scientific article; zbMATH DE number 1985576

From MaRDI portal
Publication:4429258





The author introduces a new version of the so-called Cohen-Lenstra heuristics concerning predictions how certain class groups are distributed, according to the size of their automorphism groups. His new ingredient is an action of the Galois group. Let us fix an odd prime \(p\) and a finite \(p\)-group \(\Delta \). Let \(I_ \Delta \) denote the augmentation ideal of \(\mathbb Z_ p[\Delta ]\), and \(M_ \Delta =M/I_ \Delta M\) the module of coinvariants attached to a \(\mathbb Z_ p[\Delta ]\)-module \(M\). The author proves that for every finite abelian \(p\)-group \(G\) there is an equality \(\sum _ {M_ \Delta \cong G}| \Aut_ {\mathbb Z _ p[\Delta ]}(M)| ^ {-1} =| \Aut(G)| ^ {-1}\), where \(M\) runs through all isomorphism classes of finite \(\mathbb Z_ p[\Delta ]\)-modules of projective dimension at most 1 such that \(M_ \Delta \) is isomorphic to \(G\); here \(\Aut_ {\mathbb Z_ p[\Delta ]}(M)\) means the group of automorphisms of \(M\) over \(Z_ p[\Delta ]\) and \(\Aut(G)\) is the automorphism group of \(G\). The author also proves that analogical sum without the assumption on the projective dimension is divergent if \(\Delta \) is noncyclic group. For each prime \(r\) which is congruent to \(3\) modulo \(4\) and to \(1\) modulo \(p\), let \(K_ r\) be the unique abelian extension of \(\mathbb Q\) of degree \(2p\) and conductor \(r\). The first mentioned theorem makes possible the author to state the following equivariant Cohen-Lenstra heuristics: For any given finite \(\mathbb Z_ p[\Delta ]\)-module \(M\) of projective dimension at most 1, the probability distribution of the \(p\)-part \(C^ -_ {K_ r}\{p\}\) of the minus part of the class group of \(K_ r\) to be \(Z_ p[\Delta ]\)-isomorphic to \(M\) is proportional to \(| \Aut_ {\mathbb Z_ p[\Delta ]}(M)| ^ {-1}\), i.e. \(P(C^ -_ {K_ r}\{p\}\cong M)=S^ {-1}\cdot | \Aut_ {\mathbb Z_ p[\Delta ]}(M)| ^ {-1}\), where the constant \(S\) is described in the paper. The author brings also some numerical evidence for this in case \(p=3\).











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4429258)