Year-end appeal: Please make a donation to the OEIS Foundation to support ongoing development and maintenance of the OEIS. We are now in our 61st year, we have over 378,000 sequences, and we’ve reached 11,000 citations (which often say “discovered thanks to the OEIS”).
%I #19 Jul 13 2022 21:58:31
%S 1369391,1371989,1378217,1393937,1418117,1426127,1428767,1429367,
%T 1430291,1494509,1502141,1502717,1506611,1510307,35278697,35287001,
%U 35447171,35468429,35468861,35470271,35595869,45274121,45276227,45304157,45306827,45324569,45336461,45336917
%N First member p(m) of the m-th twin prime pair such that d(m) > 0 and d(m-1) < 0, with d(k) = k/Integral_{x=2..p(k)} 1/log(x)^2 dx - C, C = 2*A005597 = A114907.
%C The sequence gives the positions, expressed by A001359(m), where the number of twin prime pairs m seen so far first exceeds the number predicted by the first Hardy-Littlewood conjecture after having been less than the predicted number before. A347279 gives the transitions in the opposite direction.
%C The total number of twin prime pairs up to that with first member x in the intervals a(k) <= x < A347279(k) is above the Hardy-Littlewood prediction. The total number of twin prime pairs up to that with first member x in the intervals A347279(k) <= x < a(k+1) is below the H-L prediction.
%H Hugo Pfoertner, <a href="/A347278/b347278.txt">Table of n, a(n) for n = 1..12135</a>
%H Wikipedia, <a href="https://en.wikipedia.org/wiki/Twin_prime#First_Hardy%E2%80%93Littlewood_conjecture">Twin prime</a>, First Hardy-Littlewood conjecture.
%H Marek Wolf, <a href="https://arxiv.org/abs/1107.2809">The Skewes number for twin primes: counting sign changes of pi_2(x)-C_2 Li_2(x)</a>, arXiv:1107.2809 [math.NT], 14 Jul 2011.
%o (PARI) halicon(h) = {my(w=Set(vecsort(h)), n=#w, wmin=vecmin(w), distres(v,p)=#Set(v%p)); for(k=1,n, w[k]=w[k]-wmin); my(plim=nextprime(vecmax(w))); prodeuler(p=2, plim, (1-distres(w,p)/p)/(1-1/p)^n) * prodeulerrat((1-n/p)/(1-1/p)^n, 1, nextprime(plim+1))}; \\ k-tuple constant
%o Li(x, n)=intnum(t=2, n, 1/log(t)^x); \\ logarithmic integral
%o a347278(nterms,CHL)={my(n=1,pprev=1,np=0); forprime(p=5,, if(p%6!=1&&ispseudoprime(p+2), n++; L=Li(2,p); my(x=n/L-CHL); if(x*pprev>0, if(pprev>0,print1(p,", ");np++; if(np>nterms,return)); pprev=-pprev)))};
%o a347278(10,halicon([0,2])) \\ computing 30 terms takes about 5 minutes
%Y Cf. A001359, A005597, A114907, A152051, A347279.
%Y a(1) = A210439(2) (Skewes number for twin primes).
%K nonn
%O 1,1
%A _Hugo Pfoertner_, Aug 26 2021