Cython Cheat Sheet Cython (https://cython.org/) is a C extension to the Python programming language that allows for static compi- lation and strong type declaration. It’s aim is to speed up Python significantly. Compiling Cython Class and function declaration Jupyter notebooks (temporary) Classes: %load_ext cython cdef class import Cython TestClass(object): # making variables accessible from Python # write code using cython # using public: # run cell cdef public int # nb! print functions have to be used a,b cdef public double # %%cython has to be written c # to each cell using it def # %%cython -a outputs function documentation __init__(self): # (which parts are compiled to C self.a=0 # and which are Python) self.b=1 self.c= 3.14 %% cython Functions: cdef int multiplyWithOne(int x): x*=1; def functionExample(int x, float y): return x # function is callable from Python and Cython print(multiplyWithOne(9)) return y*x ... pyximport (temporary) cdef functionExample(int x, float y): 1. write a file.pyx containing the source code # function is callable from Cython only 2. import (and autocompile) file.pyx using pyximport: # function is optimized return y*x # using pyximport for auto-compilation import pyximport cpdef functionExample(int x, float y): pyximport.install() # function is callable from Cython (optimized) import file #imports file.pyx # function is callable from Python (unoptimized) setup.py script return y*x 1. write a file.pyx containing the source code 2. compile it using a setup.py file containing: Further, the return type can be added to the function: from distutils.core import setup from Cython.Build import cythonize cdef float functionExample(int x, float y): # function is optimized & callable from Cython only setup(ext_modules= cythonize( ``file.pyx'')) return y*x 3. compile it: $ python setupy.py build_ext --inplace Type declaration Files .pyx Source code containing Cython code Cython offers static typing of the following types: .c C source code generated from .pyx file int A = 1 integer .so shared object file from compilation float A = 1.0 floating point (fp) .html documentation from %%cython -a double A = 1.0 double precision fp ./build/ temporary files from compilation char *A = ’1’ string (pointer (*)!!) Py_ssize_t signed integer for NumPy indices Misc Memory management Calling external C functions cdef extern from ''file.h'': Manual memory management can be done by importing double someFunctionsName(double) memory handling functions from the C standard library: from libc.math cimport cos void* malloc cdef double functionName(double x) void* realloc return cos(x) void free ©Simon Wenkel (https://www.simonwenkel.com) This pdf is licensed under the CC BY-SA 4.0 license..