Quasi-elliptic cohomology. I (Q1785782)

Elliptic cohomology associates under certain conditions to an elliptic curve over a ring \(R\) an even-periodic cohomology theory \(E\) with \(E^0(\mathrm{pt}) = R\). One of the most important examples is based on the Tate curve over \(\mathbb{Z}[[q]]\), where the corresponding cohomology theory is called \textit{Tate \(K\)-theory}. One of the early ideas in the field surrounding the Witten genus was that the Tate \(K\)-theory of a space \(X\) should be an approximation of the circle-equivariant \(K\)-theory of its free loop space \(LX\), at least if we restrict to loops ``close to constant loops''. In the extreme case of just considering the constant loops themselves we obtain the equivariant \(K\)-theory \(K_{S^1}(X) \cong K(X) \otimes \mathbb{Z}[u^{\pm 1}]\), a completion of which yields the Tate \(K\)-theory \(K_{Tate}(X) \cong K(X) \otimes \mathbb{Z}[[q]]\). \textit{J. A. Devoto} [Mich. Math. J. 43, No. 1, 3--32 (1996; Zbl 0871.55004)] and \textit{N. Ganter} [Homology Homotopy Appl. 15, No. 1, 313--342 (2013; Zbl 1277.19003)] have defined an equivariant version of Tate \(K\)-theory. In line with the general philosophy of equivariant elliptic cohomology (see e.g. \textit{J. Lurie} [in: Algebraic topology. The Abel symposium 2007. Proceedings of the fourth Abel symposium, Oslo, Norway, August 5--10, 2007. Berlin: Springer. 219--277 (2009; Zbl 1206.55007)]), this theory should for finite groups of equivariance only depend on the torsion subgroup scheme of the Tate curve. In contrast to the Tate curve itself, this is not only defined over \(\mathbb{Z}[[q]]\), but already over \(\mathbb{Z}[u^{\pm 1}]\). This suggests that equivariant Tate \(K\)-theory for finite groups of equivariance should already be defined over \(\mathbb{Z}[u^{\pm 1}]\), which is exactly what the paper under review achieves. The keyword \textit{quasi-elliptic} can be read as referring to an equivariant cohomology theory based on a torsion-scheme that ``looks like the one of an elliptic curve''. The approach of the author though is not based on algebraic geometry, but rather on the free-loop space philosophy sketched above and follows closely ideas of Ganter and unpublished notes of Rezk. Given a space \(X\) with an action by a finite group \(G\), we can consider the topological stack \(Z = X//G\). Quasi-elliptic cohomology is defined as the \(S^1\)-equivariant \(K\)-theory of the constant loops in the free loop space of \(X//G\), in the following sense: Consider first the topological substack of \(\mathrm{Map}(S^1, X//G)\) of loops that become constant after pullback along \(\mathbb{R} \to S^1\); concretely there is a presentation in which the objects are pairs \((x,g)\) with \(x\in X\) fixed by \(g\). We further restrict to those objects where \(g\) is of finite order and define \(\Lambda (X//G)\) to be the stack quotient of this stack by the rotation action of \(S^1\). The \(G\)-equivariant quasi-elliptic cohomology of \(X\) is by definition the orbifold \(K\)-theory of \(\Lambda X//G\). The author shows the expected comparision to the equivariant Tate \(K\)-theory of Ganter.