| Subject: | |
| From: | |
| Reply To: | |
| Date: | Thu, 25 Jun 2009 10:49:50 -0400 |
| Content-Type: | text/plain |
| Parts/Attachments: |
|
|
A Generalized Repunit Conjecture:
For any base b, which has prime numbers of the form Rb[n]=(b^p-1)/(b-1)
,the prime numbers will be distributed near the best fit line
Y = G * logb(logb(Rb[n])) + C
,where limit n-> Inf, G = 1/e^gamma = 0.56145948...
gamma is Euler's constant.
logb is the logarithm in base |b|.
Rb[n] is the nth sequential repunit prime number in base b.
C is a data fit constant which varies with b.
Generalizing the Wagstaff Mersenne Conjecture and the work of Pomerance
and Lenstra, as outlined by Caldwell at
http://primes.utm.edu/mersenne/heuristic.html
,we also have the following 3 properties:
1. The number of Repunit primes less than or equal to x is about
e^gamma * logb logb x.
2. The expected number of Repunit primes (b^p-1)/(b-1) with p between x
and |b|x is about e^gamma.
3. The probability that (b^p-1)/(b-1) is prime is about
e^gamma / (p log |b|).
For negative bases, p can not be 2, and the summation in 1 is closer to
e^gamma * logb(logb(x)/3), which will converge slowly. For practical
exponents, p < 1e6, G ~ 0.47.
The current linear data fits for the first few bases are listed below:
b #primes G C largest known exponent
-12 9 .48686 +.26244 24071
-11 7 .47650 +.17286 6113
-10 10 .33115 +.15048 3011
-9 9 .43930 +.66336 49223
-7 9 .63511 -.51818 106187
-6 14 .44014 +.14829 41341
-5 11 .65828 +.51129 193939
-3 23 .45540 +.86304 152287
-2 40 .47271 +.65033 986191
2 47 .55715 +.92757 43112609
3 16 .62162 +1.0636 57917
5 16 .43711 +.08611 201359
6 14 .46809 -.24593 79987
7 8 .78784 -.14225 126037
10 9 .70637 -.39537 270343
11 11 .31087 +.90528 20161
12 13 .37669 -.39149 46889
13 11 .31876 +.56306 31751
14 9 .55408 -.56477 67421
|
|
|