Prime Numbers

Prime Numbers Prime Numbers

thales.doa.fmph.uniba.sk
from thales.doa.fmph.uniba.sk More from this publisher
10.12.2012 Views

24 Chapter 1 PRIMES! 9.5.18 and Algorithm 9.5.19). This method has at least doubled the search efficiency over previous methods. It should be mentioned that modern Mersenne searching is sometimes of the “hit or miss” variety; that is, random prime exponents q are used to check accordingly random candidates 2 q − 1. (In fact, some Mersenne primes were indeed found out of order, as indicated above). But much systematic testing has also occurred. As of this writing, exponents q have been checked for all q ≤ 12830000. Many of these exponents are recognized as giving composite Mersennes because a prime factor is detected. For example, if q is a prime that is 3 (mod 4), and p =2q + 1 is prime, then p|Mq. (See also Exercises 1.47, 1.81.) For the remaining values of q, the Lucas–Lehmer test (see Section 4.2.1) was used. In fact, for all q ≤ 9040000 for which a factor of Mq was not found, the Lucas–Lehmer test was carried out twice (see [Woltman 2000], which website is frequently updated). As mentioned in Section 1.1.2, the prime M25964951 is the current record holder as not only the largest known Mersenne prime, but also the largest explicit number that has ever been proved prime. With few exceptions, the record for largest proved prime in the modern era has always been a Mersenne prime. One of the exceptions occurred in 1989, when the “Amdahl Six” found the prime [Caldwell 1999] 391581 · 2 216193 − 1, which is larger than 2 216091 − 1, the record Mersenne prime of that time. However, this is not the largest known explicit non-Mersenne prime, for Young found, in 1997, the prime 5·2 240937 +1, and in 2001, Cosgrave found the prime 3 · 2 916773 +1. Actually, the 5th largest known explicit prime is the non-Mersenne 5359 · 2 5054502 +1, found by R. Sundquist in 2003. Mersenne primes figure uniquely in the ancient subject of perfect numbers. A perfect number is a positive integer equal to the sum of its divisors other than itself. For example, 6 = 1 + 2 + 3 and 28 = 1 + 2 + 4 + 7 + 14 are perfect numbers. An equivalent way to define “perfection” is to denote by σ(n) the sum of the positive divisors of n, whence n is perfect if and only if σ(n) =2n. The following famous theorem completely characterizes the even perfect numbers. Theorem 1.3.3 (Euclid–Euler). An even number n is perfect if and only if it is of the form n =2 q−1 Mq, where Mq =2 q − 1 is prime.

1.3 Primes of special form 25 Proof. Suppose n =2 a m is an even number, where m is the largest odd divisor of n. The divisors of n are of the form 2 j d, where 0 ≤ j ≤ a and d|m. Let D be the sum of the divisors of m excluding m, and let M =2 a+1 − 1=2 0 +2 1 + ···+2 a . Thus, the sum of all such divisors of n is M(D + m). If M is prime and M = m, thenD = 1, and the sum of all the divisors of n is M(1 + m) =2n, sothatn is perfect. This proves the first half of the assertion. For the second, assume that n =2 a m is perfect. Then M(D + m) =2n =2 a+1 m =(M +1)m. Subtracting Mm from this equation, we see that m = MD. If D>1, then D and 1 are distinct divisors of m less than m, contradicting the definition of D. SoD =1,m is therefore prime, and m = M =2 a+1 − 1. ✷ The first half of this theorem was proved by Euclid, while the second half was proved some two millennia later by Euler. It is evident that every newly discovered Mersenne prime immediately generates a new (even) perfect number. On the other hand, it is still not known whether there are any odd perfect numbers, the conventional belief being that none exist. Much of the research in this area is manifestly computational: It is known that if an odd perfect number n exists, then n>10 300 , a result in [Brent et al. 1993], and that n has at least eight distinct prime factors, an independent result of E. Chein and P. Hagis; see [Ribenboim 1996]. For more on perfect numbers, see Exercise 1.30. There are many interesting open problems concerning Mersenne primes. We do not know whether there are infinitely many such primes. We do not even know whether infinitely many Mersenne numbers Mq with q prime are composite. However, the latter assertion follows from the prime k-tuples Conjecture 1.2.1. Indeed, it is easy to see that if q ≡ 3 (mod 4) is prime and 2q + 1 is also prime, then 2q + 1 divides Mq. For example, 23 divides M11. Conjecture 1.2.1 implies that there are infinitely many such primes q. Various interesting conjectures have been made in regard to Mersenne numbers, for example the “new Mersenne conjecture” of P. Bateman, J. Selfridge, and S. Wagstaff, Jr. This stems from Mersenne’s original assertion in 1644 that the exponents q for which 2 q −1 is prime and 29 ≤ q ≤ 257 are 31, 67, 127, and 257. (The smaller exponents were known at that time, and it was also known that 2 37 −1 is composite.) Considering that the numerical evidence below 29 was that every prime except 11 and 23 works, it is rather amazing that Mersenne would assert such a sparse sequence for the exponents. He was right on the sparsity, and on the exponents 31 and 127, but he missed 61, 89, and 107. With just five mistakes, no one really knows how Mersenne effected such a claim. However, it was noticed that the odd Mersenne exponents below 29 are all either 1 away from a power of 2, or 3 away from a power of 4 (while the two missing primes, 11 and 23, do not have this property), and Mersenne’s list just continues this pattern (perhaps with 61 being an “experimental error,” since Mersenne left it out). In [Bateman et al. 1989] the authors suggest a new

24 Chapter 1 PRIMES!<br />

9.5.18 and Algorithm 9.5.19). This method has at least doubled the search<br />

efficiency over previous methods.<br />

It should be mentioned that modern Mersenne searching is sometimes of<br />

the “hit or miss” variety; that is, random prime exponents q are used to check<br />

accordingly random candidates 2 q − 1. (In fact, some Mersenne primes were<br />

indeed found out of order, as indicated above). But much systematic testing<br />

has also occurred. As of this writing, exponents q have been checked for all<br />

q ≤ 12830000. Many of these exponents are recognized as giving composite<br />

Mersennes because a prime factor is detected. For example, if q is a prime<br />

that is 3 (mod 4), and p =2q + 1 is prime, then p|Mq. (See also Exercises<br />

1.47, 1.81.) For the remaining values of q, the Lucas–Lehmer test (see Section<br />

4.2.1) was used. In fact, for all q ≤ 9040000 for which a factor of Mq was<br />

not found, the Lucas–Lehmer test was carried out twice (see [Woltman 2000],<br />

which website is frequently updated).<br />

As mentioned in Section 1.1.2, the prime M25964951 is the current record<br />

holder as not only the largest known Mersenne prime, but also the largest<br />

explicit number that has ever been proved prime. With few exceptions, the<br />

record for largest proved prime in the modern era has always been a Mersenne<br />

prime. One of the exceptions occurred in 1989, when the “Amdahl Six” found<br />

the prime [Caldwell 1999]<br />

391581 · 2 216193 − 1,<br />

which is larger than 2 216091 − 1, the record Mersenne prime of that time.<br />

However, this is not the largest known explicit non-Mersenne prime, for Young<br />

found, in 1997, the prime 5·2 240937 +1, and in 2001, Cosgrave found the prime<br />

3 · 2 916773 +1.<br />

Actually, the 5th largest known explicit prime is the non-Mersenne<br />

5359 · 2 5054502 +1,<br />

found by R. Sundquist in 2003.<br />

Mersenne primes figure uniquely in the ancient subject of perfect numbers.<br />

A perfect number is a positive integer equal to the sum of its divisors other<br />

than itself. For example, 6 = 1 + 2 + 3 and 28 = 1 + 2 + 4 + 7 + 14 are<br />

perfect numbers. An equivalent way to define “perfection” is to denote by<br />

σ(n) the sum of the positive divisors of n, whence n is perfect if and only if<br />

σ(n) =2n. The following famous theorem completely characterizes the even<br />

perfect numbers.<br />

Theorem 1.3.3 (Euclid–Euler). An even number n is perfect if and only if<br />

it is of the form<br />

n =2 q−1 Mq,<br />

where Mq =2 q − 1 is prime.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!