Semi-ampleness of NQC generalized log canonical pairs (Q6098436)

A generalized pair \((X, B, M)\) consists of a log pair \((X, B)\) and an additional (push-forward of a) nef divisor \(M\). It was introduced in [\textit{C. Birkar} and \textit{D.-Q. Zhang} Publ. Math., Inst. Hautes Étud. Sci. 123, 283--331 (2016; Zbl 1348.14038)] to treat the structure from the canonical bundle formula. Many results in the minimal model program also hold for generalized pairs. Moreover, generalized pairs are successfully used in solving log-standing conjectures [\textit{C. Birkar}, Ann. Math. (2) 193, No. 2, 347--405 (2021; Zbl 1469.14085); Ann. Math. (2) 190, No. 2, 345--463 (2019; Zbl 1470.14078)]. One of the main results in this paper is the following. Theorem. Let \((X, B, M)/Z\) be a generalized lc pair and \(A \geq 0\) an \(\mathbb R\)-divisor on \(X\), such that \((X, B+A, M)\) is generalized lc and \(K_X + B + A + M \sim_{\mathbb R,Z} 0\). Then: \begin{itemize} \item[1.] \((X, B,M)/Z\) has a Mori fiber space or a log minimal model \((Y, B_Y ,M_Y)/Z\). \item[2.] If \(K_Y + B_Y + M_Y\) is nef\(/Z\), then it is semi-ample \(/Z\). \item[3.] If \((X, B,M)\) is \(\mathbb Q\)-factorial generalized dlt, then any \((K_X + B + M)\)-MMP\(/Z\) with scaling of an ample\(/Z\) \(\mathbb R\)-divisor terminates. \end{itemize} This result generalizes [\textit{C. Birkar}, Publ. Math., Inst. Hautes Étud. Sci. 115, 325--368 (2012; Zbl 1256.14012); \textit{C. D. Hacon} and \textit{C. Xu}, Invent. Math. 192, No. 1, 161--195 (2013; Zbl 1282.14027); \textit{K. Hashizume}, Manuscr. Math. 160, No. 3--4, 285--314 (2019; Zbl 1476.14037)] to the category of generalized pairs. The (1) and (3) items are already established in [\textit{J. Liu} and \textit{L. Xie}, ``Relative Nakayama-Zariski decomposition and minimal models of generalized pairs'', Peking Math. J. (to appear)]. The proof of (2) uses the canonical bundle formula for generalized pairs to perform the reduction and the gluing theory of Kollár to obtain the semi-ampleness. A similar strategy was adopted in [\textit{C. D. Hacon} and \textit{C. Xu}, Invent. Math. 192, No. 1, 161--195 (2013; Zbl 1282.14027)]. A corollary of the above result is that flips exist for generalized pairs. This result was proved in [\textit{C. D. Hacon} and \textit{J. Liu}, ``Existence of generalized lc flips'', Preprint, \url{arXiv:2105.13590}] with the additional assumption that \(M\) is \(\mathbb R\)-Cartier. The second main result asserts that if a generalized lc pair \((X, B, M)/U\) has a good minimal model over an open set \(U_0 \subset U\) and each generalized lc center intersects with \(X_{U_0}\), then \((X, B, M)/U\) admits a good minimal model. This result generalizes [\textit{C. Birkar}, Publ. Math., Inst. Hautes Étud. Sci. 115, 325--368 (2012; Zbl 1256.14012); \textit{C. D. Hacon} and \textit{C. Xu}, Invent. Math. 192, No. 1, 161--195 (2013; Zbl 1282.14027); \textit{K. Hashizume}, Manuscr. Math. 160, No. 3--4, 285--314 (2019; Zbl 1476.14037)] to the category of generalized pairs. The paper also contains examples showing that approaches which are successful in establishing semi-ampleness for usual pairs may fail for generalized pairs.