Defining integrality at prime sets of high density over function fields (Q1599512)

Let \(F\) be a function field over a finite field \(C_p\) with \(p>2\) elements, i.e. \(F\) is a finite extension of the field of rational functions \(C_p(t)\) in the variable \(t\) and with coefficients in \(C_p\). The author proves that for any finite set \(S\) of \(F\)-primes and for any \(\varepsilon >0\) there exists a set of \(F\)-primes \(W\), of Dirichlet density greater than \(1-\varepsilon \) such that \(S\subset W\) and \({\mathcal O}_{F,S}\) has a Diophantine definition over \({\mathcal O}_{F,W}\). Here \({\mathcal O}_{F,W}=\{x\in F\mid \text{ord}_qx\geq 0\), \(\forall q\not\in W\}\) and similarly for \({\mathcal O}_{F,S}\). A set is ``Diophantine'' if and only if it is positive-existentially definable in the language of rings (which contains symbols for addition and multiplication), augmented by a symbol which represents an element of \(F\) which is transcendental over \(C_p\) (that the language just described suffices for the construction does not appear explicitly in the paper but follows from the given proofs). As a consequence one obtains that if the positive-existential theory of \({\mathcal O}_{F,S}\) is undecidable (something which is expected to hold true for any \(S\) and is known in several cases) then the positive-existential theory of \({\mathcal O}_{F,W}\) is undecidable as well. Moreover, the author contrasts the result against a question of Mazur whose analogue in the case of positive characteristic would imply that \({\mathcal O}_{F,S}\) is not existentially definable over \(F\). Results for number fields, of a similar type as the article's, have been obtained by the author and B. Poonen. In the reviewer's opinion these results combined show that Mazur's question is intriguing and its answer unpredictable for the moment.