💨 Compiling with Cython

Cython (not to be confused with CPython) allows Python code to be compiled to C, offering significant performance enhancements for computational heavy tasks or when integrating with C libraries.

First, you need to install Cython. It can be installed via pip: pip install cython

Let's start with a basic example of a Python function that sums the squares of numbers up to a given value. Here’s how you might write it in pure Python:

def sum_of_squares(n):
    total = 0
    for i in range(n):
        total += i ** 2
    return total

To optimise this with Cython, you'd create a .pyx file, say example_cy.pyx, and include the types:

def sum_of_squares(int n):
    cdef int total = 0
    cdef int i
    for i in range(n):
        total += i ** 2
    return total

To compile the Cython code, you need a setup script, setup.py, containing:

from distutils.core import setup
from Cython.Build import cythonize

setup(ext_modules = cythonize("example_cy.pyx"))

Run the compilation process with:

python setup.py build_ext --inplace

This generates a shared object file (.so or .pyd, depending on the OS) that you can import into Python.

Using Cython in Jupyter Notebooks

For quick experiments, Cython can be used directly in Jupyter notebooks with the %load_ext Cython magic. Here’s how you can use it:

%load_ext Cython

%%cython
def sum_of_squares(int n):
    cdef int total = 0
    cdef int i
    for i in range(n):
        total += i ** 2
    return total

Profiling and Annotating

Cython can also generate an HTML file showing which lines of code are converted to pure C and which ones invoke the Python C-API (potentially slowing down execution). Compile with the -a option in your setup script, or use the %%cython -a magic in Jupyter notebooks to generate this annotation. Lines highlighted in yellow indicate Python interactions.

Get PDF