Talk:Fermat number/Archive 1

Old comment
It's worth noticing something. If it makes sense to talk of "probabilities" in this area, then the odds are good that Fermat's little theorem will "make" these numbers prime. That's because there is so much scope for any of the testing sequences to come up with 1 at an intermediate stage, in which case they will stick there. 1 is an attractor for the sequences. PML. — Preceding unsigned comment added by 203.202.5.75 (talk) 23:13, 23 January 2003 (UTC)

2 Squared N?
Is there a way to superscript the n above the 2 in the expression of the function? Right now it doesn't look like "Two to the two-to-the-Nth power" but "Two squared times N." That confused me greatly at first reading. Nevah 03:28, 1 May 2006 (UTC)

Regarding the phrase...
"The following heuristic argument suggests there are only finitely many Fermat primes: according to the prime number theorem, the "probability" that a number n is prime is at most A/ln(n), where A is a fixed constant. Therefore, the total expected number of Fermat primes is at most


 * $$A \sum_{n=0}^{\infty} \frac{1}{\ln F_{n}} = \frac{A}{\ln 2} \sum_{n=0}^{\infty} \frac{1}{\log_{2}(2^{2^{n}}+1)} < \frac{A}{\ln 2} \sum_{n=0}^{\infty} 2^{-n} = \frac{2A}{\ln 2}$$

It should be stressed that this argument is in no way a rigorous proof. For one thing, the argument assumes that Fermat numbers behave "randomly", yet we have already seen that the factors of Fermat numbers have special properties. Although it is widely believed that there are only finitely many Fermat primes, there are some experts who disagree. "

Name a kind of prime that the same argument suggests is finite, but that in fact is known to be infinite. Georgia guy 20:29, 7 October 2006 (UTC)

Reciprocal sum
yo, the property "the sum of two fermat number recipricals is irrational" is clearly wrong, a rational + a rational is rational
 * That quote is false. The article actually says "The sum of the reciprocals of the Fermat numbers is irrational". It's the sum of all reciprocals. A sum of infinitely many rationals can be irrational. PrimeHunter 00:43, 23 December 2006 (UTC)

F33 status
2^2^33 + 1 is known to have no prime factors below... Georgia guy 23:34, 24 February 2007 (UTC)

Luca prime
A prime divisor of Fermat numbers is called Luca prime. The first two Luca prime is 3,5. It is difficult to know whether the given prime is Luca prime. It is not even known if 7 is a Luca prime. We do know that the sum of the the reciprocals of all Luca primes converges.218.133.184.93 16:27, 22 March 2007 (UTC)
 * The answer is that 7 is definitely not. There are certain primes that no number of the form 2^n+1 is divisible by, and 7 is such a prime. The sequence:

2, 3, 5, 9, 17, 33, 65

Is congruent to 2, 3, 5, 2, 3, 5, 2 mod 7. Georgia guy 16:57, 22 March 2007 (UTC)


 * I found no mention of "Luca prime" with Google. If the term has been used then it doesn't appear notable enough to mention in Wikipedia. PrimeHunter 01:16, 23 March 2007 (UTC)

Is there any algorithm to know whether the given prime is Luca prime?218.133.184.93 08:10, 23 March 2007 (UTC)


 * See http://primes.utm.edu/glossary/page.php?sort=FermatDivisor. Your alleged term "Luca prime" is usually called a Fermat divisor. PrimeHunter 15:28, 23 March 2007 (UTC)


 * The problem is that 3 and 5 are excluded from Fermat divisors. — Preceding unsigned comment added by 218.133.184.93 (talk) 07:16, 24 March 2007 (UTC)

Factually Incorrect
From the article:

If 2n + 1 is prime, it can be shown that n must be a power of 2. (If n = ab where 1 < a, b < n and b is odd, then 2n + 1 ≡ (2a)b + 1 ≡ (&minus;1)b + 1 ≡ 0 (mod 2a + 1).) In other words, every prime of the form 2n + 1 is a Fermat number, and such primes are called Fermat primes. The only known Fermat primes are F0,...,F4.

I don't know much about Fermat numbers but I do know that 2 is a prime number of the form 2n+1 where n is 0. Zero is not a power of two so the first part cannot be correct and the "every prime of the form 2n+1 is a fermat prime" is also incorrect since 2 is of the form 2n+1 and not a fermat prime. I'm gonna try to fix this, but somebody who knows what they're doing really should. --Shadowoftime 18:53, 16 July 2006 (UTC)


 * Indeed, the statement only holds for nonzero n. -- EJ 19:06, 16 July 2006 (UTC)

I agree with Shadowoftime. I believe that 2 should be a Fermat prime because it is a prime of the form 2n+1. If a prime is of this form, then it is a Fermat prime. --PhiEaglesfan712 14:57, 17 June 2007 (UTC)


 * Wikipedia content should have reliable sources, and the reliable sources I have seen either define a Fermat prime as a prime of form 2^(2^n)+1 with integer n>=0, or the provably equivalent definition 2^n+1 with integer n>=1. PrimeHunter 21:44, 17 June 2007 (UTC)

Fermat numbers in hexadecimal representation
Hi, I added the following table of the first nine Fermat numbers in hexadecimal representation, which was now deleted as "redundant". I think, it is more easy to understand these numbers seeing them hexadecimal, because 1.0001hex and 1.0000.0001hex are much more concise than their decimal equivalents. Due to the fact that a normal brain does not see, that 4.294.967.296dec ist simply 1.0000.0001hex I dont concider the second table as redundant. Hexadecimal is the most appropriate representation for Fermat numbers, and the first table is only necessary for people who can´t read them. What are your thoughts? --Tilman Piesk 13:22, 11 October 2007 (UTC)


 * I don't think there is any reason to show hexadecimal numbers. They are rarely used in mathematics outside computer science and people who are familiar with them are likely to understand the formula 2n+1 well without listing it in hex. By the way, all your exponents from F4 are wrong. PrimeHunter 22:35, 11 October 2007 (UTC)


 * I tend to agree that there is not much of an advantage, since from 17 on,they are all of the form 16^m + 1, so they all start with a 1, end with a 1, and have an easily calculated number of zeroes in between. So there does not seem to be a great need to use space to write them down explicitly. However, hexadecimal numbers do have other mathematical interest, see for example Bailey-Borwein-Plouffe formula on the hexadecimal expansion of piMessagetolove 13:56, 12 October 2007 (UTC)

"By the way, all your exponents from F4 are wrong." This very funny hint shows, that hexadecimal numbers are indeed very strange to a considerable part of the population... 10 = seize, 20 = deux-seize, 40 = quatre-seize, 80=huit-seize, 100=nilā --Tilman Piesk 13:31, 17 October 2007 (UTC)
 * OK, I overlooked that you used hexadecimal exponents. I have never seen anybody do that before, and I often see hex numbers. If it's done (which I don't think it should be) then it should be said clearly that exponents are also hex. PrimeHunter 14:06, 17 October 2007 (UTC)

F34 status
2^2^34 + 1 is known to have no prime factors below 2^69 —Preceding unsigned comment added by PhiEaglesfan712 (talk • contribs) ("below 2^69" was added after Georgia guy's below comment)

You mean, it is known to be prime?? Where is your source?? Prove it. Georgia guy 14:57, 17 June 2007 (UTC)

See for details. —Preceding unsigned comment added by PhiEaglesfan712 (talk • contribs)
 * No, that doesn't say it is known to be prime; it says it has unknown character; if it is prime; I strongly doubt it will be proven before 2030. Georgia guy 18:53, 17 June 2007 (UTC)

However, there is a huge chasm between 2^69 and 2^131072 (the square root of 2^2^34)... If 2^2^34 + 1 was prime... not only do I win $50,000 for the first to find a 10 million digit prime... I would win $400,000 more for finding the first 100 million digit prime and 1 billion digit prime. The number 2^2^34 + 1 has nearly 5.5 billion digits! —Preceding unsigned comment added by PhiEaglesfan712 (talk • contribs)
 * No, 2^131072 is the square root of 2^262144. The square root of 2^2^34 is 2^(2^33). Georgia guy 20:57, 17 June 2007 (UTC)

If there is a reliable source to known trial factor limits for the smallest Fermat numbers with unknown status, then that may be worth mentioning in the article. The link from PhiEaglesfan712 does not mention a limit of 2^69. PrimeHunter 21:54, 17 June 2007 (UTC)

You are correct. The square root of 2^2^34 is 2^ (2 ^ 33)... I realized the mathematical error after I signed off earlier today... However, my computer tried to factor 2^2^34 to 2^69, and it said that "F34 has no factor to 2^69".


 * (a) 2^2^34 is already factored (it has 2^34 factors) ; (b) probably someone before you has already tried to factor F34 to more than 2^69 bits...&mdash; MFH:Talk 07:36, 20 June 2008 (UTC)

Applications of Fermat numbers
"The most common method used is to take any seed value between 1 and P-1, where P is a Fermat prime. Now multiply this by a number A, which is greater than the square root of P and relatively prime to P." Is the statement that A is relatively prime with respect to P necessary? P was stated to be a prime number, so as long as A is not P, they will be relatively prime. 12.13.126.253 (talk) 13:01, 27 August 2008 (UTC)Tim Forbis


 * You are right. Presumably the intended condition was that A is a primitive root mod P (which guarantees that the generator has a full period P − 1). — Preceding unsigned comment added by EmilJ (talk • contribs) 13:21, 27 August 2008 (UTC)

Apparently looseness in the term "generalized Fermat number"
The current version of this article states that:

Numbers of the form $$a^{2^{ \overset{n} {}}} + b^{2^{ \overset{n} {}}}$$, where a > 1 are called generalized Fermat numbers.

It also defines a subclass of these GFNs where b = 1, and denotes them with the expression Fn(a). (My understanding is that "ordinary" Fermat numbers can be denoted by the expression Fn(2).) However, it seems that much of the subsequent discussion of "generalized Fermat numbers" is actually only true for the untitled subclass of GFNs that can be described by Fn(a), which is a superset of ordinary Fermat numbers. For instance, the statement that:

Generalized Fermat numbers can be prime only for even a, because if a is odd then every generalized Fermat number will be divisible by 2.

is true if b = 1 (or indeed any odd number) but would seem to be false if b is any even number. In other words, this use of "generalized Fermat number" seems to be limiting itself to the set of all numbers Fn(a). I can see that the requirement to avoid Fermat numbers divisible by 2 is actually met by making sure that a+b is odd; i.e., one is odd and the other even, but that only demonstrates an unwritten assumption that we're limiting ourselves to a subclass of GFNs where b is always odd, whereas the given GFN definition makes no assumptions about the parity of either a or b.

Similarly, I also question the other two statements ("…odd prime p…" and "finitely many generalized Fermat primes for each even base") that also seem to my cursory view to be assuming that b = 1. (For the latter, doesn't a GFN have two bases, a and b?)

Could someone with sounder mathematical knowledge than I clean these statements up? Or am I just missing something? ~ Jeff Q (talk) 03:59, 22 February 2009 (UTC)

Invalid Information
The page currently notes that only up to the 11th number has been completely factorized. I think that should really not be put into the page because of how easily it can be made invalid. Using WolframAlpha, you can easily completely factor the 12th number in the sequence by typing in 2^2^12+1 and clicking more digits until there are no more. The final digit count is 1,234 digits. --12.52.229.65 (talk) 12:58, 13 June 2009 (UTC)
 * Uh, no. WolframAlpha cannot factor 22 12 +1; it can only display it. When F12 is factored, it will be news and the article will get updated. Shreevatsa (talk) 13:21, 13 June 2009 (UTC)


 * "completely factored" means that all prime factors are known. This is unrelated to all digits in the decimal expansion being known. The former is often very hard for large numbers. The latter is trivial. It follows from http://www.prothsearch.com/fermat.html#Incomplete that
 * 2^(2^12)+1 = (7*2^14+1) * (397*2^16+1) * (973*2^16+1) * (11613415*2^14+1) * (76668221077*2^14+1) * c,
 * where the 5 factors in parantheses are primes, but c is a composite number with no known prime factors. c has 1187 digits and 2^(2^12)+1 has 1234 digits, so only a small part of the prime factorization is known. PrimeHunter (talk) 20:38, 13 June 2009 (UTC)
 * What is the smallest prime factor of this C1187 known to be no less than?? Georgia guy (talk) 21:45, 13 June 2009 (UTC)
 * The proven lower limit appears to be around 20 digits but the smallest factor is probably much larger than the proven limit. I think trial division limited to prime factors of a possible form for Fermat numbers is the fastest exhaustive method to guarantee discovery of all prime factors below a given limit in an incompletely factored number. But in practice relatively little time is used on trial division. Other algorithms like ECM can find much larger factors and give high probability but no certainty that no small factor has been missed. I don't know the precise factorization effort on the C1187 but there is probably no factor below 50 digits. PrimeHunter (talk) 01:36, 14 June 2009 (UTC)

F9 contains P99
But P99 is either What does it mean in the context of a number factor ? Help me, please explain or link to the relevant page.--methodood (talk) 17:43, 26 November 2009 (UTC)
 * Embraer P99, a variant of the Embraer R99 (an aircraft)
 * Walther P99 (A Pistol)
 * The notation is apparently from . P means it is prime, while 99 seems to be a meaningless id number. They claim it has 96 digits; if you are really interested in its value, you can divide F[9] by the other two factors to get it. — Emil J. 17:53, 26 November 2009 (UTC)
 * And the result is 741640062627530801524787141901937474059940781097519023905821316144415759504705008092818711693940737. Which actually has 99 digits, so that's what the 99 in P99 presumably means. — Emil J. 17:58, 26 November 2009 (UTC)

Thank you EmilJ for updating the page. Whoever put that notation without reference is forgiven, and a link to prothsearch would have been shorter (or shortable using tinyurl, if WP policies accept that.) I think that the notability for P99 is inverse to its number of digits, else it had been well placed in the disambig. --82.227.17.30 (talk) 10:46, 28 November 2009 (UTC)

Fermat aware of Euler's result
I've added reference Křížek, Luca, Somer 2001, p. 38, Remark 4.15 to the claim that Fermat probably knew this result. However, the claim that it is "widely believed" might be too strong. The mentioned remarks says the following:
 * Remark 4.15. As mentioned in the literature (cf., e.g., [Lenstra, Lenstra, Manasse, Pollard]), Fermat probably knew the previous theorem, although its proof was given later by Euler in 1747 (see [Burton, p.219]). However, it is then unclear why Fermat did not apply this theorem to the factorization of F5. It was sufficient to divide F5 only by primes of the form 64k + 1, and for k = 10 Fermat could have convinced himself that his hypothesis on the primality of Fm did not hold. Concerning the factor 2142.27 + 1 = 274177 of F6 (see Figure 1.2), we note that there are 370 primes of the form k27 + 1 for k < 2142.

--Kompik (talk) 14:06, 27 March 2010 (UTC)

Who thinks Fermat knew Euler's result
I think the reference provided after the statement in question gives sufficient answer to who thinks this. I am going to remove the Who template. Please discuss here if you disagree. Bender2k14 (talk) 20:38, 1 May 2011 (UTC)

Other theorems about Fermat numbers
The first lemma and its cumbersome proof can easily be avoided. The only fact that is used in the subsequent discussion is that a - b divides a^n - b^n. This can be proved very easily as follows. We have a = b modulo (a - b). Therefore also a^n = b^n mod (a - b), which means precisely that a - b divides a^n - b^n.

Stickelberger (talk) 16:15, 18 March 2010 (UTC)
 * I'm not too familiar with modulo arithmetic, so a proof using summation would be useful for me. However going from line 2 to line 3, I can't figure out why $$\sum_{k=0}^{n-1}a^{k+1}b^{n-1-k}$$ = $$=a^n+\sum_{k=1}^{n-1}a^kb^{n-k}$$.  Instead, removing the k = 0 term from the sum gives me $$ab^{n-1}+ab^{-1}\sum_{k=1}^{n-1}a^kb^{n-k}$$.  Could someone show how the first 2 terms of the 3rd line follow from the second line?--Wikimedes (talk) 00:13, 23 September 2012 (UTC)

More to add to the section "Heuristic arguments for density"
That section should give an extra argument at the end showing that the factors having to be of special form has no effect on the probability of a Fermat number being prime. Only the lower bound of prime factors of a Fermat number effects the probability of a Fermat number being prime and not the property of all factors being of a special form. By Fermat's little theorem, all factors of Fn are of the form 2n + 1 + 1 and by Lucas' theorem, for any odd prime p, 2 is a square modulo p if and only if p ≡ 1 or 7 mod 8, so for any n > 1, all factors of Fn are of the form 2n + 2 + 1. Since we know that a Fermat number can never be ≡ 1 mod p for any odd prime p, assuming we don't know anything about the form of the factors of a Fermat number, the probability of p being a factor of a Fermat number is 1/(p - 1). Furthermore, by Lucas' theorem, all factors of F5 are of the form 27 + 1, but 1/26 of the prime numbers are of the form 27 + 1. Also, since 2 is a square mod 641, 2^64 must be a fifth root of unity modulo 641 so there's 1 chance in 5 that 2^64 ≡ 1 mod 641. Given that 2^64 ≡ 1 mod 641, there's 1 chance in 2 that 2^32 ≡ -1 mod 641. Thus in total, there's 1 chance in 10 that 641 is a factor of F5, 26 times higher than the probability one would expect for that number being a factor of F5, making up for the fact that 1/26 of the prime numbers are of the form 27 + 1, so the factors of Fermat numbers having to be of that form makes no difference to the probability of a Fermat number being prime. Blackbombchu (talk) 01:34, 21 November 2013 (UTC)


 * I haven't examined your argument but your post is missing k in the factor form k2n+2 + 1. PrimeHunter (talk) 02:16, 21 November 2013 (UTC)

Ninth Fermat number??
Recently, there was a discussion about the definition of the "ninth Fermat number". Look at this, as an example:


 * Ninth Floor
 * Eighth Floor
 * Seventh Floor
 * Sixth Floor
 * Fifth Floor
 * Fourth Floor
 * Third Floor
 * Second Floor
 * Ground Floor
 * Basement Floor

As you can see in the above list, the "ninth floor" is technically the tenth floor, but it is designated as the ninth. Does Wikipedia have any article about inconsistent definitions of ordinal numbers?? Georgia guy 13:45, 13 July 2007 (UTC)


 * See Zeroth. PrimeHunter 16:25, 13 July 2007 (UTC)


 * So, that means that 22 0 + 1 = 3, is counted as the zeroth Fermat number. PhiEaglesfan712 17:27, 13 July 2007 (UTC)


 * Yes. I have also seen it called "the first" but in contexts where "first" appears to mean "the initial" rather than the beginning of a systematic numbering. When numbering gets to "the ninth Fermat number" I have only seen the name about F9 = 22 9 + 1. PrimeHunter 18:15, 13 July 2007 (UTC)

But keep in mind, that 3 is one Fermat number and 3 and 5 are two. So its nearly unavoidable calling 5 the second. To avoid missunderstandings the first Fermat number should be called Fermat number zero, thats perfect. --Tilman Piesk 13:29, 11 October 2007 (UTC)

The zeroth number is not the first number, by definition. There is no such thing as the zeroth number in any list. There's only the first one and higher. If you only had the zeroth item and no others, technically you'd have an empty set and no list at all. Clearly, the first Fermat number is the first one. It's just coincidental that the generation of the list of numbers relies on a formula that uses "number in list" minus one. If this was a list made from another formula that didn't resemble counting numbers then it would be utterly clear which number was first and second and so on. That would be the kind of thinking you'd want to use here and everywhere that you'd want to label the first number in a series. — Preceding unsigned comment added by 70.193.48.154 (talk) 02:13, 11 June 2015 (UTC)

Moving Generalized Fermat Numbers to a separate article
I think the Generalized Fermat Numbers section is specific enough to be put in a separate article.

Reasons for this:


 * we are talking about a list here, it doesn't make sense to mix two (i.e. the maximum noted in the text box at the top right is for Fermat Numbers, not Generalized Fermat Numbers)
 * it seems that Generalized Fermat Numbers are not a subset of Fermat Numbers, rather a related concept / superset
 * the Generalized Fermat Numbers takes a large section, divided into subsections, which are better served on their own
 * related to this, the Generalized Fermat Numbers takes a large portion on the Fermat Numbers article, that may not be needed by the reader
 * confusion may arise, e.g. it is not clear that a Generalized Fermat Number is not (or is) a Fermat Number
 * different research / links etc — Preceding unsigned comment added by Owlstead (talk • contribs) 14:07, 18 September 2015 (UTC)

Selfridge's Conjecture
I removed the short section "Selfridge's Conjecture". The conjecture was not well explained, it is not mentioned here (probably meaning it is not very important) and the same contents is also in the wiki-page John Selfridge, where I tried to improve it. I suggest we include only (proven) theorems about the Fermat numbers in the article. Herbmuell (talk) 09:18, 14 December 2016 (UTC)

True or false??
True or false: It is likely the F_33 will be prime with current knowledge.


 * Currently, the primality status of F_33 is unknown. See for details. Maxal 05:45, 13 December 2006 (UTC)

F_34 is more likely to be prime than F_33. It has been proven that F_34 has no prime factors below 2^69.
 * But, to prove a number prime, you must verify it has no factors less than or equal to its square root. The above info would have proved it to be prime if it were less than 2^138, but it is not, its base 2 logarithm is just over 2^34. (Note that the base 2 logarithm of 2^138 is 138.) Georgia guy 20:54, 17 June 2007 (UTC)


 * If F(n) is composite, it ought to satisfy the inequality SQRT [F(n)] > 3×2^(n+3). The inequality is violated by the first five F(n). So the first five F(n) are not composite. The inequality is satisfied by the rest of F(n); that is, n ≥ 5. So there are no other Fermat primes. Ana Conda (talk) 17:28, 24 December 2016 (UTC)

How to very easily hunt for more factors of Fermat numbers
According to the prime density, the set of all prime numbers of the form 3 * 2n + 1 probably increases about as rapidly as the set of Fermat numbers and the set of Mersenne primes and the set values of n such that 3 * 2n + 1 is prime probably grows exponentially. For each n > 0, if 3 * 2n + 2 + 1 is prime, then by Lucas' theorem, since 3 * 2n + 2 + 1 ≡ 1 mod 8, 2 is a square mod 3 * 2n + 2 + 1 and by Fermat's little theorem, since 2 is a square mod 3 * 2n + 2 + 1, 23 * 2 n + 1 ≡ 1 mod 3 * 2n + 2 + 1. Furthermore, given that 3 * 2n + 2 + 1 is prime, there's 1 chance in 3 that 22 n + 1 ≡ 1 mod 3 * 2n + 2 + 1 so there's 1 chance in 3 that 3 * 2n + 2 + 1 is a factor of the Fermat number Fm for some m ≤ n. There's 1 chance in 6 that 22 n ≡ -1 mod 3 * 2n + 2 + 1 so there's 1 chance in 6 that 3 * 2n + 2 + 1 is a factor of the Fermat number Fn given that 3 * 2n + 2 + 1 is prime. In order to beat the record of the largest Fermat number known to be composite, all that has to be done is to hunt for prime numbers of the form 3 * 2n + 1. 48 Mersenne primes have been found so it shouldn't be that much harder to find prime numbers of the form 3 * 2n + 1. Prime numbers of that form are really good candidates for factors of Fermat numbers. Maybe prime numbers of that form that turn out to be factors of Fermat numbers can be added to the web page Fermat factoring status beating the record of the largest known composite Fermat number then the record listed in this article can be updated making it a better article. Blackbombchu (talk) 22:31, 22 November 2013 (UTC)


 * This is known and people have already searched primes of form 3×2n+1 for decades to test them as Fermat factors. Some of the largest known primes are of this form and it may take hundreds or thousands of computer years to find more. PrimeGrid has a distributed project to search for them. PrimeHunter (talk) 00:31, 23 November 2013 (UTC)


 * A composite Fermat number will have a prime factor f = [2^(n+2)]*(6a+1)+1 if n is even, or a prime factor f = [2^(n+2)]*(6b+5)+1 if n is odd. When a Fermat number has only two prime factors, the other prime factor is f = [2^(n+2)]*(6c+3)+1. These modified forms help speed up the search of a prime factor by a rate of 6. Ana Conda (talk) 18:08, 24 December 2016 (UTC)

External links modified
Hello fellow Wikipedians,

I have just modified 2 external links on Fermat number. Please take a moment to review my edit. If you have any questions, or need the bot to ignore the links, or the page altogether, please visit this simple FaQ for additional information. I made the following changes:
 * Added archive https://web.archive.org/web/20131224224552/http://primes.utm.edu/links/theory/special_forms/ to http://primes.utm.edu/links/theory/special_forms/
 * Added archive https://web.archive.org/web/20061002105454/http://www.spd.dcu.ie/johnbcos/fermat6.htm to http://www.spd.dcu.ie/johnbcos/fermat6.htm

When you have finished reviewing my changes, you may follow the instructions on the template below to fix any issues with the URLs.

Cheers.— InternetArchiveBot  (Report bug) 23:36, 29 September 2017 (UTC)

A product of Fermat numbers
Why does power-2 Mersenne prime $$M_{2^n} = 2^{2^n}-1$$ have (Fermat numbers) $$F_{i} = 2^{2^i}+1$$ for divisors, i=0..n-1?
 * Like prime factored examples:
 * 21-1 = 1
 * 22-1 = 1 × 3 = 3
 * 24-1 = 1 × 3 × 5 = (22-1)×(22+1)
 * 28-1 = 1 × (3 × 5) × 17 = 15 × 17 = (24-1)×(24+1)
 * 216-1 = 1 × (3 × 5 × 17) × 257 = 255 × 257 = (28-1)×(28+1)
 * 232-1 = 1 × (3 × 5 × 17 × 257) × 65537 = 65535 × 65537 = (216-1)×(216+1)
 * 264-1 = 1 × (3 × 5 × 17 × 257 × 65537) × 4294967297 = 4294967295 × 4294967297 = (232-1)×(232+1)
 * So really just $$2^{2^n}-1 = (2^{2^{n-1}}-1)(2^{2^{n-1}}+1) = (2^{2^{n-2}}-1)(2^{2^{n-2}}+1)(2^{2^{n-1}}+1) = (2^{2^{n-3}}-1)(2^{2^{n-3}}+1)(2^{2^{n-2}}+1)(2^{2^{n-1}}+1) = ... $$
 * And $$F_{n}-2 = M_{2^n} = 2^{2^n}-1=\prod_{i \mathop =0}^{n-1}(2^{2^i}+1) =\prod_{i \mathop =0}^{n-1}F_{i}.$$
 * And interestingly: $$F_{n} = 2 + \prod_{i \mathop =0}^{n-1}F_{i}.$$ And $$M_{2^n} = \prod_{i \mathop =0}^{n-1}(M_{2^i}+2).$$
 * And $$M_{2^n} = M_{2^{n-1}} * F_{n-1}$$
 * This must be written up somewhere, but I don't see anything similar so far. Tom Ruen (talk) 05:39, 26 October 2018 (UTC)

Original proof of $$2^{n+2}k+1$$
The article first attributes Euler for proving that factors of Fermat numbers must be of the form $$2^{n+2}k+1$$, but then the article later states that Édouard Lucas is the one who proved it. I'm sure both of them aren't the source. Which is it? —Preceding unsigned comment added by 72.235.147.240 (talk) 08:43, 8 October 2007 (UTC)

The statements are a little different. Euler proved that any prime factors of a Fermat number $$2^{2^{n}}+1$$ must be of the form $$2^{n+1}k+1$$ for some integer k, and Lucas proved that any prime factor of Fermat number $$2^{2^{n}}+1$$ must be of the form $$k2^{n+2}+1$$ (for some integer k) if n > 1. For example, Euler's result guarantees only that every prime divisor of the Fermat number $$2^{32}+1$$ must be of the form $$64k+1$$, but Lucas's result guarantees that every prime divisor of the Fermat number $$2^{32}+1$$ must be of the form $$128k+1$$. Messagetolove 14:59, 8 October 2007 (UTC)

Oh, I see. Scrolling up and down, I didn't catch the differences.

However, someone has now changed the article and attributes the stronger form to Euler, citing Ribenboim's book as a reference. I am unsure who has priority( I am not a number-theorist myself), but I would find it strange if Lucas proved a theorem already known to Euler, so I would appreciate more clarity from someone who knows the historyMessagetolove (talk) 16:27, 23 June 2009 (UTC)

There is an error - Euler in his article shows that $$2^{2^5}+1$$ is divided by primes in form $$64k+1$$. (In his result general form is $$k2^{n+1}+1$$, not $$k2^{n+2}+1$$) You can find it in original paper at http://math.dartmouth.edu/~euler/docs/originals/E134.pdf page 33. If somebody maintains this article, correct it please. Anino 16:00, 10 July 2009 —Preceding unsigned comment added by 87.244.194.138 (talk)

Thanks to the anonymous editor for pointing out Euler's original text, which answers my June 23 query.Messagetolove (talk) 19:26, 10 July 2009 (UTC)

The powers of 2 in F(12), F(15), F(36), and F(38) seem to deviate from the Euler-Lucas theorem. Can someone reconcile the deviation for me? Thanks. 166.48.109.119 (talk) 16:00, 7 December 2018 (UTC)