%I #16 Feb 13 2013 23:58:30
%S 0,11,11,11,7,17,13,29,29,23,41,41,37,47,43,59,53,67,61,0,97,97,97,97,
%T 89,0,107,103,127,149,109,149,149,151,137,139,167,167,163,179,173,0,
%U 227,229,229,233,229,227,223,211,199,0,0,263,263,257,0,281,281
%N Let p_n=prime(n), n>=1. Then a(n) is the maximal prime p which differs from p_n, for which the intervals (p/2,p_n/2), (p,p_n], if p<p_n, or the intervals (p_n/2,p/2), (p_n,p], if p>p_n, contain the same number of primes, and a(n)=0, if no such prime p exists.
%C a(n)<p_n if and only if p_n is Ramanujan prime (A104272).
%C a(n)=0 if and only if p_n is a peculiar prime, i.e., simultaneously Ramanujan and Labos (A080359) prime (see sequence A164554).
%H V. Shevelev, <a href="http://www.cs.uwaterloo.ca/journals/JIS/VOL15/Shevelev/shevelev19.html">Ramanujan and Labos primes, their generalizations, and classifications of primes</a>, J. Integer Seq. 15 (2012) Article 12.5.4
%F If p_n is not a Ramanujan prime, then a(n) = A104272(n-pi(p_n/2)).
%e Let n=4, p_n=7. Since 7 is not Ramanujan prime, then a(4) = A104272(4-pi(3.5)) = A104272(2) = 11.
%Y Cf. A212493, A104272, A080359, A164554.
%K nonn
%O 1,2
%A _Vladimir Shevelev_ and _Peter J. C. Moses_, May 20 2012