February 27, 2015 - 11:02
![Doug Blank's picture Doug Blank's picture](https://secure.gravatar.com/avatar/7a9581b968c198e3c08ffe747e69bb5b.jpg?d=https%3A//ns1.serendipstudio.org/oneworld/files/pictures/SerendipStudioAvatar.png&s=100&r=G)
This post captures some good reasons for having an algorithm in Python, rather than simply calling a fast library function:
https://jakevdp.github.io/blog/2015/02/24/optimizing-python-with-numpy-and-numba/
Pure-Python code is easier to read, understand, and contribute to.
Pure-python packages are much easier to install than Python-wrapped C or Fortran code.
Pure-python code often works for many data types.
Pure-python is easier to use at scale.
-Doug