A note on the Hodge index theorem (Q752100): Difference between revisions

The classical Hodge index theorem for a nonsingular complex projective surface S says that if X is a big divisor on S (i.e. \((X^ 2)>0)\) and E is a divisor on S such that \((E\cdot X)=0\) then \((E^ 2)\leq 0\), and \((E^ 2)=0\) iff E is numerically equivalent to zero [cf. \textit{O. Zariski}, Ann. Math., II. Ser. 76, 560-615 (1962; Zbl 0124.370)]. - In the note under review the author proves the following generalization of this theorem: Let V be an n-dimensional nonsingular projective complex variety, and let X be a big and nef divisor on V (i.e. \((X^ n>0\) and (X\(\cdot C)\geq 0\) for each curve \(C\subset V)\). If E is a divisor on V such that \((X^{n-1}\cdot E)=0\), then \((X^{n-2}\cdot E^ 2)\leq 0\) and \((X^{n-2}\cdot E^ 2)=0\) iff the curve \(X^{n-2}\cdot E\) is numerically equivalent to zero. To prove this result the author first verifies it in the case when X is an ample divisor with rational coefficients and proceeds with reducing the general case to this one by adding to X and E small multiples of an ample divisor on V. As an application, the author proves the following. Under the above assumptions, suppose that E is an effective divisor such that \((X^{(n-1)}\cdot E)=0\). Then \(h^ 0(X+E)=h^ 0(X)\). The former proof of this result given by the author [Am. J. Math. 111, No.3, 457-487 (1989; Zbl 0691.14005)] was more involved.