Talk:NumPy

This article is rated Start-class on Wikipedia's content assessment scale. It is of interest to the following WikiProjects:

Computing Low‑importance This article is within the scope of WikiProject Computing, a collaborative effort to improve the coverage of computers, computing, and information technology on Wikipedia. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks.ComputingWikipedia:WikiProject ComputingTemplate:WikiProject ComputingComputing Low This article has been rated as Low-importance on the project's importance scale.

This article was the subject of a Wiki Education Foundation-supported course assignment, between 14 January 2021 and 24 April 2021. Further details are available on the course page. Student editor(s): Ammarh1234. Peer reviewers: Pheiruse.

Above undated message substituted from Template:Dashboard.wikiedu.org assignment by PrimeBOT (talk) 05:29, 17 January 2022 (UTC)Reply

The discussion/comparison to Matlab (especially discussion of Simulink) seems misplaced and sort of bizarre frankly. Maybe a holdover from days of yore? What is it doing in an article on numpy? Maybe it should be in a separate article comparing Matlab and Python?

Would some knowledgeable person please explain in the article the version of Python for which NumPy works? As far as I can discover NumPy is for the tenaciously non-obsolete Python version 2.7 and not for the shiny and spanking new Python 3.x. Help please!! — Preceding unsigned comment added by Hawden (talk • contribs) 14:16, 8 March 2014 (UTC)Reply

Numpy works for Python 3.x, Python 2 isn't really supported anymore

Since numarray and numeric are deprecated, that info is no longer relevant to most people. I would like to move alot of that info to a new section, e.g. History of Numerical Python. Chrike 04:42, 25 June 2007 (UTC)Reply

Um, which info? The info in the heading section is relevant, as it tells who originally created it. The other info is already in the 'History' section. Please clarify so we can work together to improve the article. peterl (talk) 10:15, 19 October 2011 (UTC)Reply

I suggest we look into renaming this article from Numerical Python to NumPy because nowhere else (except in very old documents) is the term Numerical Python used much. See http://numpy.scipy.org or Travis Oliphant's "Guide to NumPy" Chrike 17:21, 15 October 2007 (UTC)Reply

This page has been moved from "Numerical Python" to NumPy as the result of a move request listed at Wikipedia:Requested moves. Dekimasuよ! 11:56, 22 October 2007 (UTC) Reply

The intro reads "Since Python is an interpreted language, mathematical algorithms often run much slower than they do in compiled languages like C or even Java."

Isn't Python compiled? (pyc files) See http://www.network-theory.co.uk/docs/pytut/CompiledPythonfiles.html 202.78.240.7 (talk) 04:43, 11 June 2008 (UTC)Reply

"A program doesn't run any faster when it is read from a .pyc or .pyo file than when it is read from a .py file; the only thing that's faster about .pyc or .pyo files is the speed with which they are loaded." Python Tutorial Hawryluka (talk) 14:54, 22 July 2008 (UTC)Reply

Python source code (.py files) is compiled into byte code (.pyc or .pyo files) the first time it is successfully imported. If possible the compiled byte code file is saved for future use but this is just an optimisation. The byte code itself is interpreted, so Python is regarded an interpreted language.

The article used to justify that the numpy implementation is about the same as an equivalent ("nothing fancy") C++ program does not, in fact, support the claim. According to the article, the numpy based implementation was more than 10 times slower than the naive C++ version -- to me that's very far from being close to the same speed. A naive C++ implementation of a vector or matrix library will also be far from optimal; considerable effort has to be made to make full use of a CPUs theoretical performance -- hence BLAS and similar libraries exist. A naive matrix-matrix multiplication, for example, will not use more than 20 % (at best!) of a processors potential.

Due to this exaggeration I'm removing the claim. — Preceding unsigned comment added by 212.130.79.38 (talk) 11:23, 6 May 2016 (UTC)Reply

I could not find this information. The closest was this thread in the mail-list https://groups.google.com/forum/#!topic/comp.lang.python/Ej8x3hP3wlc%5B1-25%5D

Nicoguaro (talk) 16:11, 2 February 2017 (UTC)Reply

Added. peterl (talk) 21:26, 5 February 2017 (UTC)Reply

In the page is indicated a BSD license, but on NumPy site (https://numpy.org/license.html) BSD is not explicitly mentioned (the text is almost the same of the BSD3).

Paolo Gibellini (talk) 11:22, 31 January 2020 (UTC)Reply

Thanks for pointing that out. BSD is mentioned on the main NumPy page, toward the top of the page. I have added a citation. --{{u|Mark viking}} {Talk} 11:41, 31 January 2020 (UTC)Reply

Where it mentions BSD, it links to this: https://github.com/numpy/numpy/blob/main/LICENSE.txt, which indicates that it's BSD-3. 70.50.71.2 (talk) 17:41, 8 June 2023 (UTC)Reply

The article does not explain what basic types are supported as the elements of the array. Is there an expert able to add this information ? -- Juergen 217.61.203.35 (talk) 09:37, 23 August 2020 (UTC)Reply

It can be anything.

https://numpy.org/devdocs/user/basics.types.html and https://numpy.org/devdocs/user/basics.rec.html#structured-arrays

peterl (talk) 17:37, 23 August 2020 (UTC)Reply

https://doi.org/10.1038/s41586-020-2649-2 It's open access, thus accessible to everyone. I added it to the further reading list, but I think we could use the paper to improve the article:

• check if we can exchange some existing references by the paper, after all there is nothing more solid than a open access Nature publication
• maybe add some of the included images, same as the paper they are CC-BY-4.0 licensed
• extend the text: The paper emphasizes NumPy being at the foundation of the scientific python ecosystem and that it's role of providing an API that is also used by other scientific computing packages. The importance of NumPy now extends beyond the package itself. — Preceding unsigned comment added by Elimik31 (talk • contribs) 01:29, 15 October 2020 (UTC)Reply

I now implemented the last point that I suggested, using the paper as a reference and mentioning python's role for providing an array API for other array implementation in the limitations section, since numpy's limitations are what those array packages usually aim at solving. Elimik31 (talk) 23:05, 15 October 2020 (UTC)Reply

Gui 45.251.232.185 (talk) 23:31, 19 June 2022 (UTC)Reply

Could someone fix the release version reference to Wikidata? I gave it a go but no luck. The latest release version should show as v1.23.5. Wikidata does have this value marked as the preferred rank, but I am not sure how to make this page refer to it. Thanks!

--Toastpaint (talk) 02:01, 28 November 2022 (UTC)Reply

For transparency, I am currently employed by MathWorks, who produce MATLAB. I'm writing in my own words and don't represent the views of MathWorks.

These statements (in the "Features" section) seem to be opinion-based and ill-defined, and don't seem to be reflected in the source cited. If these are quotes from a notable source comparing the languages, can they be put in quotation marks and cited properly? Otherwise I don't think they belong in an encyclopedia. 195.99.123.202 (talk) 14:00, 30 October 2024 (UTC)Reply