Talk:FRACTRAN

factorize fractions

would it be more useful to have the prime factorizes with the programs —Preceding unsigned comment added by 210.7.49.10 (talk) 04:31, 3 November 2010 (UTC)

Conway's formula for primes doesn't match the linked OEIS sequence?

This page immediately gives Conway's formula for primes as this list of fractions:

[17/91, 78/85, 19/51, 23/38, 29/33, 77/29, 95/23, 77/19, 1/17, 11/13, 13/11, 15/14, 15/2, 55/1]

The text right below then says that this outputs sequence A007542 in OEIS as [n = 2] is updated. The OEIS page links to this handy table of values for the sequence, which seems valid for the above FRACTRAN program, until you get to iteration/element 131, where the integer output of the above algorithm no longer matches the sequence. The expected values from the table do start coming back within a few more updates of n, but out of sync by a few iterations, throwing the numbering off between the two (and the pattern repeats, so they get further apart the longer you let it run).

I only noticed the discrepancy because I recently wrote a FRACTRAN interpreter, and I kept noticing deviations from the OEIS sequence while testing with Conway's primes formula; I initially thought it was a bug in my code, until I saw that the fraction list visible on the OEIS page diverged slightly from the one on Wikipedia, the second- and third-last fractions being different:

[17/91, 78/85, 19/51, 23/38, 29/33, 77/29, 95/23, 77/19, 1/17, 11/13, 13/11, 15/2, 1/7, 55]

Plugging this new fraction list into my program, it output the sequence exactly as given in the linked table.

Anyway, I don't feel comfortable editing the article myself, because I'm not really sure how it should be handled. I'n not that well-versed in FRACTRAN or Conway's work, nor the OESI database or even integer sequences in general. I'm guessing that the two different fraction lists are equivalent for the purposes of computing primes in FRACTRAN (I'm too lazy to bother factoring any of the output), but, unless I'm really off my rocker here, they don't output the same integer sequences, and so the article probably shouldn't imply that they do. 108.181.244.28 (talk) 20:16, 21 February 2013 (UTC)

==

You are right. The given fraction list was missing the important 1/7. And a lot of mathematic papers is also missing 1/7. Maybe because of this article. Anyway. I have fixed the problem. TraxPlayer (talk) 19:59, 30 September 2015 (UTC)

I have reverted to the sequence of 14 fractions that is given in Conway & Guy The Book of Numbers and in Havil Nonplussed. We can't just change stuff because we think it is wrong - we have to stick to cited sources. Gandalf61 (talk) 15:53, 2 October 2015 (UTC)

Note that the sequence given in Project Euler Problem 308 is different from the Conway & Guy and Havil version because it ends in 13/11, 15/2, 1/7, 55/1 instead of 13/11, 15/14, 15/2, 55/1. It probably achieves the same result, but the article should stick to cited sources, and published sources are more reliable than a web page.

I have added the missing 1/7 back again. My source is page 116 in the book: Genius At Play by Siobhan Roberts. Mrs. Roberts had full access to Conway's archive and on page 116 is a copy of his original hand-writing of the fractran program which includes the 1/7 term. TraxPlayer (talk) 09:20, 4 October 2015 (UTC)

I have restored the original sequence, but kept your alternative sequence in a footnote. The version that was actually published by Conway must take precedence over alternative versions in other sources. Gandalf61 (talk) 11:57, 4 October 2015 (UTC)

Sorry but you are not correct. The 1983 version was *not* published by Conway.

I can see that you have been removing the 1/7 term many terms during the last years but the 1983 published version is not the original version published by Conway. Please see also [1]. Most of the Conway & Guy book from 1983 was solely written by Guy including page 26 which contains the wrong fractions. See [2] that Guy is the solely author. My pictures shows Conway's own hand-written. Mrs. Roberts had full access to Conway's archive. So that is the primare source and not the page written by Guy. I will try to locate Conway's original paper from 1973 if that is really needed to convince you. TraxPlayer (talk) 19:04, 4 October 2015 (UTC)

Here is the first page from the original first published version which clearly shows the 1/7 term.

— Preceding unsigned comment added by TraxPlayer (talk • contribs) 22:30, 4 October 2015 (UTC) 

Third Opinion

A third opinion has been requested. I am not sure what the question is. It does appear that different sources provide different sequences. You might try the reliable source noticeboard, but they might find the question to be as esoteric as the language. Robert McClenon (talk) 20:49, 19 October 2015 (UTC)

Both sequences of fractions produce prime powers of 2 (although different numbers in between), so they are equivalent for that matter, as the article rightly says now. Both versions are supported by sources; however, the version with 1/7 is from Conway's paper about FRACTRAN, while The Book of Numbers doesn't even mention FRACTRAN in general, so I'd say that the version with 1/7 should be considered as primary. The OEIS entry is based on the variant with 1/7, not the variant with 15/14; not a single source says otherwise. So the statement from the current version of the article saying that the 15/14 variant produces A007542 is not supported by sources and is simply incorrect. --colt_browning (talk) 12:31, 22 April 2019 (UTC)

I have (hopefully) solved the above problem by explicitly including both variants in the article, citing them, and identifying their corresponding OEIS sequences. JudgeDeadd (talk) 23:42, 9 January 2022 (UTC)

Bibliography

The right reference to Conway's original work is

J. H. Conway. Fractran: A Simple Universal Programming Language for Arithmetic. In: Open Problems in Communication and Computation, pages 4–26. Springer, 1987.

The one listed (in Lagarias) is probably a reprint. — Preceding unsigned comment added by 2001:980:D78E:1:E500:1010:1A88:DFC6 (talk) 11:37, 18 January 2014 (UTC)

Proposing to add a link to a pure functional scala interpreter to the external links

• Specific text to be added or removed:A pure functional Fractran interpreter written in Scala
• Reason for the change: there isn't any pure functional example linked. Disclaimer, I'm the author of the software, it's a spare time project so there's no interest conflicting.
• References supporting change: https://en.wikipedia.org/wiki/FRACTRAN#External_links

Angleto (talk) 19:12, 14 November 2021 (UTC)

 Not done This request does not outline specific text to be added to the article. Please outline text you would like added in a "Change X to Y" or "Add X to Section Z" format. Z1720 (talk) 20:03, 25 November 2021 (UTC)