Talk:List of random number generators

Notes

Random number servers/download should be separated from product-selling companies on the page — Preceding unsigned comment added by Keith4random (talk • contribs) 13:36, 15 March 2012 (UTC)

I have removed "TheBestMoms.com Number Generator" from the list of random number servers. The other listed sites attempt to generate "true" random numbers, in quantity, and have APIs for automated access. This page on the other hand gives only a single number - embedded in the web-page, created via a javascript call to Math.random. It's not really an appropriate link for this section.Loris (talk) 10:34, 11 September 2012 (UTC)

PCG

Just wanted to mention PCG (http://www.pcg-random.org/) in case it should be included into the list. From PCG home page: "The name for the family, PCG, stands for permuted congruential generator, combining both concepts that underly the generation scheme, namely permutation functions on tuples and a base linear congruential generator." — Preceding unsigned comment added by 62.78.207.123 (talk) 09:58, 3 December 2015 (UTC)

OneRNG

I have no way of proving this, and maybe some other users can share their experience, but I suspect this device as either a complete fraud, or a crappy device embedded with malicious hardware. It registers itself initially as a modem, and the creators say its because of old manufacturer ID assignments or something similar. I believe it creates a serial over IP PPP connection and its possible it just rootkits the machine its installed in. If other users have evidence of being hacked, or ransom-wared after purchasing this device, it should be removed from Wiki. The creators of this device are now marketing a "physically secure computer" that I fear could be a "network security nightmare" and I'm only trying to save others the same headache this has caused me. Chris.horning (talk) 04:36, 12 November 2016 (UTC)

Main PRNG table

I personally think the main PRNG table should be a self-contained and relatively complete listing of PRNGs commonly encountered in industry, whether based on cryptographic primitives (ChaCha20) or otherwise.

The distinction between crypto and non-crypto is moot in many common applications, so long as it's fast and strong and not unduly encumbered.

I also think it helps to have at least one note per entry giving at least a single attribute of distinction, so we don't have to load 100% of the meaning on a randomly chosen name. — MaxEnt 00:00, 27 April 2018 (UTC)

Grab data from "Comparison of hardware random number generators"

Likely worthwhile to grab the contents of the former "Comparison of hardware random number generators" article and add it to the "Hardware (true) random number generators (TRNGs)" section. Pretty heavy overlap in the data with what that section has.

OTOH the deletionists may come for this article too. 74.104.188.4 (talk) 22:51, 14 September 2018 (UTC)

ACORN

The Additive Congruential Random Number (ACORN (PRNG)) family of generators is not trivial and should be included in this list. — Preceding unsigned comment added by Jwikip (talk • contribs) 23:58, 4 February 2019 (UTC)

I did include ACORN some time back, and further added to the notes column in the table;
I have just modified the notes as ACORN can be extended to arbitrary period length [ref to be provided in the ACORN page] jw (talk) 08:43, 3 October 2019 (UTC)

JSF is older than 2009?

I just tried to find out when Bob Jenkins' small PRNG was actually created. The table in the article says Date 2009 but for sure the algorithm must be older because archive.org has a copy from 2007, cf. http://web.archive.org/web/20071028200612/https://burtleburtle.net/bob/rand/smallprng.html Does anybody know about when the algorithm was proposed or can find out somehow? Linux-Fan 2003:C3:F704:F518:921B:EFF:FE55:6D5D (talk) 22:03, 21 May 2019 (UTC)

I emailed Jenkins and he replied that the earliest reference he could find to the present (final) version of his small PRNG is from August 2 2007. I've updated the date in the table. L0rents (talk)

Flawed implementation in Excel

• This article https://webpages.uidaho.edu/~brian/McCullough_2008.pdf discusses the improper implementation of the Wichmann–Hill generator in Excel. PhilAtWiki (talk) 11:13, 16 September 2020 (UTC)
• This article https://jamesmccaffrey.wordpress.com/2016/05/14/the-wichmann-hill-random-number-algorithm/ puts forth a possible reason for the error. PhilAtWiki (talk) 11:40, 16 September 2020 (UTC)

Romu

Is there any reliable information on the Romu family? The website is https://romu-random.org/ and the author claims that the generators pass severe tests for randomness and are fast, generally at least as good as the PCG family.

Just after I posted this I found https://rhet.dev/wheel/rng-battle-royale-47-prngs-9-consoles/. The simplest Romu generator came out on top, with the caveat that very occasionally (according to the original creator) a seed may produce a short cycle period. GammaTwo (talk) 23:59, 28 January 2021 (UTC)

• I have suspicions about it. It doesn't mix its internal state anywhere near enough for the author's argument based on the cyclicity of random mappings to even heuristically apply. Given that the "standard model" for Romu (RomuTrio) has a 192-bit state, if it does pass statistical tests, it may be because of the phenomenon that the PCG author christened the "too big to fail" problem. 175.140.63.108 (talk) 09:57, 10 January 2022 (UTC)

Thoughts regarding WELLDOC

I had never heard of the WELLDOC technique before, and I cannot see any evidence of its use. Also, looking at the paper, their results frankly don't look very good: by mixing two generators together with WELLDOC words (i.e. doubling the state size), it seems like they were able to increase the measurable quality of the LCGs a little bit. Irene Twoth (talk) 02:19, 2 January 2023 (UTC)

Date table not properly sorted

When sorting by date, ACORN shows as the newest, likely do to the date not being interpereted as a straightforward integer 82.13.112.23 (talk) 18:01, 8 November 2023 (UTC)

Thank you for pointing out this difficulty ...
As the person who originally included the information "(discovered in 1984)", I have resolved the issue, without losing that information. jw (talk) 20:03, 8 November 2023 (UTC)