![]() |
|
|
#1 |
|
Jun 2019
1000102 Posts |
a nice remark
Mersenne Number 2n-1 x2 = (2n-2)-1 mod 2n-1 there is no solution for x if 2n-1 composite |
|
|
|
|
#2 |
|
"Sam"
Nov 2016
22×34 Posts |
Claim: If 2^n-1 is not prime, then 2^(n-2)-1 is a quadratic non-residue mod 2^n-1.
The claim is false, however the contrapositive is true: If 2^n-1 is prime, then 2^(n-2)-1 is a quadratic residue mod 2^n-1. n = 2 is a special case implying that x^2 = 0 mod 3, which has solution x = 0 (trivial). In all other cases, n must be odd. For simplicity, suppose p = 2^n-1. 2^(n-2)-1 = (p - 3)/4 p = 1 mod 3, and since p is prime, quadratic reciprocity tells us that x^2 = -3 mod p is solvable. Hence, x^2 = (p - 3)/4 mod p, 4*x^2 = p - 3 mod p, or (2*x)^2 = -3 mod p. The map x --> 2*x is a bijection in Zp/Z. Done. Proven. Now what remains is to show when your claim is false (given n is composite): If every prime q | p is congruent to 1 mod 3, then the claim is false. Since q = 1 mod 3, x^2 = -3 mod q will have a solution, using the Chinese Remainder Theorem allows us to construct a solution x to x^2 = -3 mod p, so the conclusion follows. There are infinitely many composite numbers of the form 2^n-1 such that the claim is false. If n = 3^k, then 2^n-1 is a counterexample if k > 2, since every prime q | 2^(3^k) - 1, the order of 2 mod q by definition divides 3^k, and trivially cannot be 1, so q = 1 mod 3. I do not find your claim interesting at all, what I find interesting is finding the density of primes n such that each prime q | 2^n-1 is congruent to 1 mod 3: Any arbitrary integer N has about ln(ln(N)) prime factors, so 2^n-1 will have on average, ln(n) prime factors by this assumption. The probability that all of these factors are congruent to 1 mod 3 is about 1/2^(ln(n)). I suppose there is a better estimate? |
|
|
|
|
#3 |
|
Feb 2017
Nowhere
122316 Posts |
|
|
|
![]() |
| Thread Tools | |
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| Nice-to-have's | kar_bon | Prime Wiki | 1 | 2019-02-26 09:57 |
| Nice pic | Dubslow | Forum Feedback | 0 | 2012-05-02 02:13 |
| Let's do another nice big GNFS job! | fivemack | Factoring | 84 | 2011-04-26 10:22 |
| Mersene Prime and Number Theory | Ricie | Miscellaneous Math | 24 | 2009-08-14 15:31 |
| 6 digit numbers and the mersene 40 | Sykes1024 | Math | 7 | 2004-02-10 09:43 |