Remarks on normal bases

From MaRDI portal



Abstract: We prove that any Galois extension of commutative rings with normal basis and abelian Galois group of odd order has a self dual normal basis. Also we show that if S/R is an unramified extension of number rings with Galois group of odd order and R is totally real then the normal basis does not exist for S/R.


The core of this pretty note is the following proposition: if an abelian Galois extension \(S/R\) of odd degree (where \(S\) and \(R\) are commutative rings) has a normal basis, then it has a self-dual normal basis under the trace form. This is used to present elegant proofs of two results which say, quite roughly, that Galois extensions of rings \(S/R\) of odd order such that Quot\((S)\) is a totally real number field tend to have no normal basis. If \(S\) and \(R\) are just the rings of integers in \(L\) resp. \(K\), and \(L/K\) is an unramified odd order Galois extension of totally real fields, then indeed \(S/R\) never has a normal basis, unless of course \(S=R\). If \(R\) is the ring of \(p\)-integers in \(K=\mathbb{Q}(\zeta_{p^k})\) and \(S\) the ring of \(p\)-integers in a \(p\)-ramified cyclic degree \(p^n\) extension, then \(S/R\) has a normal basis if and only if the extension \(S/R\) is cyclotomic. As acknowledged by the author, the results in this note were previously proved by Brinkhuis, Kersten and Michaliček, and the reviewer, partly with a different degree of generality. The proofs given in this note are quick and very elegant.











This page was built for publication: Remarks on normal bases

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4523118)