Modification of Cohen-Lenstra heuristics for ideal class groups and numbers of certain real quadratic fields (Q1387430)

The famous Cohen-Lenstra heuristics predict the probability that the class number of a quadratic number field is divisible by a given prime number \(p\). In this paper, the authors look at families of quadratic fields whose class groups contain a potential element of order \(p\). This notion is best explained by an example: for an integer \(n \geq 1\), consider the field \(k_n = {\mathbb Q}(\sqrt{d(n)})\) with \(d(n) = n^2+3n+9\); the relation \(-27 = (2n+3)^2 - 4d(n)\) implies that the prime ideal \((3,2n+3 + 2\sqrt{d_n})\) above \(3\) has order dividing \(3\) in the class group of \(k_n\): such an ideal is called a potential element of order \(3\). Using the idea of Cohen-Lenstra, the authors predict that real quadratic fields with a potential element \(P\) of order \(p\) in their class groups have class number divisible by \(p\) with probability \(1 - (1-1/p)(1-1/p^2)\cdots\). Similarly, the probability that \(P\) itself is nontrivial should be \(1/p\). Numerical experiments for \(p = 3\) support the predictions, but the convergence is rather slow.