Finite generation of the \(\log\) canonical ring for \(3\)-folds in char \(p\) (Q1707258)

The Minimal Model Program is a very important conjectural framework in the classification of higher dimensional algebraic varieties. It was established around forty years ago for varieties with a certain class of singularities called \textit{Kawamata log terminal} (klt for short). A large part of the program has been proved to hold in characteristic zero (see [\textit{C. Birkar} et al., J. Am. Math. Soc. 23, No. 2, 405--468 (2010; Zbl 1210.14019)]). However, in characteristic \(p > 0\) very little is known. There have been several recent advances in the case of surfaces [\textit{H. Tanaka}, Nagoya Math. J. 216, 1--70 (2014; Zbl 1311.14020)] and threefolds when \(p > 5\), see for instance [\textit{C. D. Hacon} and \textit{C. Xu}, J. Am. Math. Soc. 28, No. 3, 711--744 (2015; Zbl 1326.14032)], [\textit{P. Cascini} et al., Ann. Sci. Éc. Norm. Supér. (4) 48, No. 5, 1239--1272 (2015; Zbl 1408.14020)], [\textit{C. Birkar}, Ann. Sci. Éc. Norm. Supér. (4) 49, No. 1, 169--212 (2016; Zbl 1346.14040)] and [ \textit{C. Birkar} and \textit{J. Waldron}, Adv. Math. 313, 62--101 (2017; Zbl 1373.14019)]. Nevertheless, several basic results in the MMP in characteristic \(p > 5\) remain widely open. In the paper under review, the author proves finite generation of the log canonical ring of a projective pair of dimension three with klt singularities defined over a field of characteristic \(p > 5\). The proof of this result in characteristic zero relies on the \textit{canonical bundle formula}, a fundamental result that describe the behaviour of the canonical bundle under special type of fibrations. Unfortunately, in general the canonical bundle formula is false in positive characteristic (see [\textit{J. Witaszek}, ``On the canonical bundle formula and log abundance in positive characteristic'', Preprint, \url{arXiv:1711.04380}]). The author, instead, proves finite generation via a special case of log abundance.