Cython#Example

Cython works by producing a standard Python module. However, the behavior differs from standard Python in that the module code, originally written in Python, is translated into C. While the resulting code is fast, it makes many calls into the CPython interpreter and CPython standard libraries to perform actual work. Choosing this arrangement saved considerably on Cython's development time, but modules have a dependency on the Python interpreter and standard library.

Although most of the code is C-based, a small stub loader written in interpreted Python is usually required (unless the goal is to create a loader written entirely in C, which may involve work with the undocumented internals of CPython). However, this is not a major problem due to the presence of the Python interpreter.{{Cite web |title=Basic Tutorial — Cython 3.0a6 documentation |url=https://cython.readthedocs.io/en/latest/src/tutorial/cython_tutorial.html |access-date=11 December 2020 |website=cython.readthedocs.io}}

Cython has a foreign function interface for invoking C/C++ routines and the ability to declare the static type of subroutine parameters and results, local variables, and class attributes.

A Cython program that implements the same algorithm as a corresponding Python program may consume fewer computing resources such as core memory and processing cycles due to differences between the CPython and Cython execution models. A basic Python program is loaded and executed by the CPython virtual machine, so both the runtime and the program itself consume computing resources. A Cython program is compiled to C code, which is further compiled to machine code, so the virtual machine is used only briefly when the program is loaded.{{cite web |last=Oliphant |first=Travis |date=20 June 2011 |url=http://technicaldiscovery.blogspot.com/2011/06/speeding-up-python-numpy-cython-and.html |title=Technical Discovery: Speeding up Python (NumPy, Cython, and Weave) |publisher=Technicaldiscovery.blogspot.com |access-date=21 July 2013}}{{cite journal |last1=Behnel |first1=Stefan |last2=Bradshaw |first2=Robert |last3=Citro |first3=Craig |last4=Dalcin |first4=Lisandro |last5=Seljebotn |first5=Dag Sverre |last6=Smith |first6=Kurt |year=2011 |title=Cython: The Best of Both Worlds |journal=Computing in Science and Engineering |volume=13 |issue=2 |pages=31–39 |doi=10.1109/MCSE.2010.118 |bibcode=2011CSE....13b..31B |s2cid=14292107 |url=http://research.google.com/pubs/pub36727.html|hdl=11336/13103 |hdl-access=free }}

{{cite book |last=Seljebot |first=Dag Sverre |title=Proceedings of the 8th Python in Science Conference |chapter=Fast numerical computations with Cython |year=2009 |pages=15–22 |doi=10.25080/GTCA8577 |chapter-url=http://conference.scipy.org/proceedings/SciPy2009/paper_2|doi-access=free }}{{cite journal |last1=Wilbers |first1=I. |last2=Langtangen |first2=H. P. |last3=Ødegård |first3=Å. |year=2009 |title=Using Cython to Speed up Numerical Python Programs |journal=Proceedings of MekIT'09 |pages=495–512 |url=http://simula.no/research/sc/publications/Simula.SC.578/simula_pdf_file |access-date=14 June 2011 |format=PDF |editor1-first=B. |editor1-last=Skallerud |editor2-first=H. I. |editor2-last=Andersson |archive-date=4 January 2017 |archive-url=https://web.archive.org/web/20170104170007/http://simula.no/research/sc/publications/Simula.SC.578/simula_pdf_file |url-status=dead }}

Cython employs:

• Optimistic optimizations
• Type inference (optional)
• Low overhead in control structures
• Low function call overhead{{cite web |url=http://telecom.inescporto.pt/~gjc/pybindgen-benchmarks/ |title=wrapper benchmarks for several Python wrapper generators (except Cython) |access-date=28 May 2010 |archive-url=https://web.archive.org/web/20150404154630/http://telecom.inescporto.pt/~gjc/pybindgen-benchmarks/ |archive-date=4 April 2015 |url-status=dead}}{{cite web |url=http://behnel.de/cycppbench/ |title=wrapper benchmarks for Cython, Boost.Python and PyBindGen |access-date=28 May 2010 |archive-url=https://web.archive.org/web/20160303224104/http://www.behnel.de/cycppbench/ |archive-date=3 March 2016 |url-status=dead}}

Performance depends both on what C code is generated by Cython and how that code is compiled by the C compiler.{{cite web |title=Cython: C-Extensions for Python |url=http://cython.org/index.html |accessdate=22 November 2015}}

Cython is a derivative of the Pyrex language, but it supports more features and optimizations than Pyrex.{{cite web |url=https://github.com/cython/cython/wiki/DifferencesFromPyrex |title=Differences between Cython and Pyrex|website=GitHub }}{{cite web |last=Ewing |first=Greg |url=http://mail.python.org/pipermail/python-dev/2011-March/109642.html |title=Re: VM and Language summit info for those not at Pycon (and those that are!)|type=Message to the electronic mailing-list python-dev |date=21 March 2011 |access-date=5 May 2011}} Cython was forked from Pyrex in 2007 by developers of the Sage computer algebra package, because they were unhappy with Pyrex's limitations and could not get patches accepted by Pyrex's maintainer Greg Ewing, who envisioned a much smaller scope for his tool than the Sage developers had in mind. They then forked Pyrex as SageX. When they found people were downloading Sage just to get SageX, and developers of other packages (including Stefan Behnel, who maintains the XML library {{proper name|LXML}}) were also maintaining forks of Pyrex, SageX was split off the Sage project and merged with {{proper name|cython-lxml}} to become Cython.{{cite web |author=Says Sage and Cython developer Robert Bradshaw at the Sage Days 29 conference |url=https://www.youtube.com/watch?v=osjSS2Rrvm0 |archive-url=https://ghostarchive.org/varchive/youtube/20211221/osjSS2Rrvm0 |archive-date=21 December 2021 |url-status=live |title=Cython: Past, Present and Future |via=YouTube |date=22 March 2011 |access-date=5 May 2011}}{{cbignore}}

Cython files have a .pyx extension. At its most basic, Cython code looks exactly like Python code. However, whereas standard Python is dynamically typed, in Cython, types can optionally be provided, allowing for improved performance, allowing loops to be converted into C loops where possible. For example:

1. The argument will be converted to int or raise a TypeError.

def primes(int kmax):

# These variables are declared with C types.

cdef int n, k, i

# Another C type

cdef int p[1000]

# A Python type

result = []

if kmax > 1000:

kmax = 1000

k = 0

n = 2

while k < kmax:

i = 0

while i < k and n % p[i] != 0:

i = i + 1

if i == k:

p[k] = n

k = k + 1

result.append(n)

n = n + 1

return result

File:Cython CPython Ext Module Workflow.png

A sample hello world program for Cython is more complex than in most languages because it interfaces with the Python C API and setuptools or other PEP517-compliant extension building facilities.{{technical inline|date=May 2024}} At least three files are required for a basic project:

• A setup.py file to invoke the setuptools build process that generates the extension module
• A main python program to load the extension module
• Cython source file(s)

The following code listings demonstrate the build and launch process:

1. hello.pyx - Python module, this code will be translated to C by Cython.

def say_hello():

print("Hello World!")

1. launch.py - Python stub loader, loads the module that was made by Cython.

1. This code is always interpreted, like normal Python.
2. It is not compiled to C.

import hello

hello.say_hello()

1. setup.py - unnecessary if not redistributing the code, see below

from setuptools import setup

from Cython.Build import cythonize

setup(name = "Hello world app",

ext_modules = cythonize("*.pyx"))

These commands build and launch the program:

$ python setup.py build_ext --inplace

$ python launch.py

A more straightforward way to start with Cython is through command-line IPython (or through in-browser python console called Jupyter notebook):

In [1]: %load_ext Cython

In [2]: %%cython

...: def f(n):

...: a = 0

...: for i in range(n):

...: a += i

...: return a

...:

...: cpdef g(int n):

...: cdef long a = 0

...: cdef int i

...: for i in range(n):

...: a += i

...: return a

...:

In [3]: %timeit f(1000000)

10 loops, best of 3: 26.5 ms per loop

In [4]: %timeit g(1000000)

1000 loops, best of 3: 279 µs per loop

which gives a 95 times improvement over the pure-python version. More details on the subject in the official quickstart page.{{Cite web |url=https://cython.readthedocs.io/en/latest/src/quickstart/build.html |title=Building Cython code |website=cython.readthedocs.io |access-date=24 April 2017}}

Cython is particularly popular among scientific users of Python,{{cite web |title=inSCIght: The Scientific Computing Podcast |url=http://inscight.org/2011/03/31/episode_/ |type=Episode 6 |access-date=29 May 2011 |archive-url=https://web.archive.org/web/20141010032300/http://inscight.org/2011/03/31/episode_/ |archive-date=10 October 2014 |url-status=dead}}{{cite journal |last1=Millman |first1=Jarrod |last2=Aivazis |first2=Michael |year=2011 |title=Python for Scientists and Engineers |journal=Computing in Science and Engineering |volume=13 |issue=2 |pages=9–12 |doi=10.1109/MCSE.2011.36 |bibcode=2011CSE....13b...9M |url=https://escholarship.org/uc/item/93s2v2s7}} where it has "the perfect audience" according to Python creator Guido van Rossum.{{cite web |author=Guido Van Rossum |url=http://mail.python.org/pipermail/python-dev/2011-March/109634.html |title=Re: VM and Language summit info for those not at Pycon (and those that are!) |type=Message to the electronic mailing-list python-dev|date=21 March 2011 |access-date=5 May 2011}} Of particular note:

• The free software SageMath computer algebra system depends on Cython, both for performance and to interface with other libraries.{{cite book |last1=Erocal |first1=Burcin |last2=Stein |first2=William |title=Mathematical Software – ICMS 2010 |chapter=The Sage Project: Unifying Free Mathematical Software to Create a Viable Alternative to Magma, Maple, Mathematica and MATLAB |year=2010 |volume=6327 |pages=12–27 |publisher=Springer Berlin / Heidelberg |doi=10.1007/978-3-642-15582-6_4 |url=http://wstein.org/papers/icms/icms_2010.pdf |series=Lecture Notes in Computer Science |isbn=978-3-642-15581-9 |citeseerx=10.1.1.172.624}}
• Significant parts of the scientific computing libraries SciPy, pandas and scikit-learn are written in Cython.{{cite web |url=http://docs.scipy.org/doc/scipy/reference/release.0.7.2.html |title=SciPy 0.7.2 release notes |access-date=29 May 2011 |archive-date=4 March 2016 |archive-url=https://web.archive.org/web/20160304052117/http://docs.scipy.org/doc/scipy/reference/release.0.7.2.html |url-status=dead}}{{cite journal |last1=Pedregosa |first1=Fabian |last2=Varoquaux |first2=Gaël |last3=Gramfort |first3=Alexandre |last4=Michel |first4=Vincent |last5=Thirion |first5=Bertrand |last6=Grisel |first6=Olivier |last7=Blondel |first7=Mathieu |last8=Prettenhofer |first8=Peter |last9=Weiss |first9=Ron |last10=Dubourg |first10=Vincent |last11=Vanderplas |first11=Jake |last12=Passos |first12=Alexandre |last13=Cournapeau |first13=David |title=Scikit-learn: Machine Learning in Python |journal=Journal of Machine Learning Research |year=2011 |volume=12 |pages=2825–2830 |arxiv=1201.0490|bibcode=2011JMLR...12.2825P }}
• Some high-traffic websites such as Quora use Cython.{{better source needed|date=October 2018}}{{cite web |url=https://www.quora.com/Is-Quora-still-running-on-PyPy/answer/Alex-Yakunin |title=Is Quora still running on PyPy?}}

Cython's domain is not limited to just numerical computing. For example, the {{proper name|lxml}} XML toolkit is written mostly in Cython, and like its predecessor Pyrex, Cython is used to provide Python bindings for many C and C++ libraries such as the messaging library ZeroMQ.{{cite web |url=http://www.zeromq.org/bindings:python |title=ØMQ: Python binding}} Cython can also be used to develop parallel programs for multi-core processor machines; this feature makes use of the OpenMP library.

• PyPy
• Numba

{{reflist|30em}}

• {{Official website}}
• {{GitHub|cython}}

{{Python (programming language)}}

Category:Articles with example Python (programming language) code

Category:Python (programming language)

Category:Python (programming language) implementations

Category:Software using the Apache license

Category:Source-to-source compilers