Genus theory and \(\varepsilon\)-conjectures on \(p\)-class groups (Q2329282): Difference between revisions

Let \(K\) be an algebraic number field of degree \(d\) and for prime \(p\) denote by \(h_K[p]\) the number of ideal classes \(X\) in \(K\) satisfying \(X^p=1\). It is conjectured (\textit{the \(\varepsilon\)-conjecture\/}) that for every \(\varepsilon>0\) one has \[ h_K[p]\le C(\varepsilon,p,d)D_K^\varepsilon,\tag{1} \] where \(D_K\) is the absolute value of the discriminant of \(K\). For \(p=2\) this follows from the Gaussian theory of genera. \textit{J. S. Ellenberg} and \textit{A. Venkatesh} [Int. Math. Res. Not. 2007, No. 1, Article ID rnm002, 18 p. (2007; Zbl 1130.11060)] proved that the General Riemann Hypothesis implies the bound \[ h_K[p]\le D_K^{1/2-a+\varepsilon}\tag{2} \] with \(a=1/(2p(d-1))\) and obtained also \(h_K[3]\ll D_K^{1/2+\varepsilon}\) for \(d=2,3\). Later \textit{J. Ellenberg} et al. [Algebra Number Theory 11, No. 8, 1739--1778 (2017; Zbl 1398.11136)] established (2) for almost all fields of degree \(d\le5\) and large \(p\), assuming in the case \(d=4\) that the Galois group of the normal closure of \(K\) is not dihedral. The author establishes the conjecture (1) for all fields in the family \(\mathcal{F}_p\) defined in the following way. For prime \(p\) let \(l_1<l_2<\cdots\) be the sequence of all primes congruent to \(1\) mod \(p\). The family \(\mathcal{F}_p\) consists of all cyclic extensions \(K/Q\) of degree \(p\) having its conductor of the form \(\prod_{i=1}^Nl_i\) (\(N=1,2,\dots\)), assuming additionally in the case \(p=2\) that \(2\) is unramified in \(K\). The proof is based on previous results of the author on the structure of finite \(Z_p[G]\)-modules for \(G\) being the cyclic group of \(p\) elements [\textit{G. Gras}, Proc. Indian Acad. Sci., Math. Sci. 127, No. 1, 1--34 (2017; Zbl 1410.11122)]. He discusses also the \textit{strong \(\varepsilon\)-conjecture} \[ \#(H_K\otimes Z_p)\ll D_K^\varepsilon, \] and in the last part of the paper presents computations supporting this conjecture.