Genericity for Mathias forcing over general Turing ideals (Q503254)

Mathias forcing is one of the best-known notions of forcing from set theory for adding a real number to the ground model. The conditions of Mathias forcing are pairs \((F,X)\) of a finite set \(F\subseteq\omega\) and an infinite set \(X\subseteq\omega\) whose minimal element is larger than the maximum of \(F\). \((F^{\prime},X^{\prime})\) is said to extend \((F,X)\) if and only if \(F\subseteq F^{\prime}\subseteq F\cup X\) and \(X^{\prime}\subseteq X\). Intuitively, \(F\) is a set of elements that will definitely belong to the new set, while \(X\) consists of possible candidates for further additions. Various notions of forcing for adding real numbers have become relevant in computability theory as well. In particular, Mathias forcing is frequently used, e.g. in the reverse mathematics of combinatorial principles. Here, the conditions are additionally constrained to satisfy a certain requirement of effectivity (such as being recursive) and genericity is restricted by, e.g. merely demanding that all dense sets definable with a formula of a certain complexity are intersected. The paper under review investigates Mathias forcing when the conditions come from an arbitrary countable Turing ideal \(\mathcal{I}\), the genericity requirements considered being \(n\)-\(\mathcal{I}\)-genericity for \(n\in\omega\) (usually \(n\geq 3\)), which means that every set definable from the parameter \(\mathcal{I}\) by an arithmetical \(\Sigma_{n}\)-formula is met or avoided, and \(\mathcal{I}\)-genericity, which means \(n\)-\(\mathcal{I}\)-genericity for all \(n\in\omega\). We mention some of the results. It is shown that, for any ideal \(\mathcal{I}\) and any \(A\notin\mathcal{I}\), there is an \(\mathcal{I}\)-generic \(X\) such that \(A\) is not Turing-reducible to \(X\) (Proposition 3.2). As an application, a new proof is given of the existence of \(X,Y\subseteq\omega\) such that \(X\) grows faster than any function recursive in \(Y\), while \(Y^{\prime\prime}\) is not Turing-reducible to \(X^{\prime}\) (Corollary 3.3). More information on the reals computable from \(n\)-\(\mathcal{I}\)-generic reals is obtained in Corollary 4.3: If \(G\) is \(n\)-\(\mathcal{I}\)-generic, \(n\geq 3\) and \(X\) is \(\Delta_{n}\) in \(\mathcal{I}\) and computable from \(G\), then \(X\) belongs to \(\mathcal{I}\). On the other hand, when \(X\) is \(3\)-\(\mathcal{I}\)-generic and \(Y\subseteq X\) is \(3\)-generic for the ideal generated by \(X\), then \(X\) is not computable from \(Y\). Finally, if \(n\geq 3\) and \(G\) is \(3\)-\(\mathcal{I}\)-generic, then a set \(X\) which is generic for the ideal of recursive sets is computable from \(G\) (Theorem 5.6). The related claim that, for any ideals \(\mathcal{I}\subseteq\mathcal{J}\) and every \(\mathcal{J}\)-generic, there is some \(3\)-\(\mathcal{I}\)-generic real computable from it remains an open question (Question 5.5). The paper is carefully written and briefly recalls the basic notions from the theory of forcing. Still, the reader is required to have a considerable background of computability theory.