp/2013_fall_ASTR599.git

This will create a directory 2013_fall_ASTR599.

For your Python installation, I recommend using [Anaconda](https://store.continuum.io/], a free all-in-one Python installer for Windows, Mac, and Linux. It contains up-to-date versions of all the packages we will use in this course. Even if you have a local Python installation on your laptop, installing Anaconda can make your life easier.

Note that after installing Anaconda, you should type

$ which python

and make sure it points to the anaconda version of the Python program. If not, you'll have to adjust your $PATH variable to point to this version.

Once you have Python installed, you can use the script provided in this repository to check the versions of the various packages. In the scripts subdirectory of 2013_fall_ASTR599, there is a script called check_installation.py. This will check the versions of various Python packages you have installed on your system. At the command line, type

$ cd 2013_fall_ASTR599/scripts
$ python check_installation.py

If you get any warnings, you should either upgrade those packages yourself, or switch to an Anaconda build.

We'll be making use of IPython notebook, a browser-based Interactive Development Environment (IDE) for Python. To make sure it's set up, type

$ ipython notebook

A browser window should open to your IPython dashboard. If it doesn't, I'd recommend installing Anaconda Python; this comes equipped with everything you need for IPython notebook.

The last thing you'll need is a good text editor. This should not be a word processor, but a program that lets you write simple text files. Preferably it should also have syntax highlighting for code. Something simple like gedit or textmate is fine. Many people like emacs, vim, and other more powerful options. Find one that works for you and make sure you can create, edit, and save text files.