[seqfan] On infinitude of A090918-A090919

[Vladimir Shevelev]

Dear SeqFans,
 
Consider the following interesting sequences.
A090918: Beginning with 3, least prime, greater than the previous term, such that the arithmetic mean of first n terms is a prime;
A090919: Primes arising as the arithmetic mean of first n terms of  A090918.
The question of the infinitude of these sequences is reduced to unsolved
question, whether there are infinitly many primes in every "progression"
of the form a*prime(n)+b, gcd(a,b)=1. Indeed, I found the following recursive algorithm of parallel  calculation of these sequences.
Let a(n)=A090919(n),  b(n)=A090918(n). Then a(1)=b(1)=3 and, if for n>=2, we know a(n-1), b(n-1), then a(n) is the smallest prime x, such that 
n*x-(n-1)*a(n-1) is prime more than b(n-1). Now, knowing a(n), we have 
b(n)=n*a(n)-(n-1)*a(n-1).
 It is clear, that gcd (n, (n-1)*a(n-1))=1, so the infinitude of these sequences is reduced to above mentioned problem. 
Are there other (maybe, probabilistic) arguments ?
 
Regards,
Vladimir

 Shevelev Vladimir‎