Higher equivariant excision (Q509665)

Homotopy calculus of functors, developed by Goodwillie, is a theory that aims to approximate a homotopy functor \(F:\mathcal C\to\mathcal D\) between model categories by a tower of functors \(F\longmapsto(\ldots\to P_nF\to P_{n-1}\to\ldots \to P_1F)\). The functor \(P_nF\) is \(n\)-excisive, which means that it takes strongly cocartesian cubical diagrams to cartesian cubical diagrams; this is the sense in which \(P_nF\) is a polynomial functor. Goodwillie also classifies homogeneous functors for certain categories, namely the homotopy fibers of the transformations \(P_nF\to P_{n-1}F\). This theory has since its inception seen many applications and has been developed in many directions. The goal of the paper under review is to set up an equivariant homotopy calculus and prove equivariant versions of many of the main theorems in the field. The author does this successfully and thoroughly in a style that is comprehensive and readable. When the action of a group is trivial, he recovers the original Goodwillie calculus, so this is genuinely a generalization. The paper is lengthy, but this is not unexpected since a substantial amount of homotopy-theoretic machinery is required for establishing the main results. One helpful factor is that the author has already developed the theory of equivariant cubical diagrams and their homotopy (co)limits in two previous papers [the author, Algebr. Geom. Topol. 16, No. 2, 1157--1202 (2016; Zbl 1339.55015); the author and \textit{K. Moi}, ibid. 16, No. 1, 325--395 (2016; Zbl 1341.55001)], and this theory is used throughout. In a little more detail, the author sets up calculus of functors for homotopy functors \(F: \mathcal C^G\to\mathcal D^G\) between categories of \(G\)-objects (such as \(G\)-spaces or \(G\)-spectra) where \(G\) is a finite group. He starts by defining strongly cocartesian \(J_+\)-cubes where \(J\) is a \(G\)-set with a disjoint basepoint. Then \(\Phi: \mathcal C^G\to\mathcal D^G\) is \(J\)-excisive if it takes such cubes to cartesian \(J_+\)-cubes. In analogy with ordinary functor calculus, one of the main theorems of the paper is that such \(J\)-excisive approximations \(P_J\Phi\) exist for any \(\Phi\) and come with a transformation \(\Phi\to P_J\Phi\). The next step is to build the Taylor tower of approximations, but now this is more complicated since one has transformations \(P_J\Phi \to P_K\Phi\) for \(K\subset J\) and also when there is a \(G\)-map \(K\to J\) that induces an injection on orbits \(K/G\to J/G\). The tower is thus a diagram that has \(P_G\Phi\) as its final object, and this diagram becomes the ordinary tower when the action is trivial. One of the results of the paper is the convergence, interpreted in a suitable way, of this diagram for functors \(\text{Top}_*^G\to\text{Top}_*^G\) that commute with fixed points. One more crucial piece of the original theory is the classification of homogeneous functors in various settings, and the author proves that result for functors \(\text{Top}_*^G\to\text{Top}_*^G\). More precisely, he shows that such a functor \(\Phi\) is equivalent to \[ \Omega^{\infty J}(E_\Phi\wedge X^{\wedge n})_{h\Sigma_n}, \] where all the objects are suitably defined equivariant versions of the objects appearing in the original Goodwillie classification result. One of the big successes of functor calculus has been its usefulness in the context of the identity functor, and the author tackles the analogous equivariant situation as well. Namely, he gives a description of the layers of the equivariant tower of the identity functor on \(G\)-spaces (and now one indeed gets a tower) by looking at the equivariant version of the Snaith splitting in the style of \textit{G. Arone} and \textit{M. Kankaanrinta} [in: Stable and unstable homotopy. Proceedings of workshops held during 1996 at the Fields Institute, Waterloo, Canada. Providence, RI: American Mathematical Society. 1--30 (1998; Zbl 0908.55007)]. Furthermore, the author gives ``tom Dieck decompositions'' of these layers by delooping \(J\)-homogeneous functors, and this is in fact what leads to the classification result mentioned above. The author does a lot more in this impressive paper. He provides all the necessary pieces for the generalization of homotopy calculus to this important new equivariant setting and does so in an enviably organized and thorough manner.