Milne's correcting factor and derived de Rham cohomology. II (Q683761)

Let \(X\) be a separated scheme of finite type over \(\mathbb{F}_q\) of dimension~\(d\), let \(Z(X,t) := \prod_{x \in X_0} \left(1-t^{\mathrm{deg}(x)}\right)^{-1}\) denote its zeta function and let \(n \in \mathbb{Z}\). Generalising work of \textit{J. S. Milne} [Am. J. Math. 108, 297--360 (1986; Zbl 0611.14020)] and \textit{S. Lichtenbaum} [Compos. Math. 141, No. 3, 689--702 (2005; Zbl 1073.14024)], \textit{T. Geisser} has proved in [Duke Math. J. 133, No. 1, 27--57 (2006; Zbl 1104.14011)] that the leading coeffixient \(Z^*(X,q^{-n})\) in the Taylor expansion of \(Z(X,t)\) at \(t=q^{-n}\) is equal to \(\pm \chi(H^*_{W,c}(X,\mathbb{Z}(n)), \cup e) \cdot q^{\chi_c^{eh}(X/\mathbb{F}_q, \mathcal{O},n)}\) provided every scheme over \(\mathbb{F}_q\) of dimension at most \(d\) admits resolution of singularities and the Tate-Beilinson conjectures hold. Here, \(H^*_{W,c}(X,\mathbb{Z}(n))\) denotes Geisser's `arithmetic cohomology with compact support', \(\cup e\) is the cup product with the fundamental class \(e \in H^1(W_{\mathbb{F}_q},\mathbb{Z})\) and \(\chi_c^{eh}(X/\mathbb{F}_q, \mathcal{O}, n)\) is defined to be \(\sum_{i\leq n, j} (-1)^{i+j} (n-i)\mathrm{dim}_{\mathbb{F}_q} H^j_c(X_{eh}, \Omega^i)\) where the so-called \(eh\)-topology is generated by étale coverings and abstract blow-ups. If \(X\) is smooth and projective, \(H^*_{W,c}(X, \mathbb{Z}(n))\) coincides with Weil-étale motivic cohomology and \(q^{\chi_c^{eh}(X/\mathbb{F}_q,\mathcal{O}, n)}\) coincides with Milne's correcting factor \(q^{\chi^{\text{Zar}}(X/\mathbb{F}_q,\mathcal{O}, n)}\). Moreover, \textit{B. Morin} has shown in [Doc. Math. 21, 39--48 (2016; Zbl 1346.14060)] that Milne's correcting factor is equal to the Euler characteristic of the derived de Rham complex \(R\Gamma(X_{\text{Zar}}, L\Omega^*_{X/\mathbb{Z}}/F^n)\) modulo the \(n^{\text{th}}\) step \(F^n\) of the Hodge filtration and that the formula for \(Z^*(X,q^{-n})\) given above can be restated in terms of a certain fundamental line. In the paper under review, Morin proves the precisely analogous results in general, replacing the Zariski topology in the definition of the derived de Rham complex with the the \(eh\)-topology. The core part here is to prove that \(\mathcal{H}^{i-n}(L\Lambda^n_{\mathcal{O}^{eh}}L_{\mathcal{O}^{eh}/\mathbb{Z}})\) is isomorphic to \(\Omega^i_{\mathcal{O}^{eh}/\mathbb{F}_q}\) if \(i \leq n\) and vanishes if \(i > n\).