OFFSET
1,1
COMMENTS
Let D be a fundamental discriminant (only the case where D is fundamental is considered because {Kronecker(D,k)} forms a primitive real Dirichlet character with period |D| if and only if D is fundamental), it seems that Sum_{primes q <= p} Kronecker(D,p) <= 0 occurs much more often than its opposite does. This can be seen as a variation of the well-known "Chebyshev's bias". Sequence gives the least prime that violates the inequality when D runs through all negative discriminants.
For any D, the primes p such that Kronecker(D,p) = 1 has asymptotic density 1/2 in all the primes, so a(n) should be > 0 for all n.
Actually, for most n, a(n) is relatively small compared with A003657(n). There are only 127 n's in [1, 3043] (there are 3043 terms in A003657 below 10000) such that a(n) > A003657(n). The largest terms among the 127 corresponding terms are a(1) = 608981813029 (with A003657(1) = 3), a(1955) = 24996194023 (with A003657(1955) = 6240) and a(847) = 1694759239 (with A003657(847) = 2787).
LINKS
Jianing Song, Table of n, a(n) for n = 1..3043
Wikipedia, Chebyshev's bias
EXAMPLE
Let D = -A003657(18) = -52, j(k) = Sum_{primes q <= prime(k)} Kronecker(D,q).
For k = 1, Kronecker(-52,2) = 0, so j(1) = 0;
For k = 2, Kronecker(-52,3) = -1, so j(2) = -1;
For k = 3, Kronecker(-52,5) = -1, so j(3) = -2;
For k = 4, Kronecker(-52,7) = +1, so j(4) = -1;
For k = 5, Kronecker(-52,11) = +1, so j(5) = 0;
For k = 6, Kronecker(-52,13) = 0, so j(6) = 0;
For k = 7, Kronecker(-52,17) = +1, so j(7) = 1.
The first time for j > 0 occurs at the prime 17, so a(18) = 17.
PROG
(PARI) b(n) = my(i=0); forprime(p=2, oo, i+=kronecker(n, p); if(i>0, return(p)))
print1(608981813029, ", "); for(n=4, 300, if(isfundamental(-n), print1(b(-n), ", ")))
CROSSREFS
KEYWORD
nonn
AUTHOR
Jianing Song, Feb 19 2019
STATUS
approved