The distribution of \(p\)-torsion in degree \(p\) cyclic fields (Q782332): Difference between revisions

The Cohen-Lenstra heuristics for quadratic fields (extended by \textit{H. Cohen} and \textit{J. Martinet} [Math. Comput. 48, 123--137 (1987; Zbl 0627.12006)] to number fields of general degrees) predict the average growth functions of \(p\)-parts of class groups for primes \(p\). These are the beacons leading to the state of the art study of class groups and there is compelling, mounting evidence supporting the conjectures even if no case of the conjectures is completely known yet. The conjectured heuristics avoided predicting the case when \(p\) divides the degree (hence, deal with \(p\) odd in case of quadratic fields) and it was \textit{F. Gerth III} [Invent. Math. 77, 489--515 (1984; Zbl 0533.12004)] who extended these conjectures to the case of \(p\)-torsion of class groups of degree \(p\) cyclic fields. \textit{É. Fouvry} and \textit{J. Klüners} [Lect. Notes Comput. Sci. 4076, 40--55 (2006; Zbl 1143.11352)] proved Gerth's predicted generalization on the behaviour of \(4\)-rank of class groups of quadratic fields. Gerth also conjectured for each prime \(p\), a distribution for a certain subgroup of \(Cl_K[p]\) for cyclic \(p\)-extensions \(K\). If \(G\) is the Galois group of \(K\), generated by \(\sigma_K\) say, then \(\phi_K:=1-\sigma_K\) acts on \(Cl_K[p]\) and Gerth considered the natural filtration \[ Cl_K[p]^G=\mathrm{ker}(\phi_k) \subseteq\mathrm{ker}(\phi_K^2)\subseteq \cdots \subseteq\mathrm{ker}(\phi_K^{p-1}) = Cl_K[p]. \] Gerth conjectured a distribution for the \(p\)-rank of \(\phi_K(\mathrm{ker}(\phi_K^2))\). The paper under review has two main theorems as follows. Theorem. The density of cubic fields \(K\) where the \(3\)-rank of \(Cl_K[3]/Cl_K[3]^G\) is \(s\), equals the expression \[ \frac{\prod_{n=1}^{\infty} (1 - 1/3^n)}{\prod_{i=1}^s (1- 1/3^i)\prod_{j=1}^{s+1} (1- 1/3^j) 3^{s(s+1)}}. \] Further, let \(p>3\) and assume GRH for Artin L-functions. Then, the density of degree \(p\) cyclic fields \(K\) where the \(p\)-rank of \(\phi_K(\mathrm{ker}(\phi_K^2))\) is \(s\), equals the expression \[ \frac{\prod_{n=1}^{\infty} (1 - 1/p^n)}{\prod_{i=1}^s (1- 1/p^i)\prod_{j=1}^{s+1} (1- 1/p^j) p^{s(s+1)}}. \] To state the second main result from which the above theorem is deduced, define for each positive integer \(k\): \(M_k(p) = \lim_{X \rightarrow \infty}\frac{\sum_{K,D_K < X} p^k~\mathrm{rank}_p \phi_K(\mathrm{ker}(\phi_K^2))}{\sum_{K,D_K < X} 1}.\) Theorem. Let \(k \geq 1\), and \(p\) be an odd prime. If \(p>3\), assume that GRH holds for Artin L-functions. Then \[ M_k(p) = N(k+1,p)- N(k,p) \] where \(N_k(p)\) is the number of subspaces of \(\mathbb{F}_p^k\). The author uses the technique employed by Fouvry and Klüners to prove the above theorem.