Hereditarily non-Pythagorean fields

From MaRDI portal
Publication:1998964



Abstract: We prove for a large class of fields F that every proper finite extension of Fpyth, the pythagorean closure of F, is not a pythagorean field. This class of fields contains number fields and fields F that are finitely generated of transcendence degree at least one over some subfield of F.


A field \(F\) is non-pythagorean if there exists a sum of squares of elements of \(F\) that is not itself the square of an element of \(F\), and \(F\) is hereditarily non-pythagorean if every proper finite extension \(E\) of \(F\), with \(E/F\) not purely inseparable, is non-pythagorean. The pythagorean closure \(F_{\mathrm{pyth}}\) of \(F\) is the direct limit of all finite extensions of \(F\) that consist of chains of quadratic extensions obtained by iteratively adjoining square roots, inside of a fixed algebraic closure of \(F\), of sums of two squares of elements in the preceding field. The main results of this paper are that if \(F\) is either an algebraic number field or a field of characteristic different from 2 that is finitely generated of transcendence degree at least one over some subfield, then \(F_{\mathrm{pyth}}\) is hereditarily non-pythagorean. The authors also consider the cases when \(F\) is an infinite dimensional algebraic extension of a number field or an infinite dimensional algebraic extension of a function field of characteristic different from 2, and prove that the same result holds under the additional assumption that \(F\) is a Galois extension over some number field or some function field of characteristic different from 2.











This page was built for publication: Hereditarily non-Pythagorean fields

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1998964)