Geometric models for higher Grothendieck-Witt groups in \(\mathbb A^1\)-homotopy theory (Q2516889): Difference between revisions

This is a highly important paper. The authors prove a foundational geometric representability result in motivic homotopy theory. Recall the story from topology. To a (nice) topological space \(X\), one may associate the Grothendieck group of complex vector bundles \(K_0(X)\), or the Grothendieck group of real vector bundles \(KO_0(X)\), or many other variants. For now we shall concentrate on \(KO_0(X)\). It turns out that this functor is representable: there exists a space \(\mathbb{Z} \times BO\) such that the set \(KO_0(X)\) is naturally in bijection with homotopy classes of maps from \(X\) to \(\mathbb{Z} \times BO\). Moreover the space \(BO\) is nicely geometric: it is the infinite real Grassmannian. Additionally the representing space \(\mathbb{Z} \times BO\) is an infinite loop space (in fact with periodic deloopings), so one obtains a (co)homology theory \(KO_*(X)\). One would like to extend this picture into the motivic world. Certainly for a (nice) scheme \(X\) one may define groups \(K_0(X)\), \(GW_0(X)\) analogous to the topological case. These are known as algebraic \(K\)-group and the Grothendieck-Witt group. These have been known for a long time, and the desire to ``complete'' them into cohomology theories existed even before the advent of motivic homotopy theory. So in fact there are already defined spaces \(K(X)\), \(GW(X)\) such that the homotopy groups define (higher) algebraic \(K\)-theory/hermitian \(K\)-theory. The next step would be to ask if there is \textit{some} motivic space \(\mathcal{GW}\) such that \(\mathbb{A}^1\)-homotopy classes of maps from \(X \wedge S^n\) to \(\mathcal{GW}\) coincide with \(GW_n(X)\). Since a motivic space is essentially a presheaf of simplicial sets (spaces), and \(GW_n(X)\) is defined as the homotopy groups of some space, this is fairly plausible. The fact that this is possible is a formal consequence of the properties of the (co)homology theory \(GW_*\) and has been established by \textit{J. Hornbostel} [Topology 44, No.~3, 661--687 (2005; Zbl 1078.19004)]. (Actually, hermitian \(K\)-theory is doubly indexed, but we ignore this for sake of exposition.) The more difficult question tackled in this article is whether the space \(\mathcal{GW}\) can be given a geometric interpretation as \(\mathbb{Z} \times BO\). Here again the motivic question is more subtle, because there are at least two reasonable interpretations of \(BO\). One could either mean the space \(B_{et} O\), which is an essentially simplicial construction, or an algebraic version of the orthogonal Grassmannian. The main result of the paper under review is that both of these potential models are equivalent as motivic spaces, and in fact equivalent to \(\mathcal{GW}\), so represent hermitian \(K\)-theory. The paper also ties up loose ends: one may repeat the story with symplectic or linear instead orthogonal groups; then one obtains geometric representability of other versions of algebraic \(K\)-theory. The authors obtain explicit periodic deloopings of the spaces involved, thus lifting the entire picture of real Bott periodicity into the motivic world.