Strong \((\delta,n)\)-complements for semi-stable morphisms (Q2214399): Difference between revisions

Let \(X\) be a normal quasi-projective variety over an algebraically closed field \(k\) of characteristic \(0\), and \(B\) an effective Weil divisor on \(X\). Let \(\pi\colon X' \rightarrow X\) be a projective birational morphism and let \(M = \pi_*M'\) for \(M'\) a \(\mathbb{Q}\)-Cartier nef divisor on \(X'\), such that \(K_X + B + M\) is \(\mathbb{Q}\)-Cartier. We say \((X, B+M)\) is a generalized pair, with boundary part \(B\) and moduli part \(M\). Suppose the generalized pair \((X,B+M)\) has \(\epsilon\)-log canonical singularities, and let \(X \rightarrow Z\) be a contraction of normal quasi-projective varieties. A Weil divisor \(B^+\) on \(X\) is a strong \((\delta,n)\)-complement over \(z \in Z\) if the following conditions hold over some neighborhood of \(z\): (a) \((X,B^+ + M)\) is a \(\delta\)-log canonical generalized pair with boundary part \(B^+\); (b) \(n(K_X + B^+ + M) \sim 0\); and (c) \(nB^+ \geq nB\). \textit{C. Birkar} [Ann. Math. (2) 190, No. 2, 345--463 (2019; Zbl 1470.14078)] proved boundedness of strong \((0,n)\)-complements for log canonical pairs, which he used in the proof of the BAB conjecture. In the paper under review, the authors conjecture a form of boundedness of strong \((\delta,n)\)-complements for generalized pairs: Conjecture 1: Let \(d,p \in \mathbb{N}\), \(\epsilon \in [0,1)\), and \(\Lambda \subset \mathbb{Q}\) a set satisfying the ACC condition with rational accumulation points. There exist \(n \in \mathbb{N}\) and \(\delta \in \mathbb{R}_{>0}\) only depending on \(d\), \(p\), \(\epsilon\) and \(\Lambda\) satisfying the following. Let \(X \rightarrow Z\) be a contraction between normal quasi-projective varieties, and \((X,B+M)\) be a generalized \(\epsilon\)-log canonical pair of dimension \(d\) such that: (1) \(-(K_X + B + M)\) is nef over \(Z\); (2) \(X\) is of Fano type over \(Z\); (3) \(\mathrm{coeff}(B) \subset \Lambda\); and (4) \(pM'\) is Cartier (notation as above). Then, for every point \(z \in Z\), there exists a strong \((\delta,n)\)-complement for \((X,B+M)\) over \(z\). Moreover we can pick \(\delta > 0\) if \(\epsilon > 0\). In this paper, the authors prove conjecture 1 if \(\delta = \epsilon = 0\), and in case \(Z = Spec(k)\), they prove it if \(\epsilon = 0\), or \(M'\) is trivial, or \(\Lambda\) is finite, and in any case they can take \(\delta = \epsilon\). They also prove the conjecture, fixing \(m \in \mathbb{N}\), if \(\epsilon > 0\), \(\Lambda\) is finite, \(M'\) is trivial, \(mK_Z\) is Cartier and \(X \rightarrow Z\) is a semi-stable morphism for the pair \((X,B)\). In this case, \(\delta > 0\) depends on \(m\) as well. They also give some applications of these theorems. The first one is an effective version of the generalized canonical bundle formula, and also an effective version of generalized adjunction to exceptional generalized log canonical centers. Finally, they show that conjecture 1 implies a conjecture due to McKernan concerning the singularities of the base a Mori Fiber Space.