Higher arithmetic \(K\)-theory (Q2580241)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Higher arithmetic \(K\)-theory |
scientific article |
Statements
Higher arithmetic \(K\)-theory (English)
0 references
18 January 2006
0 references
The aim of this paper is to give a new definition of higher \(K\)-theory in Arakelov geometry. Let \(X\) be a regular scheme which is flat and proper over \(\mathbb{Z}\). After Gillet and Soule defined the arithmetic Grothendieck group \(\widehat K_0(X)\), there have been several attempts to define higher arithmetic \(K\)-groups as the homotopy groups of the homotopy fiber of the regulator map: \[ K_{n+1}(X)@>\rho>> \bigoplus_p H_{\mathcal D}^{2p-n-1}(X,\mathbb{R}(p)) \] where \(H_{\mathcal D}^n(X,\mathbb{R}(p))\) is the real Deligne cohomology. This means that there should be a group \(KM_n(X)\), for all \(n\geq 0\) fitting into a long exact sequence: \[ \cdots \to K_{n+1}(X)@>\rho>>\bigoplus_pH_{\mathcal D}^{2p-n-1}\bigl(X,\mathbb{R}(p) \bigr)\to KM_n(X)\to K_n(X)\to\cdots\tag{1} \] In this paper the author gives a different definition of higher arithmetic \(K\)-groups \(\widehat K_n (X)\), which are given as an extension of the algebraic \(K\)-groups by the cokernel of the regulator map. These groups fit into a fundamental exact sequence: \[ K_{n+1}(X) \to\widetilde{\mathcal D}_{n+1}(X)\to\widehat K_n (X)\to K_n(X)\to 0 \] where \(\widetilde{\mathcal D}_{n+1}(X)={\mathcal D}_{n+1}(X)/d_{\mathcal D}\) and, for a compact complex algebraic manifold \(M\) (i.e., the compact manifold \(X(\mathbb{C})\) consisting of \(\mathbb{C}\)-valued points of \(X)\) the complex \(({\mathcal D}^*(M,p),d_{\mathcal D})\) is a complex of \(\mathbb{R}\)-vector spaces computing Deligne's cohomology, i.e, \[ H^n \bigl({\mathcal D}^*(M,p),d_{\mathcal D}\bigr)=H^n_{\mathcal D}\bigr(M.\mathbb{R}(p) \bigr), \] for \(n\leq 2p\). There are Chern class maps: \(\text{ch}_n:\widehat K_n (X)\to{\mathcal D}_n(X)\) such that, by defining \(KM_n(X)=\text{Ker}\, \text{ch}_n\) one obtains a long exaxt sequence as in (1). These groups have products: \(\widehat K_n(X)\times\widehat K_m(X)\to\widehat K_{n+m} (X)\), pull-backs \(\widehat f^*\) for arbitrary morphisms \(f:X\to Y\) and push-forwards \(\widehat f_*\) for smooth and projective morphisms. A projection formula for \(\widehat f^*\) and \(\widehat f_*\) holds. From the above properties one gets a non-canonical decomposition as follows: \[ \widehat K_n(X) \simeq K_n(X)\oplus{\mathcal D}_{n+1}(X)/ \text{Ker} \,d_{\mathcal D}\oplus\left(\bigoplus_p \frac{H_{\mathcal D}^{2p-n-1}\bigl(X,\mathbb{R} (p)\bigr)}{\operatorname{Im}\rho}\right). \] According to Bass' conjecture the first term should be a finitely generated abelian group, while the last one, according to Beilinson's conjectures, should be a real torus. The mid term is an infinite dimensional real vector space.
0 references