OFFSET
1,1
COMMENTS
Conjecture: The two most frequent ratios between consecutive prime gaps are 2 and 1/2, and both ratios occur with about the same frequency.
LINKS
Harvey P. Dale, Table of n, a(n) for n = 1..1000
FORMULA
Conjecture: lim_{n->inf} n/primepi(a(n)) > k > 0 for some k.
EXAMPLE
The first three consecutive primes are 2, 3 and 5, and (5-3)/(3-2)=2, so the first term is a(1)=2, that is, the first prime of (2,3,5).
The next three consecutive primes are 3, 5 and 7, and (7-5)/(5-3)=1, so the first prime of (3,5,7) is not in the list.
The next three consecutive primes are 5, 7 and 11, and (11-7)/(7-5)=2, so the second term is a(2)=5, that is, the first prime of (5,7,11).
The prime 13 is also in the list because (19-17)/(17-13)=1/2.
MATHEMATICA
b={};
Do[If[Abs[Log[2, (Prime[j+2]-Prime[j+1])/(Prime[j+1]-Prime[j])]]==1, AppendTo[b, Prime[j]]], {j, 1, 200}];
Print@b
Select[Partition[Prime[Range[250]], 3, 1], (#[[3]]-#[[2]])/(#[[2]]-#[[1]]) == 2||(#[[3]]-#[[2]])/(#[[2]]-#[[1]])==1/2&][[All, 1]] (* Harvey P. Dale, Mar 14 2022 *)
PROG
(PARI) isok(p) = my(q = nextprime(p+1), r = nextprime(q+1), f = (r-q)/(q-p)); (f == 2) || (f == 1/2);
forprime(p=2, 1000, if (isok(p), print1(p, ", "))); \\ Michel Marcus, Apr 23 2018
CROSSREFS
KEYWORD
nonn
AUTHOR
Andres Cicuttin, Apr 18 2018
STATUS
approved