|
Found a web site that estimates the number of Generalized Fermat Primes in a sequence:
http://yves.gallot.pagesperso-orange.fr/primes/stat.html
N = 32768
b = 3,093,872 to 5,984,848, primes found by PRPNet = 33
Expected number of GF primes: 33.44 => error = -0.44
Poisson distribution:
Chance of 33 primes: 6.91%
Expected number of GF primes for 6,000,000 < b < 6,100,000 is 1.13
N = 65536
b = 684,736 to 2,894,394, primes found by PRPNet = 12
Expected number of GF primes: 26.37 => error = -14.37
Suspect PRPNet range tested or primes found not correct in stats.
Expected number of GF primes for 2,900,000 < b < 3,000,000 is 1.15
N = 262144
b = 2,900 to 864,200, primes found by PRPNet = 4
Expected number of GF primes: 3.39 => error = -0.61
Poisson distribution:
Chance of 4 primes: 18.54%
Expected number of GF primes for 864,000 < b < 1,000,000 is 0.49
N = 524288
Chance of any GF prime with b > 700,000 doesn't hit 50% till we've tested up to 1,080,000
Playing with the N and b ranges gives you a good idea of expected hit rates. History has shown this estimation to be pretty good.
____________
 | 
