On a conjectural filtration on the Chow groups of an algebraic variety. I: The general conjectures and some examples (Q1309344): Difference between revisions

In this article, the author describes a conjectural filtration on the Chow group of a smooth projective variety. He begins by giving a nice review of the construction of the Chow motives [see also the recent paper of \textit{A. J. Scholl} in Motives, Proc. Summer Res. Conf., Univ. Washington, Seattle 1991, Proc. Symp. Pure Math. 55, Part 1, 163-187 (1994)]. Let \(X\) be a smooth \(d\)-dimensional projective irreducible variety defined over a field \(k\). We say that \(X\) has a Chow-Künneth decomposition, if there are projectors \(\pi_ i \in \text{Chow}^ d (X \times X,\mathbb{Q})\) for \(0 \leq i \leq 2d\) such that (a) \(\pi_ i \circ \pi_ i = \pi_ i\) and \(\pi_ i \circ \pi_ j = 0\) if \(i \neq j\), where the compositions are the compositions of correspondences. (b) \(\Delta=\) the diagonal class \(=\sum \pi_ i\) modulo homological equivalence is the usual Künneth component \(\Delta (2d - i,i)\). In this setting, \((X, \Delta) =\bigoplus (X, \pi_ i)\) in the category of Chow motives. Next the author introduces the following sequence of conjectures on the structure of the Chow groups. Conjecture A. \(X\) always has a Chow-Künneth decomposition. Conjecture B. For each \(j\) \((0 \leq j \leq d)\), \(\pi_ k\) acts as zero on \(\text{Chow}^ j (X)\) for \(k=0,1,\dots,j-1\), and \(k=2j + 1,2j + 2, \dots, 2d\). Now assuming conjectures A and B, then the author defines a decreasing filtration on \(\text{Chow}^ j(X)\) as follows: \(F^ 0 = \text{Chow}^ j(X)\), \(F^ 1= \ker (\pi_{2j}) \dots\), \(F^ k = \ker (\pi_{2j - k + 1} | F_{k-1}), \dots, F^{j+1} = 0\). Conjecture C. The filtration above is independent of the choices of the projectors \(\pi_ k\). Conjecture D. \(F^ 1=^{Ch}\text{hom}(X,\mathbb{Q})=\) kernel of the cycle map, \(\text{Chow}^ j(X,\mathbb{Q}) \to H^{2j}_{et} (X,\mathbb{Q})\). By the earlier work of the author, it is known that the conjectures are true for curves and surfaces. For abelian varieties, conjecture A is true. Conjecture B is equivalent to a conjecture of Beauville.