Standard conjecture of Künneth type with torsion coefficients (Q2404126)

Let \(X\) be a complex smooth projective variety of dimension \(n\). Let \(cl_{X,\mathbb{Q}}^j:\text{CH}^j(X)\rightarrow H^{2j}(X,\mathbb{Q})\) denote the cycle class map from the Chow group of codimension-\(j\) cycles on \(X\). For each integer \(i\), one has the rational Künneth projector \(\pi_{X,\mathbb{Q}}^i\in H^{2n}(X\times X,\mathbb{Q})\), which acts as 1 on \(H^i(X,\mathbb{Q})\) and as 0 on \(H^j(X,\mathbb{Q})\) for all \(j\neq i\). Grothendieck's standard conjecture of Künneth type asserts that the Künneth projectors \(\pi_{X,\mathbb{Q}}^i\) are algebraic, namely \(\pi_{X,\mathbb{Q}}^i\in \text{Im}(\text{cl}_{X\times X,\mathbb{Q}}^n)\otimes\mathbb{Q}\) for all \(i\). The present paper is concerned with the question, raised by A. Venkatesh, which asks whether the mod-\(p\)-analogue of the standard conjecture holds true or not. More precisely the problem is whether or not the mod-\(p\)-Künneth projectors \(\pi_{X,\mathbb{F}_p}^i\), which is defined similarly by replacing the coefficient \(\mathbb{Q}\) by \(\mathbb{F}_p\), belong to \(\text{Im}(\text{cl}_{X\times X,\mathbb{F}_p}^n)\) for all \(i\). One of the main result of this article is to answer the question in the negative by showing that, if \(H^*(X,\mathbb{Z})\) has nontrivial \(p\)-torsion, then Venkatesh's question has a negative answer for \(X\). Furthermore he shows that for any integers \(i>0\) and \(n\geq i+1\), there exists a projective smooth variety \(X\) of dimension \(n\) such that \(\pi_{X,\mathbb{F}_p}^i\) is not in the image of the mod-\(p\) cycle class map. On the other hand, he looks into the question in the case of abelian varieties, and shows that Venkatesh's question has an affirmative answer for all primes \(p\), if \(X\) is the Jacobian of a curve.