Balanced line bundles on Fano varieties (Q1789361)

Given an ample, adelically metrized, line bundle \(L\) on a smooth projective variety \(X\) defined over a number field \(F\), \(L\) induces height functions on the set of \(F\)-rational points. Denote the number \(N(X(F), L, B)\) to be the number of \(F\)-rational points in \(X\) with height less than the number \(B\). Manin's Conjecture predicts the asymptotic behavior of this number and says that \(N(X^o(F'), L, B)\sim c B^{a(X, L)}\log(B)^{b(X, L)}\) as \(B\to \infty\) as long as we replace \(X\) by a suitable Zariski open set \(X\) and \(F'/F\) a finite extension. Here \(c>0\) is an constant, while \(a(X, L)\) and \(b(X, L)\) are two geometric invariants associate to \((X, L)\). This conjecture predicts a compatibility between the \(a\)- and \(b\)- invariants of \((X, L)\) and those of the subvarieties of \(X\). Namely, a subvariety \(Y\) of \(X\) with \(Y\cap X^o\neq \emptyset\) must satisfies \[ (a(Y, L), b(Y, L))\leq (a(X, L), b(X, L)) \] in the lexicographic order under the prediction of Manin's conjecture. In other words, all subvarieties violate the above inequality are supposed sit in a fixed closed proper subset of \(X\). The paper under review studies this geometric compatibility problem and investigate the geometric properties of \(a\)- and \(b\)- invariants. Note that the study of those invariants will provide potential counterexamples to Manin's Conjecture, or provide candidates to the choice of \(X^o\). All results are over an algebraic closed field of characteristic 0. For a smooth projective variety \(X\) and an ample line bundle \(L\) on \(X\), \(a(X, L)\) is defined to be the minimal \(t\) such that \(tL+K_X\) is pseudo-effective, and \(b(X, L)\) is defined to be the codimension of the minimal supported face containing \(a(X, L)L+K_X\) of the pseudo-effective cone. The first result is about the compatibility of \(a\)-invariants. It is shown that all subvariety \(Y\) of \(X\) satisfying \(a(Y, L)>a(X, L)\) is contained in a proper subset \(V\) of \(X\), which is a countable union of proper closed subsets. Moreover, if one assume that the so called Weak BAB Conjecture holds in dimension less that \(\dim X\), then one can show that in fact this \(V\) can be chosen to be a proper closed subset. Next, balanced line bundle or balanced pair is studied. The pair \((X, L)\) is called weakly balanced if there exists a proper closed subset \(V\) of \(X\) such that every subvariety \(Y\not\subset V\) satisfies \[ (a(Y, L), b(Y, L))\leq (a(X, L), b(X, L)) \] in the lexicographic order. It is called balanced if the above inequality is strict. It is shown by the minimal model program that if \((X, L)\) is balanced, then \(K_X+a(X, L)L\) is numerically equivalent to a rigid effective divisor, more precisely, \(X\) is birational to a log-Fano variety and \(K_X+a(X, L)L\) is numerically equivalent to an exceptional divisor. Finally, the techniques developed in this paper are used to study many examples whether \((X, L)\) is balanced, where \(X\) is a smooth Fano \(3\)-fold of Picard number 1 or a primitive Fano \(3\)-fold of Picard number 2, and \(L=-K_X\).