Nori 1-motives (Q2255282): Difference between revisions

The category \(\mathcal{M}_1\) of 1-motives was defined in [\textit{P. Deligne}, Publ. Math., Inst. Hautes Étud. Sci. 44, 5--77 (1974; Zbl 0237.14003)]. In one of his letters Deligne recounts what the 1-motives he invented are for. \textit{The motivic philosophy tell us that the groups \(H^n(X)\) attached to a variety \(X\) should come from a `motive' \(H^n_{\mathrm{motivic}}(X)\). For \(X\) projective and smooth and \(n=1\) abelian varieties play this role \(\dots\) The 1-motives play the same role, still for \(n=1\), when \(X\) is no more assumed to be projective and smooth. Grothendieck's theory of pure motives relies on conjectures which remain inaccessible \(\dots\) 1-motives need no conjecture, are related only to \(H^1\), should be a small corner of a theory of mixed motives.} One such a theory has been constructed by Nori (see \textit{N. Fakhruddin}, Notes of Nori's Lectures on Mixed Motives. TIFR, Mumbai (2000)]. It is the abelian category \(\mathbf{EHM}\) of effective homological mixed motives. Fix a base field \(k\) with an embedding \(\sigma:k\hookrightarrow \mathbb C\), consider a \(k\)-scheme \(X\) and a closed subset \(Y\subset X\), we have motives \(\widetilde{H}_i(X,Y;\mathbb Z) \in \mathbf{EHM}\) for all \(i\in \mathbb N\). Their Betti realisations are the usual relative homology groups \(H_i(X(\mathbb C),Y(\mathbb C);\mathbb Z)\). The category \(\mathbf{EHM}_1\) of Nori 1-motives is defined in the present work to be the abelian subcategory of \({\mathbf{EHM}} \) which is generated by the images of \( \widetilde{H}_i(X,Y;\mathbb Z) \in\mathbf{EHM}\) for \(i \leq 1\). The major result of the paper is the proof that \(\mathbf{EHM}_1\) is naturally equivalent to \({}^t\mathcal{M}_1\), this is the category of Deligne's 1-motives with torsion, a generalization of \({}^t\mathcal{M}\) produced by L. Barbieri-Viale, A. Rosenschon and M. Saito [Ann. Math. (2) 158, No. 2, 593--633 (2003; Zbl 1124.14014)]. The roadmap of the demonstration is by a chain of equivalences, \(\mathbf{EHM}_1 \simeq\mathbf{EHM}'_1 \simeq \mathbf{EHM}''_1\), where the generators for \(\mathbf{EHM}'_1 \) satisfy the further condition that the dimension of \(X\) is at most \(1\), while for \(\mathbf{EHM}''_1 \) the request is that \(X\) is a smooth and affine curve. The key point of the paper is the existence of a canonical functor \(\mathbf{EHM}''_1 \to {}^t\mathcal{M}_1 \) which is in fact an equivalence of categories. Morphisms in \({\mathbf{EHM}} \) are formally defined by means of the abstract properties of relative cohomology, so their geometric content is possibly mysterious. One of the merits of this article is to provide such understanding for this corner of Nori's theory.