N. N. Luzin's problems on imbeddability and decomposability of projective sets (Q802551)

\textit{N. N. Luzin} [Collected works, Vol. II (Russian) (1958; Zbl 0088.262)] posed several questions concerning the possibility of carrying over results established for sets of the first order in the hierarchy of projective sets to classes of higher orders. Among these questions are the following four, stated here as propositional functions, the investigation of which is the aim of the present article: 1.1 Each single-valued element of \(\Sigma^ 1_ n\) can be embedded in a single- valued element of \(\Delta^ 1_ n\); 1.2. Each countable-valued member of \(\Sigma^ 1_ n\) can be embedded in a countable-valued member of \(\Delta^ 1_ n\); 1.3 (1.4) Each countable-valued element of \(\Delta^ 1_ n\) \((\Pi^ 1_{n-1})\) can be expressed as a union of a countable number of single-valued members of \(\Delta^ 1_ n\) \((\Pi^ 1_{n- 1})\). For \(n=1\), the truth of the first three corresponding propositions, as has been noted, was established by Luzin; the truth value of 1.4 is unknown. In their commentary on the work of Luzin, contained in volume 2 of the collected works, Novikov and Keldish indicated that 1.1 is false for \(n=2\) and, at the time of that writing, the remaining questions remained open, even in the case \(n=2\). In the present work, the author shows that 1.2, 1.3, and 1.4 are all false for \(n=2\). For values of n three or larger, the author observes that there seems to be little hope for the resolution of these questions within the framework of the Zermelo-Frankel axioms for set theory, either with or without the axiom of choice. Thus, he undertakes another line of investigation in order to determine whether these propositions be consistent with the classical axioms. With the additional assumption of the axiom of constructibility, the author is able to show that, for all \(n\geq 3\), the propositions 1.1- 1.4 are all false. Since the axiom of constructibility is consistent with ZFC, these results imply that the ZFC theory is insufficient to establish the truth of any of these four propositions for any \(n\geq 3\). Moreover, the logical consistency of the negations of 1.3 and 1.4 with ZFC can be put in even stronger form: Theorem 5. If the additional hypothesis IC, concerning the existence of strongly inaccessible cardinals, does not contradict the ZFC theory, then this theory is consistent with the following proposition: There exists a countable-valued \(\Pi^ 1_ 1\)- set that is not the union of a countable number of single-valued projective sets of any class. The proof of Theorem 5 makes use of a model due to Levy and Solovay in which every projective set is Lebesgue measurable. The author also makes use of this model in showing that, under the conditions of Theorem 5, proposition 1.2 is consistent with ZFC for all \(n\geq 3\). Whether or not the propositions 1.1, 1.3, 1.4 be consistent with ZFC remains unknown. Finally, the author shows that these last two theorems can be strengthened by replacing the axiom of choice with the principle of dependent choices.