Specialization of cycles and the \(K \)-theory elevator (Q2418754): Difference between revisions

Let \({\bar\pi}: {\mathcal X}\rightarrow {\mathcal S}\) be a proper, dominant morphism of smooth quasi-projective varieties over \(\mathbb C\) with \(\dim{\mathcal S}=1\) and smooth restriction \({\pi}: {\mathcal X}^{*}\rightarrow {\mathcal S}^{*}={\mathcal S}- \{s_0\}.\) Let \(X_s={\pi}^{-1}(s),\) \({\mathbb V}:=R^{2p-r-1}{\pi}_*{\mathbb Q}(p)\) with the monodromy operator \(T\) about \(s_0.\) Let \({\mathcal Z}^*\in {\mathrm{CH}}^p({\mathcal X}^*,r)_{{\mathbb Q}}\cong H^{2p-r}_{\mathcal M}({\mathcal X}^{*},{\mathbb Q}(r))\) be a higher Chow cycle. If \(r=0\) assume that the restrictions \(Z_s={\iota}^{*}_s {\mathcal Z}^{*} \) are homologous to zero. In this situation one can associate an admissible normal function \(\nu \in {\mathrm{ANF}}^r_{{\mathcal S}^{*}}({\mathbb V}).\) This function is on fibers of \({\pi}\) given by the values of the regulator maps: \(AJ^{p,r}_{X_s}(Z_s)\in {\mathrm{Ext}}^1_{\mathrm{MHS}}({\mathbb Q}, H^{2p-r-1}({ X}_s,{\mathbb Q}(p)).\) It is difficult to show that the normal function is nonzero - this is possible in some cases but requires advanced techniques. General relation between regulator maps are given in [\textit{M. Kerr} et al., Compos. Math. 142, No. 2, 374--396 (2006; Zbl 1123.14006)]. In the absence of singularity at \(s_0\) one can consider the limit of the normal function at \(s_0.\) In order to compute this the authors use the specialization i.e. if \({\mathcal Z}^*\) lifts to \({\mathcal Z}\in {\mathrm{CH}}^p({\mathcal X},r)_{\mathbb Q}\) the one gets a class \({\iota}^*_{s_0}\in H^{2p-r}_{{\mathcal M}}(X_{s_0}, {\mathbb Q}(p)).\) For a specialized cycle, the authors use a semi-simplicial hyper-resolution of \(X_{s_0}\) to compute its Abel-Jacobi class in the absolute Hodge cohomology \(H_{\mathcal H}^{2p-r}(X_{s_0}, {\mathbb Q}(p))\cong {\mathrm{Ext}}^1_{\mathrm{MHS}}({\mathbb Q}, H^{2p-r-1}({ X}_s,{\mathbb Q}(p)).\) The main result od the paper of general nature is that the image of this class under the Clemens retraction computes \({\nu}(s_0).\) It is worth noting that the paper is very well-written, contains many worked out examples as well as the motivation for this research.