Decomposition of small diagonals and Chow rings of hypersurfaces and Calabi-Yau complete intersections (Q2437431)

The present paper is an interesting sequel to \textit{C. Voisin}'s work [Pure Appl. Math. Q. 4, No. 3, 613--649 (2008; Zbl 1165.14012)] on the intersection properties of the Chow ring of Calabi Yau hypersurfaces. In the first part the author generalizes Voisin's results to cover the case of a general complete intersection \(X\) in a projective space, under the hypothesis that \(X\) is still of C-Y type. The method centers around Voisin's idea of studying the rational class of the small diagonal on \(X \times X \times X \) by means of its relation with the geometry of the correspondence \(V\) of triples of points on \(X\) which are collinear. The situation of complete intersections is more complex than the one Voisin was dealing with, because now some of the cycles which appear in the game have the wrong dimension. Fu must appeal to Fulton's excess intersection formula, which he uses in a skillfull and efficient way. In fact he deals with the more general situation of when \(X\) is the zero locus of a conveniently ample vector bundle of rank \(r\) on \(\mathbb P^{n+r} \), but the main result about the decomposition of the small diagonal can be proved entirely only for complete intersections. The author uses the terminology that on a variety \(X\) of dimension \(n\) a \(0-\)cycle class with rational coefficients is said \textit{decomposable} if it is contained in the subspace of \(\mathrm{CH}_0(X)_{\mathbb Q} \) which is spanned by the images of the intersections of cycles of complementary dimensions, \( \mathrm{Im}(\mathrm{CH}^m (X)_{\mathbb Q} \times \mathrm{CH}^{n-m} (X)_{\mathbb Q} \to \mathrm{CH}_0(X)_{\mathbb Q} \), where \( 1 \leq m \leq (n-1)\). The main consequence of Fu' s work on the small diagonal of a general complete intersection of C-Y type says that a decomposable class is trivial if and only if it has degree \(0\). At this point the author gives a contrasting example. With remarkable geometrical prowess he finds a non singular surface \(S\) in \(\mathbb P^3\) of general type such that the image of the intersection product of divisors in \(\mathrm{CH}_0(S)_{\mathbb Q} \) is strictly larger that the copy of \(\mathbb Q\) generated by the intersection with a line. Next Fu extends Voisin's results for Calabi Yau hypersurfaces in a second different direction, he inquires what remains true for the case of a hypersurface \(X\) of general type, of degree \(d\geq n+3\) in \(\mathbb P^{n+1}\). We have recalled above that Voisin's main tool was a decomposition in \(\mathrm{CH}_n(X^3 )_{\mathbb Q} \) of the class of the small diagonal. Nori suggested that there should be a generalization also for the case of when \(X\) is of general type. The author can accomplish this by a careful and clever analysis of the geometry of collinear points on \(X\). In the end he is able to compute a precise form for the desired decomposition of the small diagonal in \(\mathrm{CH}_n (X^k)\), where \(k = d+1 - n\), and he proves that either this decomposition holds or that there is a smallest integer \(3 \leq m < k\) for which there is a simpler decomposition for the small diagonal on \(\mathrm{CH}_n (X^m)\). Again, as a corollary, a statement parallel to the preceding one holds for decomposable classes in \(\mathrm{CH}_0(X)_{\mathbb Q} \). A final remark explains that this last fact is an expected consequence of the Bloch-Beilinson conjecture.