Endomorphisms of elliptic curves and undecidability in function fields of positive characteristic. (Q1427400)

The author proves that the full first order theory of a field of rational functions over a field of characteristic \(\geq 5\) is undecidable in the ring language with a predicate for the variable. The result was proven by \textit{G. L. Cherlin} for perfect ground fields [Logic colloquium '82, Stud.\ Logic Found.\ Math.\ 112, 85--95 (1984; Zbl 0551.03027)]. However, the techniques in the present paper are different and based on the arithmetic of elliptic curves, and everything is set up to guarantee that as many of the formulae as possible are actually existential in the given language. The proof works as follows. The endomorphism ring of a suitably chosen elliptic curve over the ground field \(F\) can be seen as the \(F(t)\)-rational points of a twist of that curve (up to 2-torsion), and is at the same time a quadratic order \(I\) in a number field. Many relations (addition, certain divisibilities, powers of the Frobenius) on \(I\) can thus be transported to the twisted elliptic curve and become (existential) relations in the language of \(F(t)\). For multiplication (or equivalently, squaring) it is necessary to rely upon a universal quantifier, and this is the only obstruction to giving a negative answer to Hilbert's tenth problem for such function fields.