A model with Suslin trees but no minimal uncountable linear orders other than \(\omega_1\) and \(- \omega_1\) (Q2327970)

Structure theorems for linear orders, such as \textit{R. Laver}'s theorem [Ann. Math. (2) 93, 89--111 (1971; Zbl 0208.28905)], or \textit{J. T. Moore}'s five elements base theorem [Ann. Math. (2) 163, No. 2, 669--688 (2006; Zbl 1143.03026)], are among the great triumphs of infinitary combinatorics. In [in: Ordered sets. Proceedings of the NATO Advanced Study Institute held at Banff, Canada, August 28 to September 12, 1981. Dordrecht-Boston-London: D. Reidel Publishing Company. 239--277 (1982; Zbl 0506.04003)], \textit{J. E. Baumgartner} studied the structure of linear orders of cardinality \(\aleph_1\), focusing on minimal and universal types, under various set theoretical scenarios. This analysis established that a minimal linear order of size \(\aleph_1\) is either \(\omega_1\), \(\omega_1^*\), an \(\omega_1\)-dense subset of the reals, or a Specker line. Moreover, Specker lines are related to Aronszajn trees (similarly to the relation between Suslin lines and Suslin trees). In order to obtain a model in which the only minimal orders are \(\omega_1\), \(\omega_1^*\), one needs to make sure that no subset of the reals of size \(\omega_1\) is minimal and that there is no minimal Specker line, or equivalently there is no club-minimal Aronszajn tree. The non-existence of a minimal real type follows from \(\mathrm{CH}\), by a result of Sierpiński. For Specker lines the situation is more involved, as Baumgartner [loc. cit.] proved that a strengthening of \(\mathrm{CH}\), \(\diamondsuit^+\), implies the existence of a club-minimal Aronszajn tree, and thus a minimal Specker line. Baumgartner asked whether the existence of a Suslin tree is already sufficient in order to obtain a minimal Specker line. In [Mich. Math. J. 55, No. 2, 437--457 (2007; Zbl 1146.03037)], \textit{J. T. Moore} constructs a model of \(\mathrm{CH}\) in which there is no club-minimal Aronszajn tree. He achieves this by uniformazing colourings on ladder systems of Aronszajn trees and showing that if \(A\) is a minimal Aronszajn tree and a \(2^{\aleph_0} < 2^{\aleph_1}\), then there is a colouring of a ladder system on \(A\) that cannot be uniformized. It is currently unknown whether there is a non-special Aronszajn in Moore's model, or even whether this uniformization property is consistent, where \(A\) is Suslin. In the paper under review, the author answers Baumgartner's question [loc. cit.] negatively, by providing a model in which a restricted version of Moore's uniformization principle holds while a specific rigid Suslin tree \(R\) is preserved. The forcing which is used in the paper is very different from Moore's iteration. The author uses a variant of Jensen's method for specialization of all Aronszajn trees without adding reals, as explained in [\textit{K. J. Devlin} and \textit{H. Johnsbraten}, The Souslin problem. Berlin-Heidelberg-New York: Springer-Verlag (1974; Zbl 0289.02043)]. The crux of this method is that the iterations is embedded into Suslin trees, and thus does not add reals. The author utilizes a lot of the machinery that was developed in [\textit{U. Abraham} and \textit{S. Shelah}, Isr. J. Math. 50, 75--113 (1985; Zbl 0566.03032)], in order to preserve the Suslinity of \(R\) during the iteration. In particular, in the obtained model each Aronszajn tree \(A\) is either a subtree of the finite power of \(R\) or that every colouring on a ladder system is uniformized on \(A\). In both cases, the tree \(A\) cannot be minimal, by two completely different reasons. While this iteration method is usually considered obsolete by the further developments of \textit{S. Shelah} of the general theory of completeness systems [Proper and improper forcing. 2nd ed. Berlin: Springer (1998; Zbl 0889.03041)] the author shows that some of its features can be used in order to define a better behaving successor step, replacing parts of the arguments that uses completeness systems by chain condition and distributivity arguments. Indeed, the author points to the exact technical issue that once resolved would enable one to prove the same theorem using plain countable support iterations. The paper concludes with some natural open problems and suggestions for possible ways to attack them.