Definable MAD families and forcing axioms (Q2660156)

The main result of the paper under review is that under \(\mathrm{ZFC}+ \mathrm{BPFA}\) (\(\mathrm{BPFA}\) stands for the bounded proper forcing axiom) if \(\omega_1\) is not remarkable in Gödel's constructible universe \(\mathbf L\), then there is an infinite \(\mathbf{\Pi}^1_2\) maximal almost disjoint (\(\mathrm{MAD}\)) family of reals. The main body of the proof of this theorem consists of a series of technical lemmas, where they essentially stand on two major pillars. One is that the authors use a consequence of \(\mathrm{BPFA}\), namely the existence of a \(\Sigma_1\)-well-ordering of the reals of length \(\omega_2\) with the nice property that initial segments can be defined by a single \(\Sigma_1\)-formula with a parameter in \(\mathcal P(\omega_1)\) (this is, of course, of independent interest). The authors extract this property from the \(\Sigma_1\)-well-ordering of the reals introduced and proved to be a consequence of \(\mathrm{BPFA}\) by \textit{A. E. Caicedo} and \textit{B. Veličković} [Math. Res. Lett. 13, No. 2--3, 393--408 (2006; Zbl 1113.03039)]. The other pillar bases upon Schindler's work on remarkable cardinals, that is if \(\omega_1\) is not remarkable in \(\mathbf{L}\) and \(\mathrm{BPFA}\) holds, then there is a real \(r\) with \(\omega_1=\omega_1^{\mathbf{L}[r]}\). These two major steps are exactly what they need to get from \(\mathrm{BPFA}\) and the non-remarkability of \(\omega_1\) in \(\mathbf L\). The rest of the proof involves certain coding lemmas to enable the authors to perform an inductive construction of the desired infinite MAD family. As a corollary of the main theorem they show that the following theories are equiconsistent. \begin{itemize} \item \(\mathrm{ZFC} ~+\) there exists a \(\Sigma_1\)-reflecting cardinal above a remarkable cardinal. \item \(\mathrm{ZFC+ BPFA}~+\) there is no infinite \(\mathbf{\Pi}^1_2\) MAD family. \item \(\mathrm{ZFC+ BPFA}~+\) there is no infinite MAD family in \(\mathbf{L}(\mathbb R)\). \end{itemize} They also show that if for every real \(r\), \(r^{\sharp}\) exists (a consequence of \(\mathrm{BMM}\)), then there is no infinite \(\Sigma^1_3\) MAD family.