Effective adjunction theory (Q1989476)

Consider firstly \(X\) a normal complex projective variety with at most canonical singularities, \(K_X\) stands for its canonical divisor and \(H\) is a Cartier divisor on \(X\). The notion of \textit{termination of adjunction} is not standard, several meanings are proposed in the literature: (A) for every (for some) \(H\) big there exists a natural number \(m_0(H)>0\) such that \(mK_X+H\) is not pseudoeffective for \(m \geq m_0(H)\); (B) for every \(H\) big there exists a natural number \(m_0(H)>0\) such that \(m_0(H)K_X+H\) is not effective; (C) same as (B) but \(H\) very ample; (D) for some (for every) \(H_0\) big and for every\(k>0\) there exists \(m_0(k)>0\) such that \(m_0(k)K_X+kH_0\) is not effective. It is clear that (A) implies (B), (B) implies (C) and (C) implies (D). The authors prove in this paper that the four definitions are in fact equivalent and moreover adjunction terminates if and only if \(X\) is uniruled (Theorem 3, Corollary 1 and 2). They also conjecture that a more general definition is also equivalent to those presented above. To be precise: for \(H\) an effective Cartier divisor \textit{adjunction terminates in the classical sense for \(H\)} if there exists an integer \(m_0 \geq 1\) such that \(H+mK_X\) is not effective for \(m \geq m_0\). In Proposition 2 a partial result towards this conjectire is shown: if adjunction terminates in the classical sense then \(X\) has negative Kodaira dimension. Now consider \((X,L)\) a quasipolarized projective variety with at most terminal \({\mathbb Q}\)-factorial singularities, that is, \(L\) is a nef and big Cartier divisor on \(X\). In Conjecture 2, the authors conjecture that the non effectivity of \(K_X+tL\), \(1 \leq t \leq s\) is equivalent to the non pseudoeffectivity of \(K_X+sL\) and they are able to prove it for \(s=\dim(X)\) (cf. Proposition 4). The case \(s=\dim(X)-1\) was previously known (see the Introduction of the paper and references therein) and, in Theorem 7, it is presented a explicit list of pairs such that \((X,L)\) is birationally equivalent to one of its elements. Several results on \(\dim(X)=4\) are also provided.