A note on Iitaka's conjecture \(C_{3,1}\) in positive characteristics (Q1750698): Difference between revisions

The birational geometry of an algebraic variety \(X\) is governed to a large extent by the behaviour of the canonical divisor \(K_X\). Therefore, if \(X\) is somehow related to an other variety \(Z\), it is crucial to find relations between \(K_X\) and \(K_Z\). For instance, in the case where \(Z\) is a prime divisor in a smooth variety \(X\), the following relation, known as adjunction formula, holds: \((K_X + Z)|_Z = K_Z\). When \(X\) admits a fibration onto an other variety \(Y\), that is a surjective morphism with connected fibers \(f : X \to Y\), Iitaka conjectured the following fundamental relation between the Kodaira dimension of \(X\), \(Y\) and the generic fiber of \(f\), denoted with \(X_{\eta}\): \[ k(X) \geq k(Y) + k(X_{\eta}). \] The statement is referred to as \(C_{n, m}\), where \(n\) and \(m\) detone, respectively, the dimension of \(X\) and the dimension of \(Y\). Over an algebraically closed field \(k\) of characteristic zero, the generic fiber \(X_{\eta}\) is smooth over \(k(\eta)\) and Iitaka conjecture has been proved in many important cases, see for instance [\textit{C. Birkar}, Compos. Math. 145, No. 6, 1442--1446 (2009; Zbl 1186.14015)] and [\textit{J. A. Chen} and \textit{C. D. Hacon}, Invent. Math. 186, No. 3, 481--500 (2011; Zbl 1234.14006)] and references therein. Over a field of characteristic \(p > 0\), instead, the conjecture is more sublte, due to separability issues and the fact that the generic fiber is not necessarily smooth. In the paper under review, the author proves the statement \(C_{3, 1}\) for varieties defined over an algebraically closed field of characteristic \(p > 5\) under the assumption that the canonical divisor of the generic fiber is big. The proof combines methods of the minimal model program in dimension three and two and the study of vector bundles over elliptic curves.