The scientific python stack

The scientific python stack#

Unlike Matlab, the set of Python tools used by scientists does not come from one single source. It is the result of a non-coordinated, chaotic and creative development process originating from a community of volunteers and professionals.

In this chapter I will shortly describe some of the essential tools that every scientific python programmer should know about. It is not representative or complete: it is just a list of packages I happen to know about, and I surely missed many of them.

Python’s scientific ecosystem#

The set of python scientific packages is sometimes referred to as the “scientific python ecosystem”. I did not find an official explanation for this name, but I guess that it has something to do with the fact that many packages rely on the others to build new features on top of them, like a natural ecosystem.

Jake Vanderplas made a great graphic in a 2015 presentation (the video of the presentation is also available here if you are interested), and I took the liberty to adapt it a little bit:

img

The core packages#

Numpy provides the N-dimensional arrays necessary to do fast computations, and SciPy adds the fundamental scientific tools to it. SciPy is a very large package and covers many aspects of the scientific workflow. It is organized in submodules, all dedicated to a specific aspect of data processing. For example: scipy.integrate, scipy.optimize, or scipy.linalg. Matplotlib is the traditional package to make graphics in python.

Essential numpy “extensions”#

There are two packages which I consider essential when it comes to data processing:

They both add a layer of abstraction to numpy arrays, giving “names” and “labels” to their dimensions and the data they contain.

Domain specific packages#

There are so many of them! I cannot list them all, but here are a few that you will probably come across in your career:

Geosciences/Meteorology:

  • MetPy: the meteorology toolbox

  • Cartopy: maps and map projections

  • xESMF: Universal Regridder for Geospatial Data

  • xgcm: General Circulation Model Postprocessing with xarray

  • GeoPandas: Pandas for vector data

  • Rasterio: geospatial raster data I/O

Statistics/Machine Learning:

Miscellaneous:

Contents